July 29, 2026 – I’ve written several posts in the “How to Lie with Personal Finance” series, all dealing with common misconceptions and, well, sometimes outright lies in the personal finance world: General Personal Finance Lies, Homeownership Lies, and Diversfiction Lies. Today I collected a set of lies for another interesting topic. Recently, I’ve heard and read a lot about an ostensibly innovative asset allocation strategy, Risk Parity, and its advantages, especially for retirees. Among some of the purported benefits are lower volatility, less stock market exposure, and higher sustainable withdrawal rates. The rationale for the superiority of this asset allocation is that it covers all the bases and hedges against different economic regimes, i.e., high growth vs. low growth and high inflation vs. low inflation. Some folks claim you can raise your safe withdrawal rate from 4% to 5% if you use Risk Parity in your retirement portfolio. So, why haven’t I proclaimed victory over Sequence of Return Risk yet? Mainly because there is a lot of hype, false advertising, and misunderstandings about Risk Parity. Here are several reasons to be skeptical…
1: Risk Parity is no Super-Secret Hedge Fund Weapon
As a former finance professional, I laughed out loud when I heard that Risk Parity is being sold as some super-secret weapon that only elite hedge fund tycoons like Bridgewater’s Ray Dalio figured out. At least that’s what Tony Robbins insinuates in his book “MONEY Master the Game: 7 Simple Steps to Financial Freedom.” If you don’t want to buy the book, here’s a short summary. Tony Robbins sells Risk Parity as having extracted the secret sauce from one of the famous hedge fund tycoons, and he’s now letting us in on this financial superweapon. That’s a bunch of hooey! Every finance student will learn about Risk Parity in Finance 101. It’s like someone asked the Ferrari chairman what the secret to building great cars is, and he responds, “A combustion engine and four wheels.” While Ferraris certainly have four wheels and a combustion engine (except for that ugly new electric Ferrari Luce that looks like a Volvo), the chairman didn’t really give away any proprietary information about their engines, transmissions, aerodynamics, design, etc. Only the most useless generic information that will certainly not give you a heads-up. Of course, Tony Robbins still runs with this and beats the drum (literally and figuratively) about Risk Parity.
In the FIRE/Personal Finance community, we got our friend Frank Vasquez, who touts Risk Parity as a brilliant and innovative way to think about diversification on his Risk Parity Podcast and on several guest appearances on other podcasts. He’s made that point on Bigger Pockets, Forget About Money, ChooseFI, Afford Anything, and likely more podcasts. But that’s all mostly bogus. Quite the opposite: naive Risk Parity relies solely on the variance-covariance matrix and will not cut it in the highly competitive field of asset management and high finance. Truly competent asset managers will consider risk and expected return estimates to trade off risk versus return; more on that below.
Most importantly, while Risk Parity may be one ingredient in Bridgewater’s hedge fund, you don’t grow your hedge fund empire to Bridgewater’s size with only Risk Parity principles. Hedge funds are all about alpha, not some static strategic asset allocation that anyone can replicate. You don’t need 1,300 employees (as of the most recent 2025 estimate, according to Wikipedia) to run something as trivial as Risk Parity, which any intelligent college student with basic Python or MATLAB skills can calculate on their home computer. Put differently, a Risk Parity strategy is only as good as the underlying returns. If you feed in a bunch of zero expected return streams into Risk Parity, you still have zero expected returns. Risk reduction is useless if the returns are not there. Of course, Bridgewater has all the know-how to squeeze additional alpha out of the financial market, whether through market timing, stock picking, sector rotation, etc., to add enough extra returns to Risk Parity to make it competitive. If you implement the “dumb” Risk Parity model floating around on the interwebs, you get subpar results.
Also, the timing is peculiar. Risk parity was all the rage twenty years ago in the mid-2000s because we were living in the “Goldilocks” economy where all major asset classes were doing well in the Post-Dot-Com-Crash era, and a Risk Parity portfolio garnered quite attractive returns. Then, the global financial crisis struck; equities took a nosedive, and the commodity bubble burst around the same time. In fact, the bellwether GSCI commodity index is still about 50% below its peak in nominal terms (and the GSG ETF more than that). Adjusted for inflation, the picture is even bleaker. Recently, the decade-long bond bull market came to an abrupt end with the inflation shock of 2022 and the swift Federal Reserve rate hikes. So, Risk Parity fell out of favor. A simple Stock/bond portfolio would have given you all the diversification you needed. I worked in institutional asset management until 2018, and Risk Parity was a non-entity at that time. But hey, this is the perfect time to drop a new old asset management catch phrase on the naive and gullible retail investor, after the smart money has long ago said goodbye. The Dunning-Kruger financial influencers are ready to spread this nonsense!
Side Note: What exactly is Risk Parity?
A standard exercise in basic finance and portfolio management is to calculate the risk contributions of different portions of a portfolio. It’s an easy application of elementary statistics and matrix algebra. The contribution has a pretty neat and intuitive interpretation, and we can derive it in two different ways: the absolute contribution to the portfolio variance and the marginal contribution to the portfolio risk, which both give you the same result. I put together a brief technical note in PDF format and posted it here.
Let’s go through a simple numerical example with just two asset classes, stocks and bonds. Assume stocks and bonds have annualized standard deviations of 16% and 6%, respectively, and a correlation of 0.10. If we plot the marginal risk contribution of equities as a function of the equity weight, we notice something peculiar. Of course, a 100% equity portfolio should have 100% risk coming from equities. But adding bonds, and even substantial percentages of bonds, to the portfolio hardly makes any dent in the equity contribution. The curve is almost flat between 60% and 100% equity weight. Even a 60/40 portfolio still has more than 90% of its risk coming from stocks. What’s going on here? This is an artifact of how we aggregate variances: even though equities have a standard deviation of “only” about 16/6=2.67 times the bond standard deviation, the equity variance is 7.1x the bond variance (=2.67 squared!), so equities will always dominate the risk attribution in a stock/bond portfolio, unless you vastly overweight bonds. Thus, in this example, you’d need to go down to roughly a 27% equities and 73% bond portfolio to equalize the risk contribution and achieve Risk Parity.
But alas, this chart above is really only a cheap party trick. It could be abused by charlatans to tell me that something needs to be done about the massive equity risk in my portfolio, when it really doesn’t. Rational and honest investors will instead look at the risk budgeting math the following way. In the chart below, let’s plot the actual portfolio risk (standard deviation) as a function of the equity weight and then also add the equity contribution (i.e., multiply the blue line by the percentage contribution in the previous chart). A 60/40 portfolio has only about 10% annualized risk, roughly 37% less than the 100% equity portfolio. So, diversification with 40% bonds lowered your risk by almost exactly that number. True, as a percentage of that reduced risk, equities still contribute 92% of the risk, but the risk is certainly much lower than before. Bonds helped with diversification much more than what a gullible retail investor would have deduced from the previous chart.
The danger of Risk Parity is that you apply too much diversification, by moving out of high-return equities and into low-return bonds and commodities. Diversification then becomes Di-WORSE-fication. Also, leverage doesn’t help, more on that later in item #7!
For the record, though, I generally like the idea of hedging against different economic risks, but the performance in practice, for the average retail investor without the advanced toolkit of Ray Dalio’s hedge fund, is disappointing; more on that later.
2: Risk Parity is rife with “Hindsight Bias”
My fraud sensors go off whenever someone mentions a new portfolio allocation innovation. Then, I always ask myself, “Did you optimize this portfolio with knowledge unavailable at the beginning of the simulation period or with the benefit of hindsight?” That is certainly the case with some (or all?) of the various Risk Parity approaches floating around. The Hindsight Bias in Risk Parity portfolios is visible in at least two dimensions: 1) what asset classes to include, and 2) what weights we pick for those asset classes. Then I ask myself how this new strategy performed “out-of-sample,” i.e., did the outperformance continue when we exposed this new portfolio to subsequent return patterns? Most of the time, the results are disappointing.
For example, the All-Weather, aka All-Seasons, portfolio suffers from this hindsight bias. Tony Robbins unapologetically admits that the All-Weather strategy was optimized by examining the 1984 to 2013 time span and tailoring the returns to what worked best then. The Golden Butterfly portfolio has also existed since the early 2010s. It probably originated from the PortfolioCharts site, which uses data starting in 1970. The Golden Butterfly weights also seem to work well over the 1970s to early 2010s time span, but then sputter noticeably once people went from the benefit of hindsight to putting actual money at stake in a real portfolio.
But, alas, once people popularized those new portfolio weights, the returns didn’t look so great anymore when you go from the benefit of hindsight to putting actual money on the line. Let me demonstrate that with the All Weather portfolio: I’ll calculate the returns between 1984 and 2013, then compare how you’d have done since then (Dec 31, 2013 to June 30, 2026). Also, for completeness, I’ll throw in the return stats for pre-1984 and the entire horizon. And quite impressively, the All Weather portfolio blew everything else out of the water during 1984-2013. It had a Sharpe Ratio of 0.71 and close to 7% returns with only 8% volatility. That’s roughly the same return as the S&P fetched over the entire period, but with half the equity risk. I’m very impressed! But alas, since this amazing portfolio made the rounds on YahooFinance, the performance has been lackluster, significantly trailing the 75/25 and 60/40 portfolios since 2013. The average return was only 3.29% above inflation, and the Sharpe Ratio was very weak.
I find this recent All Weather performance quite astonishing because we did observe all of the economic regimes: high growth and low growth, i.e., one of the longest economic expansions on record up to the pandemic. Then a deep recession in 2020 and a slowdown in 2022. Likewise with inflation: We’ve had low and stable inflation and high inflation. Specifically, the two equity bear markets covered both bases, i.e., a demand shock with disinflation during the pandemic in 2020 and an inflation spike and rapid FOMC policy rate hike in 2022-2023. Even in this ideal testing ground for the Risk Parity approach, the venerable old 60/40 portfolio mopped the floor with the All-Weather portfolio. The 75/25 did even better. It’s peak-time hindsight bias: the overfitted portfolio that did so well with the benefit of hindsight sputters when you run it “out of sample.”
Also noteworthy: Over the entire 155-year time span from 1871 to 2026, the All Weather strategy is just mediocre, both in total returns (CPI+4.16%, almost a full two percentage points behind the 75/25) and in risk-adjusted returns: 0.29 Sharpe vs. 0.33 Sharpe for the 60/40 and 75/25 simple stock/bond portfolios. Also, All-Weather didn’t perform that well in the 1871-1983 era. How would an investor in 1984 have known that All-Weather would do so well over the subsequent 30 years? Without a time machine, nobody in the 1980s would have found this appealing.
3: Most Risk Parity strategies posted on the web aren’t really Risk Parity at all
The asset allocations that are often sold as Risk Parity aren’t Risk Parity in the mathematical sense at all. Risk Parity means the marginal contribution to the overall portfolio standard deviation is equal across all asset classes, as described in the section above. None of the allocations I’ve seen satisfy that criterion. For example, in the table below, I calculate the 120-month (4/2016-3/2026) risk attribution for several Risk Parity strategies. None of the purported RP strategies spread their risk equally across the asset classes. The All Weather portfolio comes close to 33% equity risk attribution, but bonds (57%) are well above the target, while commodities (7.4%) are below one third, no doubt a result of hindsight bias (see item 2 above) because riding the duration bet since the early 1980s was so much more lucrative than volatile commodities in the 1984-2013 window. Likewise, the Golden Butterfly and Golden Ratio portfolios have around two-thirds of their risk come from equities, but only around 16-19% from bonds, 11-17% from commodities, and -1% for the DBMF (trend-following) ETF. That’s not a typo; a risk contribution can indeed be negative.
Another giveaway that all these strategies are Risk Parity in name only is that their portfolio weights never seem to change. And I am not talking about occasionally replacing one ETF with another. The asset class weights should move substantially because the variance-covariance (VCV) matrix used to calculate those weights changes over time, both due to variances and covariances/correlations. For example, the chart below shows the risk parity portfolio weights calculated from a 120-month rolling VCV, using the three assets S&P 500, Long-Term Treasury Bonds, and Commodities (mimicked through SPY, TLT, GSG ETF returns and simulations before actual returns start). The weights are all over the place, unlike the completely static allocations recommended by risk parity fans. In fact, there is some financial research (Moreira, A. and Muir, T. (2017), Volatility-Managed Portfolios. The Journal of Finance, 72: 1611-1644. https://doi.org/10.1111/jofi.12513) that shows that dynamically shifting asset weights in response to volatility changes can generate more attractive return profiles. But that potential all goes out the window if you force the portfolio weights to stay constant over time, as done in the fake-Risk-Parity portfolio marketed on the web.
Of course, I’m not saying that “risk non-parity” in Frank’s Risk Parity portfolios is a huge problem. Quite the opposite, it’s not a bug; it’s likely a feature because some of the asset return results would have looked even worse if you had insisted on strict Risk Parity weights. Specifically, because the Golden Butterfly and Golden Ratio portfolios still maintain a decent equity allocation (40-42% weight), they will perform reasonably well. The fact that the All Weather portfolio has a risk contribution of only 35% from equities creates a huge drag in my retirement withdrawal simulations. More on that later.
Thus, a better, more appropriate label for these strategies should then be “Risk Budget(ed)” or “Risk Aware” or “Risk-Managed” strategies, rather than Risk Parity. That’s because Risk Parity is a mathematically well-defined term and none of the strategies satisfy this very unambiguous condition. The fact that Risk Parity fans don’t know or don’t want to know this is a red flag. I would not trust my money or my readers’ money to advice from this “finger painting” corner of the personal finance community.
4: Risk Parity fans ignore and/or misunderstand their own principles
Isn’t it peculiar that the Risk Parity proponents lecture the rest of us about proper risk budgeting, but then don’t even realize that they put their own portfolios together without much thought about risk budgeting. Take, for example, Frank Vasquez’s Golden Ratio portfolio. Frank wants to allocate 42% to equities, which I find a bit lean, but so be it. He likes to allocate this to both large-cap growth and small-cap value, and he then picks 21% each. That seems ad hoc. The VIOV fund now accounts for 37.2% of the portfolio risk, while the VUG only accounts for 33.1%. Wouldn’t it have been more appropriate to use risk parity in the equity bucket, i.e., allocate a bit more to VUG than VIOV to account for the different volatility levels?
The same story, only more extreme, is present in the Golden Butterfly portfolio. Here too, 20% each goes to large-cap blend and Small-Cap Value. But the risk contributions are 28.6% for large-cap and 37.4% for the much more volatile small-cap fund. Is this intentional? Why do you want more risk budget on the VIOV? And if you thought things couldn’t get worse, the All Weather portfolio takes the cake with two such inconsistencies: First, both gold and commodities get 7.5% of the portfolio weight, but gold adds much more to the portfolio risk (5.8%) than commodities (1.6%), and that doesn’t even count the fact that part of the commodity index includes gold, so get even more gold risk through the backdoor. We find a similar mismatch in the bond portion: Long-term bonds account for 48.5% of the risk, but intermediate bonds only 8.5%.
So, not only do the across-asset-class risk budgets deviate wildly from Risk Parity, but even the within-asset-class risk budgeting is completely ad hoc. It’s like they pull these portfolio weights out of their nose, risk budgeting and Risk Parity principles be damned. Either the inventors of these Risk Parity strategies are so thick that they didn’t notice it, or they knew but were too lazy or incompetent to properly calculate the weights according to Risk Parity principles – you be the judge.
Side note: all these results clearly depend on the return window you use for the calculation of the VCV. Quantitatively, the results will differ if you use a different horizon, different VCV construction methods (e.g., exponentially-weighted moving average, GARCH, TGARCH, etc.), but qualitatively, they all produce similar inconsistencies.
5: Risk Parity looks unattractive for retirees
Frank Vasquez proposes a 5%+ safe withdrawal rate due to the purported improvement in the diversification of Risk Parity. I disagree with that claim and recently wrote about this topic in my Safe Withdrawal Rate Series, Part 64. Risk Parity alone produces mostly disappointing results in safe withdrawal rate simulations for early retirees. You can significantly improve the results by adding Small-Cap Value stocks. But then any improvement in retirement safety is not due to Risk Parity, but only because of the stock-picking alpha inherent in the Fama-French SMB and HML factors. Actually, a simple 75/25 portfolio with a mild SCV bias would have easily outperformed the Risk Parity + SCV portfolio. So even if you’re truly convinced that the Small-Cap Value style comes back into fashion, you’d be better off skipping the Risk Parity method and rather implementing SCV into your simple 75/25 portfolio. From the retirement safety perspective, the Risk Parity really faces a catch-22: If SCV works, you should do 75/25 + SCV rather than Risk Parity. If SCV doesn’t work, the simulation results look worse than under the traditional 75/25 portfolio.
But just for the record: I showed in my post last year (Part 63) that mixing in a momentum strategy can improve the SWR simulation results. Only 10% is a bit too lean, but 50% momentum and the remaining half in the S&P 500 had very impressive results. But again, no Risk Parity is necessary to achieve that.
6: Recent Risk Parity returns look disappointing
I have a very simple rule: if you want to show me your portfolio allocation expertise, don’t bother showing me simulated returns of how your portfolio would have done in the past. I can top that by pretending I had invested in Nvidia ten years ago, which could have easily afforded me a 20+% safe withdrawal rate. If you deviate in any way from a very generic allocation like 60/40 or 75/25, only actual live returns matter. So, I looked at the returns of the eight strategies that Frank Vasquez posts on his site Risk Parity Radio. For the longest time, he published his live strategy returns in table format. The last time I saw detailed returns was in early 2025, and I was able to take screenshots of the monthly returns between July 2020 and February 2025.
But Frank has since discontinued that transparency, likely due to disappointing results. So, I took Frank’s returns up to 2/2025 and added the simulated returns via the testfol.io links he provided. One other major assumption: Since the Levered Golden Ratio started a year later, I backfill the first twelve months with the testfolio simulations, so that all strategies have the same starting point. I ignore the OPTRA strategy because that one only started in 2024. So I am left with seven Risk Parity strategies. I also ignore the partial month of 7/2020 and start the simulations on 7/31/2020. I also construct my 75/25 and 60/40 portfolios. For people who are interested, I posted the return data on Google Sheets; please see here. Please note that this sheet is read-only for you.
Here are the return stats:
- The Risk Parity portfolios look disappointing in the time period that Frank actually traded them with his own money. In all relevant stats, they often significantly lag the venerable 75/25 portfolio.
- The Golden Butterfly and Golden Ratio portfolios did OK, though they still slightly lag the 75/25 in almost all stats: lower return, lower Sharpe ratio, longer drawdowns. Admittedly, the Golden Butterfly portfolio beats the 60/40 and 75/25 in one single stat: it had a 23.92% drawdown, instead of 24.68% and 25.87% in the 60/40 and 75/25, respectively.
- Contrary to the narrative of better risk budgeting and more diversification, the Golden Ratio and RP Ultimate portfolios have equity betas similar to a 60/40 portfolio. So, you’re not at all shielded from equity volatility.
- The other five Risk Parity portfolios are all a complete clown show:
- All Seasons, Accelerated Permanent Portfolio, and Aggressive 50-50 still have not recovered from the 2022 bear market. Their drawdown is still ongoing, and this is just the buy-and-hold portfolio; accounting for withdrawals, the results are worse.
- The Sharpe Ratio of these five Risk Parity portfolios ranges from -0.10 to 0.28, which is pathetic considering that we captured two bull markets and only one shallow bear market, not even a recession!
- The volatility and drawdowns of the Accelerated PP and Aggressive 50-50 are terrible.
- The leveraged Risk Parity portfolios (Accelerated PP, Aggressive 50-50, and Leveraged Golden Ratio) all have volatilities and S&P 500 betas higher than the 75/25 portfolio. So much for building “resilient retirement portfolios.”
- If we model 5% annualized withdrawals, the 60/40 and 75/25 portfolios do much better than the Risk Parity portfolios. Six years into retirement, only the Golden Butterfly and Golden Ratio portfolios hold up reasonably well. The worst disaster is the Aggressive 50-50 portfolio, which now has lost 52% of its initial value (CPI-adjusted). If you keep up 5% withdrawals, you will run out of money well before the 30-year mark. I doubt you will even make it to the 15-year mark.
In the chart below, I plot the drawdowns over time: CPI-adjusted returns, but only buy-and-hold, no withdrawals yet! The 60/40, 75/25, Golden Ratio, Golden Butterfly, and All Seasons all had similar drawdowns in 2022, nearly down to the %. But the Risk Parity portfolios took longer to recover. As mentioned above, the All Seasons portfolio is still underwater from the 2022 bear market. The more adventurous Risk Parity portfolios fared even worse. For example, the Aggressive 50-50 had a 60+% drawdown and is still more than 35% below the peak. I just hope no one used this garbage portfolio in their own retirement.
Drawdowns look even worse if you model the Risk Parity portfolios with regular monthly withdrawals. Here are the portfolio values over time if you had retired in late July 2020. Only the 75/25 portfolio recovered from the drawdown. 60/40, Golden Butterfly and Golden Ratio also still look OK, but most other portfolios are seriously damaged and will likely run out of money unless the current bull market continues indefinitely.
Risk Parity fans might object that the last six years are not really a representative sample. I agree with that, but not in a way they will like. In these 71 months we’ve experienced most of the 4/2020-12/2021 bull market, an entire bear market 1/2022-10/2022, and the subsequent amazing bull market since the October 2022 trough. So, we’ve seen more than one full market cycle. If anything is not representative, it’s that Risk Parity performs so poorly over a slightly biased sample with essentially two bull markets and only one shallow, garden-variety bear market without a recession. Wait until you run this in retirement and you actually experience five bull markets and five bear markets. You will likely do a lot worse!
Also, the bear market, which was inflationary and certainly took a toll on the simple stock/bond portfolios, should have catapulted the Risk Parity portfolios ahead of the 60/40 and 75/25. But Risk Parity sputtered, despite purportedly hedging against all the different macroeconomic risks. If your exotic Risk Parity portfolio gets clobbered during such calm times, wait until we go through the next bear market! Can you imagine how the 3x leveraged equity ETFs do if we ever were to go through a repeat of the dot-com crash?
So, long story short, I understand why Frank Vasquez needed to take down his live sample portfolio return stats. They are such an unmitigated disaster that now he only links to the testfol.io simulated results that start with the backfilled data (mostly in January 2000). But his live returns are atrocious. Especially considering that he wants to withdraw not 5% but 6%-8% of some of his now-decimated portfolios, he will face the difficult decision when he wants to retire (pardon the pun) some of the underperforming portfolios on his blog. Unsavory financial actors sometimes play this game, i.e., start multiple sample portfolios and run them for a few years. Then quietly retire the ones that performed poorly and market their financial prowess with the one or two that did well. But Frank is really zero for seven in his sample.
Most importantly, stay away from all of the exotic Risk Parity portfolios peddled by Frank, especially the ones involving leveraged ETFs.
7: Risk Parity portfolios are inefficient
Risk Parity’s inefficiency doesn’t come only from some of the terrible ETF choices, like expensive leveraged ETFs with a lot of churning and transaction costs (UPRO, TMF, UDOW, UTSL, etc.) or overpriced and underperforming trend-following ETFs (e.g. KMLM, DBMF, etc.). There is another way in which Risk Parity violates fundamental financial principles. Because we consider only the variance-covariance (VCV) matrix in portfolio construction and ignore expected returns, we are bound to overweight low-return assets and underweight high-return assets. The effect of this assumption is that with almost mathematical precision, Risk Parity portfolios must be inefficient.
In fact, one can prove mathematically that a Risk Parity portfolio lies on the efficient frontier if and only if all assets have the same Sharpe Ratio and all correlations are equal. This is not the case in the real world. Equity ETFs have much higher Sharpe Ratios than commodity funds. Correlations are very different across asset classes: likely negative between commodities and bonds and slightly positive between equities and commodities. Because the necessary and sufficient condition for efficiency is violated in all practical applications, we can be assured that Risk Parity is mathematically, certifiably inefficient.
Let’s look at the following numerical example to quantify this inefficiency. Assume we take as our asset universe the equity funds SPY (S&P 500 = large-cap blend), VIOV (Small-Cap Value), and VUG (large-cap growth), SHY (T-bills), IEF (intermediate Treasury), TLT (Long-term Treasury), GLD (gold), and GSG (commodities). I use the 10-year backward-looking risk matrix and assume the following (nominal) expected returns going forward:
- Equities: SPY=9%, VUG=8.5%, VIOV=10%. Essentially, assume 2% inflation, 7% equity returns for the S&P 500, a little higher for small-cap value, and a little lower for large-cap growth.
- Fixed Income: T-bills = 4%, intermediate bonds = 4.5% and long-term bonds = 5%.
- Commodities: both GSG and gold at 4.5%, so about 2.5% real return.
Let’s plot the efficient frontier and also the expected returns and risk of different portfolios. For the Risk Parity portfolio, I assume 1/3 of the risk budget coming from equities (thus, 1/9 shares from each of the equity funds), 1/3 of the risk coming from fixed income (1/6 each from IEF and TLT), and 1/3 of the risk coming from commodities (1/6 each from GSG and GLD). Notice that this portfolio skipped the T-bill fund. I also include one “Risk Parity w/SHY” portfolio, which will spread the fixed income portion equally among the SHY, IEF, and TLT funds, but even though SHY only has 1/9 of the risk budget, it gets more than 50% of the weight under Risk Parity, due to the low volatility of the T-bill return series. Not a very sensible portfolio; please take that with a grain of salt. In any case, here’s the efficient frontier chart; see below:
- The Risk Parity portfolios (both actual risk parity and the two so-called risk parity portfolios) lie significantly inside the efficient frontier, i.e., they generate inefficient allocations. In other words, we could have found allocations with lower risk and/or higher expected returns. This is an artifact of ignoring expected returns and only focusing on risk.
- Also noteworthy: the risk parity portfolios all have relatively low expected returns. Not surprising because the equity portion is so low. What’s worse, though, is that even the risk-adjusted return (Sharpe Ratio) is also low. So, if Risk Parity proponents now try to use leverage to increase the expected return (e.g., four of Frank’s portfolios: Accelerated Permanent, Aggressive 50-50, Levered Golden Ratio, and OPTRA), you only lever up the inefficiency, especially when using the expensive levered ETFs like UPRO, TMF, etc.
- I also plot the efficient frontier when only using the S&P 50o and intermediate bonds. True, that’s not exactly efficient, but it doesn’t matter because at the spot of the 75/25 portfolio, we’re still really near the overall efficient frontier.
Ignoring expected returns and potentially reserving too much space in your portfolio for assets with low average returns makes Risk Parity a suboptimal investment. So, let’s repeat everybody: Risk Parity is an inefficient portfolio allocation method.
8: We don’t have enough uncorrelated assets!
The ultimate dream of financial professionals is to find uncorrelated return sources and then use them to build ever more impressive and efficient portfolio return stats. If you could find 100 uncorrelated return streams, each with a Sharpe Ratio of 0.5, then an optimally weighted portfolio would fetch a Sharpe Ratio of 0.5 times the square root of uncorrelated assets, i.e., 5.0. What’s not to like about that? The sad state of the world is that most retail investors are really stuck with two, maybe three major asset classes that are worthwhile feeding into a Risk Parity process. Number 1 and 2 are equities and bonds, respectively, with decent Sharpe Ratios, about 0.3-0.4 in equities and 0.2-0.3 in bonds. For very technically astute and experienced investors, I recommend trading the volatility premium; see my options trading landing page.
But beyond that, as a retail investor, you quickly run out of options (pun intended). A long time ago, you could have used some stock-picking flavors (small-cap, value, momentum, quality, etc.) as additional factors. However, they became so popularized that any reliable outperformance is now arbitraged away. Commodities in general, and gold in particular, come to mind, but both are highly volatile and have low average returns. Both commodities in general and gold in particular have occasional decade-long drawdowns. For example, gold had a real, CPI-adjusted drawdown between 1980 and 2024 (536 months, almost 45 years), with a peak-to-bottom fall of 83%. Commodities have certainly rallied recently, but make no mistake: Commodities peaked in 2008 and have been in an 18-year drawdown, with a peak-to-trough drop of 89%. Despite the recent rally, commodities are still almost 70% below the 2008 peak today. So, commodities in general and gold in particular don’t exude much confidence. I’m fine with my 75/25 portfolio and supplementing that with my options trading alpha!
Of course, the Risk Parity crowd will show you portfolios with numerous additional ETFs, say, preferred shares, high-yield bonds, international stocks, emerging market stocks, REITs, China A shares, etc., but these aren’t exactly uncorrelated assets. They are simply combinations of your existing asset classes. For example, preferred shares are a mix of stock and fixed-income returns, and not at all a new, uncorrelated asset class (pro tip: check their performance during the Global Financial Crisis!). Emerging markets are a mix of equities and commodity exposure. So, recycling existing financial risk factors and throwing in ever more low-quality, high-expense exotic ETFs with low expected returns will not help the Risk Parity cause; it will only make the low-expected-return problem even worse.
9: Bonus Lie – You can create extra returns out of nowhere (new: 7/31/2026)
The newest Risk Parity shtick I’ve heard is that through an intriguing artifact in the construction of geometric average returns, you can actually increase the expected return of a diversified portfolio. For free and out of nowhere. Basically, Frank Vasquez wants to use this technique to boost the expected returns of Risk Parity portfolios. Specifically, during a recent discussion, Frank claimed that he miraculously and significantly raised Risk Parity returns by applying the principle of Shannon’s Alpha (sometimes called Shannon’s Demon, but I try to avoid that term and the negative connotation). It has to do with the mechanics of how geometric returns aggregate over time and how volatility penalizes those returns (remember “mu – 0.5 sigma^2”). To keep things simple and keep the flow, I will leave out the mathematical details, though.
First, Shannon’s Alpha does not apply here because it’s used in cases where you have only one single risky asset plus a risk-free asset and where frequent rebalancing back to target weights seemingly creates phantom returns, as if out of nowhere. But a similar principle, volatility pumping, would generate a qualitatively similar effect in portfolios with several risky and imperfectly correlated assets, which could be applicable to Risk Parity. So let’s get our terminology straight and study that effect. One application would be to spread the portfolio’s equity portion across several different ETFs. How much extra alpha can I generate with volatility pumping? How about very little in the best case in theory and negative alpha in practice?
First, even in the ideal case with perfect laboratory-condition return patterns, the alpha will be quite small. Let’s take an example of two assets with log-normal returns, each with 18% volatility and a correlation of 0.85. The continuous rebalance premium is 0.25 x 0.18^2 x (1 – 0.85) = 0.001215 = 0.1215% annualized. It’s not zero, but a far cry from the 6% or so additional returns that some people generate with their ticky-tacky numerical examples on the internet, usually assuming far more volatile returns (50%+) and zero correlation. So, let’s say that a 75/25 portfolio might have a 5% real expected return p.a. over the long-term (6% from equities, 2% from bonds) and a Risk Parity portfolio like Golden Butterfly with expected returns of 6%, 6%, 2%, 1.5%, and 2% real returns in the VTI, VIOV, TLT, SHY, GLD ETFs, respectively, will get you a 3.5% expected return. Add to that a rebalancing alpha of 0.12% spread over 40% of the portfolio, then you get only about 0.05% extra return. That’s not enough to overcome 1.50 percentage points in lower returns relative to the 75/25 portfolio.
But it’s even worse, because the tiny volatility pumping effect would be swamped in the real world by at least a number of other effects:
- Potentially higher expense ratios when shifting from a cheap, blended, and broad ETF to smaller, exotic funds, e.g., one growth and one value fund. Or one large and small-cap fund, etc.
- Transaction costs from rebalancing continuously, i.e., commissions, bid-ask spreads, tax drag, time and effort, bookkeeping and tax-reporting costs, etc.
- Other effects, like momentum in asset returns, can make frequent rebalancing counterproductive. In other words, if there is even the slightest asset return momentum, there is an advantage to letting deviations from target weights run rather than rebalancing too frequently that can far outweigh the tiny vol pumping alpha. And notice that there is noticeable asset class momentum, as I pointed out recently in the SWR Series, Part 63.
To showcase how there is no alpha left after accounting for all those pesky non-laboratory-perfect conditions, let’s compare the returns of holding just a passive 100% portfolio of the IVV ETF (iShares S&P 500, blended) and two equal portions of S&P 500 Value and S&P 500 Growth, which in equal shares make up the S&P 500 again. Let’s check the simulation results. I’m using the Frank-Vasquez-approved site testfol.io, which he uses for his own simulations. I calculate the returns between 5/31/2000 (first available full month for all three ETFs) and 6/30/2026, both for a 100% IVV and a 50% IVW + 50% IVE with different rebalancing assumptions: 1) daily, 2) weekly, 3) monthly, 4) quarterly, 5) semi-annually, 6) annually, 7) every two years, 8) every five years, and 9) no rebalancing at all. The link to the simulations: Part 1 and Part 2. I currently have the free version of testfol.io only, where I can simulate only five portfolios at a time; hence, the stats are spread over two parts. But I quite like the simulation tool, so I might even sign up for the paid version.
Here are the results:
- There is no gain from splitting up the IVV fund into two. Quite the opposite, the IVE+IVW portfolio performs significantly worse at all rebalancing frequencies, between 0.16% and 0.30% annualized.
- The differences cannot be explained by the expense ratio differential alone, which is 0.15%.
- There is no clear trend in how different rebalancing frequencies help or hurt you. The vol pumping scheme should work best with the most frequent rebalancing (i.e., daily) and worst without rebalancing. True, there is a drop-off between 2Y to 5Y and to no rebalancing, but daily rebalancing also works poorly. Thus, volatility pumping cannot cause these variations because the trend would be a clear monotone decrease in performance: highest benefit from vol-pumping with daily rebalancing, and then worse results as the rebalancing intervals rise.
- Needless to say, the simulations here ignore transaction costs. If you were to add those, you would eliminate any gains from more frequent rebalancing.
- Really delusional proponents of Risk Parity and Volatility Pumping may still argue that it works because the no-rebalancing portfolio looks worse, then claim that the static, blended index portfolio must also look worse than the rebalanced portfolios, right? Wrong: the static index is the IVV ETF, and that performed best.
Well, Frank will now object that Volatility Pumping works better among the entire portfolio, not just the 40% equities. And it’s true that the vol pumping effect is stronger with less correlated assets. OK, let’s check that case too. Here are the simulation results for three portfolios: Golden Butterfly, 60/40, and 75/25. We have data from 1993 until 2026. I copied and pasted Frank’s own testfolio link and used the same assumptions ($10,000 starting capital, adjusted for inflation, $42 a month in distributions, etc.) to simulate all three portfolios at the nine rebalance frequencies (1=daily, 2=weekly, …, 9=never). The links to the simulations are here: GB Part 1 and Part 2, 60/40 Part 1 and Part 2, 75/25 Part 1 and Part 2.
Here are the return stats. Notice that for this exercise with cash flows, we should look at the money-weighted returns (MWRR), i.e., the internal rate of return (IRR) of the cash flows over time (initial investment, withdrawals, and final portfolio value). Though the CAGR would produce qualitatively similar results, shifted by just a few basis points. Again, the results are disappointing:
- First note that 75/25 beats the Golden Butterfly. Just saying!
- There is no strong relationship between the rebalance strategy and the IRR. If anything, you get a little bump by going from annual to every two years, and you get the best results from never rebalancing. (opposite effect as with the IVV vs IVE+IVW, go figure!)
- Between the more standard rebalancing assumptions, anything between daily and yearly shows no obvious trend. The lines are horizontal with some variation due to noise.
- Also notice that the lines for 75/25, 60/40, and GB are just mostly parallel shifts. If there was any effect of volatility interacting with the rebalance frequency, it appears to impact all three portfolios about the same way, so any positive effect on the convoluted Risk Parity portfolio also shows up in the simple 75/25 and 60/40 portfolios. So, aside from the fact that your vol pumping can’t compensate for the massive expected return difference, it certainly can’t do so if vol-pumping also lifts the expected returns of the 75/25 portfolio.
To summarize, the Shannon Alpha doesn’t apply here. Volatility Pumping may apply but is a red herring, because a) it’s too small to matter, so momentum matters more than rebalancing alpha, and b) even if it were larger, vol-pumping also lifts the 75/25 expected returns, so it cannot bridge the massive expected returns between Risk Parity and traditional portfolios. Also notice the inconsistency: Frank only rebalances his Risk Parity portfolios once a year, invalidating any use of vol-pumping. This is a recurring pattern we see with trolls: they pick up something on the internet. They don’t understand the math, statistics, finance, or economics behind it. They don’t understand the limitations. They take it out of context and then run with it!
Risk Parity has become a Risk Parody!
This vol-pumping issue is a prime example of Brandolini’s Law, a.k.a., the Bullshit Asymmetry Principle. It takes me about ten times, maybe 100x more time and effort to refute the nonsense spread by internet trolls than it takes them to come up with it. Of course, I can’t right every wrong on the web. But this was a fun project!
Conclusions
The general idea of Risk Parity isn’t entirely bad; I’ve occasionally found it useful in intra-asset-class allocation decisions, e.g., allocating a certain percentage to various equity styles with very similar Sharpe Ratios and correlations, where the Risk Parity method is very close to efficient. But most investors should stay away from it.
Risk parity can potentially do the most damage to folks still accumulating because your equity allocation will likely be too meek. Most young investors will do best with a 100% equity portfolio: they should actually embrace volatility and use the occasional deep drawdowns to dollar-cost average, i.e., use Sequence of Returns Risk to their advantage. That’s what I did in my personal accumulation history; in the roughly 18 years of my high-earning career at the Federal Reserve and on Wall Street between 2000 and 2018, the S&P 500 performed below average (about 3.2% if adjusted for inflation and including dividends), but by keeping up with regular investments during the steep drawdowns in 2002/3 and 2008/9, I vastly improved my investment results, buying the dips.
For retirees, not all risk parity portfolios will be that bad. Especially over shorter horizons, say 30 years, some of the Risk Parity portfolios fared all right. The Golden Butterfly and Golden Section portfolios would have performed about as well as a 75/25 portfolio in the standard Bengen or Trinity study, i.e., a 30-year horizon and zero final asset value if you simulate the returns with the best-case scenario with historical Fama-French factor returns. That said, even in this scenario, with very unrealistic expectations that small-cap value outperformance could repeat, a 5% withdrawal rate would seem too aggressive. And once you scale back the unrealistic assumptions about the small-cap value premium, you’re back to roughly the same safe withdrawal rate as with a simple 75/25 portfolio.
Intriguingly, the “OG” Risk Parity portfolio, i.e., All Weather/All Seasons, as popularized by Ray Dalio and Tony Robbins, and likely the closest you can get to the mathematical Risk Parity weights, would have performed quite poorly in every arena: over both the 30-year and 50-year retirement horizon in long-term simulations, as well as in live returns on Frank’s webpage. It’s the stereotypical throwing-out-the-baby-with-the-bathwater issue, i.e., you reduce short-term volatility, but you replace it with the long-term risk of running out of money due to poor average real returns. The All Seasons portfolio hasn’t even recovered from the 2022 bear market!
In my opinion, Risk Parity is a useless concept unleashed on unsuspecting, naive retail investors hungry for financial and technical buzzwords. Stay away from it and simplify your portfolio. 100% stocks is all you need while accumulating, and 75/25 is likely all you need in retirement. That portfolio has been the most robust and reliable portfolio in retirement for the last 150 years. Most complications beyond that simple portfolio will be a crapshoot in the best case and a drag on your retirement safety in the worst case. For the record, I do recommend an options trading overlay to the technically skilled investors, as described in the series on that topic, but it’s not necessary for the typical investor.
Do I believe I will convince Frank or some of the other Risk Parity true believers? Probably not, because they painted themselves into a corner. Frank called his podcast “Risk Parity Radio,” so I don’t think he’s going to walk back anytime soon. He falls into the “foolish consistency” trap he so often points out. But maybe I can save a few unsuspecting investors from the foolish finance trap that is Risk Parity.
Update 8/15/2026: My response to Risk Parity Radio ep. 532
Frank posted his response. Or non-response. For example, he doesn’t link to my specific blog posts, which is quite unprofessional. Obviously, he doesn’t want his listeners to go to my site and check out the well-researched posts. In contrast, I have no hesitation pointing to his work, because I have nothing to hide from my readers. He can’t address any of the specific issues I raised. His rant is a collection of red herrings and strawman arguments. My detailed responses to Frank’s podcast:
Fake comment: I had an unfortunate incident with one of Frank’s supporters who made a comment under Frank’s name. I always take down inappropriate comments as soon as I become aware of them, as I did here. Frank makes a lot of hay out of this fake comment issue because Frank cannot mathematically refute anything I wrote. Frank doesn’t have the math, statistics, and finance skills to discuss at my level; hence the distraction. Also noteworthy: that same poster, from the same IP address, threatened me with violence; see these two comments:
“you better not f* up or there will be consequences in real life. I hope you’re mature and responsible enough to understand that and own it”
“So Karsten, you better be right because if you aren’t, I’m going to find you in real life!”
So, I would ask Frank to scale down his toxic and hateful rhetoric and also tell his supporters to chill when they post on my site.
Small-Cap Value discussion: The SCV discussion is a strawman argument. As I mentioned in both of my Risk Parity posts, they are not about Small-Cap Value (SCV) stocks. If you believe that SCV outperforms as impressively as it has historically, then, as I showed in my post, you should use a 75/25 portfolio with SCV bias, which would have handily outperformed both the standard 75/25 portfolio and Risk Parity as well. So, to answer Frank’s question, if I now want to attack Morningstar because they see SCV potentially outperforming, the answer is no. My beef in this discussion isn’t about SCV; it’s about Risk Parity. Frank likes to conflate these two issues and then claim he has the support of all sorts of internet influencers when he doesn’t.
Also, I’m writing this again for the gaslighting-damaged Frank fans: I don’t hate value stocks. I allocate 50% to value and 50% to growth in my blended index funds. I can even see value stocks partially catching up again, offsetting some of the losses over the last 20 years. I have simulated that scenario in my SWR Series, Part 62, and even with a generous SCV premium (small stocks at 1.2% annually & value stocks at 0.8% annually), the extra risk isn’t worth the extra return. I wouldn’t bet my retirement on it; hence, I employ an unbiased portfolio in blended index funds that covers all bases, i.e., a continued AI-fuelled bull market, a revival of value stocks, and everything in between.
Gold: The same issue applies here: I wrote a post in my SWR Series, Part 34, where I simulate safe withdrawal rates when adding 10-15% gold. Frank has publicly admitted that he likes that blog post. Well, either he lacks the reading comprehension to understand it, or he didn’t read it til the end, where I again point out this important caveat:
“So, the widely-cited exotic [Risk Parity] portfolios don’t exactly deliver any notable improvement in the safe withdrawal stats. I’d stay away from them! If you want to use gold to hedge against sequence risk, shift some of the equity portion [of your 75/25 portfolio] into gold. But stay away from the “sexy” portfolio allocations recommended by the internet gurus and motivational speakers!”
Risk Parity doesn’t have a monopoly on gold. If gold (or SCV or momentum, or any other exotic flavor) has a positive marginal effect, it doesn’t create an automatic rationale to do Risk Parity. In all cases I have studied so far, you’re still better off adding that style to a simple 75/25 portfolio, but avoiding Risk Parity.
Bill Bengen’s new book: Bill Bengen doesn’t recommend value stocks, only small-cap stocks. With the historical outperformance of small-cap stocks (which he also extrapolates to micro-cap stocks), Bill believes he can squeeze out more expected returns. The same caveat as with SCV and gold applies again: If you believe the small-cap revival hype (I don’t, FYI), shift some of your 75/25 portfolio into small-cap stocks, but stay away from Risk Parity. Also noteworthy, Bill Bengen very explicitly notes that the elevated equity price/earnings ratios require caution and lower withdrawal rates, refuting Frank’s longstanding religious belief that the CAPE doesn’t matter. Finally, much of the increase in Bengen’s SWR isn’t due to a better asset allocation, but to shifting the success criterion from 100% safe to mostly safe. No miraculous new research on retirement safety has emerged. It’s mostly shifting the goalposts, accepting higher failure rates, and forcing more flexibility.
Creepy Uncle Frank: Frank discusses my consumption and withdrawal rate patterns at length, as if he knew my life in retirement intimately. He doesn’t, but nevertheless, Frank is serving his listeners creepy stuff like this:
“He’s got this option strategy that he’s been working on and now has opened a financial advisory practice and has registered with the SEC for that. And according to their most recent disclosure, they’ve already got $24 million under management and are charging a 0.8% AUM fee on it. […] He’s just living off dividends and interest and then adding a side hustle to that. And if you’re only spending about 2% of your invested assets, you really don’t need a safe withdrawal rate strategy or any big analysis to do that.” RPR, Ep. 532
I have made this point many times: I initially used my blog as my personal (and public) notebook for early retirement planning between 2016 and my early retirement date in 2018. I could just call it a day now and shut down my blog because I’ve clearly solved my own retirement-safety issue. Out of intellectual curiosity and as a public service to the community, I keep going. I had incredibly perfect timing, retiring in 2018. Today’s retirees, especially those who don’t have the same post-retirement earnings options I had, may not be as lucky and may still benefit from my content.
Regarding the “stage 4 retirement police” comment, I should note that I make a modest amount of money from my business activities, but the bulk of my withdrawals comes from my portfolio. I consider neither the blog, nor the advisory business, nor the options trading income to be a permanent and guaranteed income stream, at least not for decades into retirement and certainly not for the entire duration of my retirement. I enjoy the extra income as long as it lasts, but I’m prepared to shut it down to zero and live completely off my portfolio at any point. This is how I structured my retirement budget, factoring in a safe withdrawal rate of about 3.25-3.50% from the portfolio to sustain a 50+-year retirement. If I make something extra, it’s only a windfall and will not materially increase my consumption behavior. For folks who are trained in economics, you will recognize this as the permanent income hypothesis. I’m sure Frank studied it too but never internalized it.
So, Frank’s warnings like “don’t listen to ERN, he lives like a miser” are really laughable. It’s just gaslighting. My wife and I live very luxuriously, and we don’t constrain our spending. It’s made-up fantasy land, like so much else on his podcast.
Summary: Frank suffers from “confirmation bias disease.” He reads bits and pieces from around the internet, runs with the parts he likes, and ignores everything he doesn’t like or doesn’t understand. This has been true for gold, SCV, Bengen’s book/small-cap stocks, Shannon’s Alpha, and many more issues.
Just like a liar can’t keep all his lies straight, Frank’s “analysis” is rife with logical inconsistencies: insisting on mean-reversion in valuations within equities, i.e., according to Frank’s crystal ball, cheap SCV stocks will outperform, but the CAPE ratio, a valuation metric for the overall stock market (and statistically much more reliable and stable than SCV, by the way) is suddenly completely irrelevant, because he found one counterexaple in 2011 (which isn’t really a counterexample). Or, invoking Shannon’s Alpha/Vol Pumping, which relies on fast rebalancing of portfolio weights, but insisting also on rebalancing only once a year in his sample portfolios and simulations. Or insisting on Shannon’s alpha, which relies on an iid return distribution without serial correlation, but then also promoting momentum strategy ETFs like KMLM or DBMF, which heavily rely on trending returns, which would create a negative alpha in Shannon’s model.
So, Frank has really painted himself into a corner. He can’t refute anything I wrote. Maybe he should just focus on different aspects of personal finance and leave the quantitative parts to, you know, retired quants like me.
Thanks for stopping by today! I’m looking forward to your comments and suggestions below!
Title Picture Credit: pixabay.com
There should be 1000’s of “Thank you notes sent to you.”
Just to make sure that we remember this.
“100% stocks is all you need while accumulating, and 75/25 is likely all you need in retirement. That portfolio has been the most robust and reliable portfolio in retirement for the last 150 years.
Thanks again and again and again! (Not much more one can say.)
Couldn’t agree more with this sentiment. Thanks again Karsten for your sterling work.
Thanks for the feedback! 🙂
Thanks, Karl!
Thank you for this. I knew Risk Parity was off but didn’t have the deep knowledge to refute it among all the online supporters. Great work!
Thanks! Same here: sometimes intuition matches the science and the facts. I still got it in me. So do you!
I have been tracking “All Weather” since inception and while 2020 it did okay, it got destroyed in 2022 and as mentioned, it still has not recovered. A few comments on portfolio construction that was not mentioned. I have been following Keller’s Hybrid Asset Allocation since the paper as been published and it has done well. What was not mentioned in this post was the tactical strategies that move weighting/and or risk off components in times of stress. It would be good to see that data.
Second, I am waiting for your 70/30 data to see how VOO/VIOV/DBMF would have performed relative to the 75/25 as mentioned in your momentum post.
Last, often what is not mentioned is the drawdown of a 75/25 and Ulcer index. Can people hangin there if the market drops 50 percent? If a retiree has amazed 4 million, would they stay in the market at 2 million? I don’t think so.
I appreciate your post. Always great and balanced and you shake out the bad actors, when warranted .
I’m not familiar with Keller.
What exactly are you waiting for here: “70/30 data to see how VOO/VIOV/DBMF” ? Not sure if I am supposed to deliver anything.
Ulcer index: FYI, the Risk Parity portfolios had worse drawdowns and longer ones too. The idea of a 75/25 portfolio is that a 50% equity market drop will be much less a 50% portfolio drop.
How would a 75/25 portfolio consisting of VTI/BND perform vs your recommended S&P500/ Treasuries portfolio?
According to my SWR sheet, where I post ETF factor exposures, the BND and AGG ETFs behave (roughly) like a portfolio with 10% stocks, 59% 10y bonds, 3% 30y bonds, 28% T-bill
It will behave similarly, but will have some additional stock exposure through the backdoor.
Hi Karsten,
Absolutely great article as I loved these two RP posts so thank you.
But a few questions; when you say 75/25 is all most retirees need you are using the 10 year UST for the 25% fixed income right? So I assume an ETF like IEF would fit well for that? But have you written or modeled other bond allocations for the FI side? Like dividing the 25% allocation between UST LT/MT/ST durations? And if using only one fund would substituting something like Total US Bond Market so etfs like BND or AGG be similar enough to IEF? I’d prefer to use IEF or solely UST but some 401K’s do not offer that option so TBM can be “closest” in terms of duration. I wouldn’t intuitively think it would make a large difference in SWR but curious if you have modeled? Lastly many academic researchers and others recommend TIPS for some or all of FI side, have you written or modeled that and do you recommend also?
Thanks again for the great posts! Much appreciated.
Yes, the 10y UST is closely approximated by the IEF. VGIT is also good (lower expense ratio, but also a shorter duration)
You can play around with different bond allocation. Traditionally, the 10t UST has done better diversifying equity risk than both the 30y and the T-Bill. So, I’m not a fan or barbell strategies.
TIPS haven’t been around for long enough. I like the inflation protection. But TIPS were also less of a diversifier in 2008/9 than UST. So, TIPS may not be that attractive. The worst equity bear markets are when we have deflation (1929, 2008/9), so nominal bonds are still the best in that case.
Excellent, thorough research on risk parity. I really appreciate you doing the deep research on this. The pro crowd has gotten insane. As has the guy you mentioned multiple times. Grateful to hear an evidence based approach to this and that all (most) roads lead back to Fama and French!
Good summary! They have gone insane because insanity is all they have; they can’t argue with theory or data.
So Adding some managed futures and Gold is complicating too much a portfolio? Man, you really think we’re monkeys don’t you? complicating things would be trading options like you do
True.
OMG thank you so much.. I’ll bump up my DBMF allocation to 40% after I read
[-1% for the DBMF (trend-following) ETF. That’s not a typo; a risk contribution can indeed be negative.]
At 40% DBMF will no longer have a negative marginal contribution, just FYI.
Note by Karsten on 7/30/2026: The comment below was not actually written by Frank Vasquez. I agree that this is not Frank’s style. I wanted to just delete the entire thing, but that would also delete the follow-up comments, one by a reader Ryan and one by Frank himself. So I keep the comment and reply below to Ryan and Frank separately.
Well, sadly, this is the world we live in: And this is not even one of those new deep fakes!
Since the initial comment, three additional troll comments from the same culprit (different names but from the same IP address) came in on 7/30/2026. Two of the comments included threats of violence against me and my family, one stating
The other said:
I deleted this garbage and blocked this IP address. This is clearly not the typical ERN blog reader. 99.99% of the interactions I’ve had here have been very cordial, and I hope that violent people like that stay away. Stay strong, everyone!
K.
[original comment below, but redaced]
Karsten,
[redacted]
Best – Frank
If only there was a podcast where you could discuss the differences in methodology… And you have a forum here to address some the parts of this analysis that you don’t agree with. I think we’d all benefit from that discussion. I don’t care where it takes place, just make it happen. Please. 😁
We did a podcast recording yesterday on Bigger Pockets Money with Mindy and Scott. To be published on August 11.
Is it still getting released today? Haven’t seen it yet in Apple Podcast.
I got some bad news: BiggerPockets decided not to publish this episode. Which is too sad; I think I handily “beat” Frank in this debate because I brought the receipts (based on two well-researched 5,000+ word blog posts on the topic), while Frank was stumped and had to resort to rambling and yelling. It’s too bad that it came to this. It’s a version of a “heckler’s veto” when a disruptive podcast guest shuts down the debate.
That’s super disappointing. I don’t regularly listen to Bigger Pockets, but I tune in when I see an interesting topic and I was looking forward to this one. They’re missing an opportunity for an episode that I imagine a lot of people would download.
It’s a real catch-22 for the podcasters: cancel the episode to bury this personal fight, or broadcast it and create more animosity. It’s too bad; we’re missing an informative episode.
Hi Karsten,
Did BiggerPockets tell you reason they are not publishing the recent episode with you and Frank? Thanks.
I will not post the email back and forth here, of course. My theory is that they revisited the recording and realized Frank has gone too far off the reservation. Add to that the fake post issue and the death threats from one of Frank’s fans against me, and the whole topic got a little too hot for them to touch.
wow..Imagine the low level of this conversation for them to decide to not air it. What a shame!
All Frank’s fault, of course. I had a ton of interesting insights.
Well, this is peculiar. It appears that I have an imposter, as I did not make this post in my name. As I mentioned last month (RPR Episode 517), I do not ordinarily follow this blog anymore (or any other blogs for that matter), and only found out about this imposter comment because a couple of listeners emailed me about it yesterday evening.
So I don’t real know why it’s here, but can think of two explanations off the top of my head:
1. Benign Intentions. Someone is trying to play “guardian angel” of some sort to defend me. If so, I appreciate the sentiments, but please make your posts in your own name! I do note that this post appears to be AI-generated. The “m dashes” tend to give that away, particularly at the beginning of some of the sentences. Normal people do not write that way, so you may want to edit those out next time.
2. Malignant Intentions. The post was made to give the impression that I actually read this blog post (or the other one) on July 29 or before. In fact, I have not (nor the other recent one) as of July 30, but will treat it like the other blog posts here that listeners ask me about and answer their questions on the podcast in due course. (As the quote from Office Space goes, “It’s not that I’m lazy, . . . ). In this regard, the time stamp of the imposter post (12:51 pm) is especially suspect. If that is my time, that was 9 minutes before I started recording a podcast for Bigger Pockets Money with Karsten. I certainly was not doing anything but setting up a microphone and camera then. If that is Pacific time, it was just after the recording was finished and I was not whipping up any posts then either.
When you hear that podcast in two or three weeks, you will hear Karsten talk about this and his other recent post a lot, as if I (or Scott or Mindy) would have been expected to have seen them prior to the recording and be responding to those references in real time. And he makes a specific point of referencing them at the end. In fact, it is clear that the release of the two blog posts was designed to coincide with his appearance on the podcast in some kind of coordinated effort on his part.
So be it, but that combined with our imposter here leads one to question the whole apparently coordinated hit-piece enterprise. I mean, what kind of person does stuff like that outside of politics?
Anyway, I’ll deal with the substance of this and the other one on the podcast in the next month or two, and you’ll all just have to wait until then.
(The Real) Frank Vasquez
Sorry to hear that, (Real) Frank! It’s the world of trolls we live in. I sensed the comment from the fake Frank was off because it wasn’t your style; too polite. Take solace in the fact that imitation is the greatest form of flattery. I still haven’t seen any fake “Big Earn” Twitter accounts without the blue checkmark, so I haven’t reached your level of fame yet.
A few thoughts about your comment, especially this part:
You mean to say that in preparation for a debate about Withdrawal Rates and Asset Allocation that will inevitably lead to the topic of Risk Parity, I actually put together my Risk Parity notes, simulations, formulas, Python code, charts, tables, etc., and actually prepared for the debate? Wow, you got me there, Sherlock Holmes! Guilty as charged. It’s called preparation. I prepared for that debate. Unlike you, by the way – more on that later. In fact, had I not published those pieces before the podcast recording but only afterward, then you would have accused me of trying to catch you off-guard and not revealing all my facts beforehand.
Got me there again, Sherlock Holmes; that’s exactly what I expected. Certainly, that’s true for the Monday post that was published 57 hours before the recording, but even the Wednesday post is fair game. I am a prominent researcher on safe withdrawal rate topics, and I know that people who interview me subscribe to my blog feed. Is it too much to ask for people to read my post(s) before the recording? Do I have to send everyone the links to publicly available posts beforehand? I’m neither your nanny nor your tutor; you have to do your homework yourself, buddy!
In fact, I was so courteous as to show my hand ahead of time, but you were too lazy and complacent and didn’t bother to read my posts. To wit, if you had read them, it would have spared you the deer-in-the-headlights moment when I pointed out that your Golden Ratio portfolio was down 27% peak to trough in 2021/22. You yelled at me and called me a liar on the show for pointing out a verifiable fact. I trust the good people at the Bigger Pockets Money podcast to edit out your psycho antics and make you look civil in the published version.
You are insinuating that I coordinated with the fake reply poster (or maybe even accuse me of posting the fake comment myself) to create a “coordinated hit-piece enterprise.” That is outrageous! That’s sillier than the “accusation” that I prepared for the podcast and dared to publish my notes as two blog posts. You’re simply embarrassed that I caught you on the wrong foot in the podcast recording. And now you use the fake comment kerfuffle to distract from the real issues.
Frank,
I’m not sure why you wouldn’t have expected ERN to come with detailed analysis and numbers. Everything he does is detailed analysis and numbers.
The last time ERN debated someone related to his SWR series, he did so with Fritz Gilbert in a respectful but detailed manner, even inviting Fritz to write up his analysis on bucket strategies on this blog. He had previously blogged about why the Bucket Strategy is really just window dressing, but ERN clearly has a constructive and respectful relationship with Franz.
I don’t know why you would agree to debate with ERN without reading his recent detailed thoughts on Diversification and Small Cap Value. Although his analysis on those things is not identical to your Risk Parity portfolio, knowing his analysis on those topics should have left you well prepared to know what arguments he would make related to Risk Parity.
I am sad that Bigger Pockets decided to can the episode. I would have liked to have heard it.
Thanks for this comment. I agree wholeheartedly. I’ve debated Fritz on buckets and Paul Merriman on SCV without incident.
Have I missed something really important? As I understand it, most of the portfolios on the RPR website, especially those that use leverage, were not designed to actually be used and they are not even recommended as such. I understand this to be a feature, not a bug. They are designed simply for educational purposes to show how different types of (usually) non correlated assets act relative to each other in various market environments. To criticize them for how they’ve performed return wise (and especially over a very limited time period) seems to miss the point entirely about why they were created in the first place.
Also, as far as I know, the risk parity website is only encouraging the application of RP principles to decumulation portfolios and actually encourages 100% equities in the accumulation phase.
Admittedly, I have only quickly perused the whole RPR website and this SWR post but seems like there are some red herrings at play
I don’t read it that way. In fact, it was my olive branch to include the possibility of accumulation because there you can survive worse drawdowns and use dollar-cost averaging. If you had any sense of the history of RP, you’d know that it was not specifically designed for retirement. But hey, if you take away accumulation and only want to allow Risk Parity in retirement, that makes the approach even less useful.
Yes, you definitely have missed something. Of course, the sample portfolios were intended to market as being used. He used $10,000 of his own money to fund them. If they had outperformed, Frank would be running around touting their success in building “resilient retirement portfolios.” But alas, now that they underperform and they are, well, just for fun, nothing to see here. This is peak gaslighting.
I went and listened to some periodic episodes starting from the beginning and heard the following:
1. He seems to regularly refer to the leveraged portfolios as things to “not try at home” or only for those who “have a gambling problem.” I’m now more sure that he included those originally just for educational purposes, not as recommended portfolios one should follow in retirement.
2. His website underwent a major renovation in early 2025 with the help of a Canadian listener to streamline the website. The charts you suggested he withdrew to hide poor performance seem to have been taken out at that time for more benign reasons.
Perhaps your cynicism is correct, but I have not found any evidence that your two arguments I referenced above have merit. Indeed, the contrary.
1: All I can say is that after the leveraged portfolios have bombed so spectacularly in 2022, yes, he will certainly downplay the seriousness of those portfolios. The question is, how did he market those portfolios before the bear market of 2022? Please show me specific quotes from the podcasts in 2020 and 2021, where he downplays these sample portfolios and explicitly warns his listeners, “Please don’t do this at home.” Until then, I’d need to assume that the various portfolios were most certainly set up to show how well they do; otherwise, why would he have put up the $10,000 of his own money (each!!!)? If they had outperformed, he would be running victory laps now. But this whole issue is just a red herring, because even his preferred two styles, Golden Ratio and Golden Butterfly, sucked in the live return period.
2: I didn’t see any effects of any major renovation on the specific portfolio info pages. The only thing that changed there is that he took down the tables with the monthly returns. They were there before; they disappeared sometime after February 2025, because I still have his actual return data from that time. I even took and saved screenshots of the return tables from 2020 to 2024.
So, your two arguments are pretty weak.
I’m a chimp with a calculator compared to you guys, so massive respect to you and Frank. I love your work, at the same time I also feel like Frank is a friend that I have never met.
Your’re so harsh, rather direct that at somebody like Dave Ramsey and his 8% withdrawal rate.
You leave out lot and manage to twist things in a way that I find unnecessary. Play the ball, not the man. Frank calls some of his portfolios “hideous experiments”. If somebody didn’t know his work and only read yours they’ll think he is selling these portfolios.
Given the FUD this has created in our little community, I can only imagine the FUD that exists for a normal person that uses a financial services company.
Is it too late to just “be lekker”?
One similarity between Karsten and Dave Ramsey is that they are not holding back when they feel something is wrong. Karsten carefully explains why he feels Frank’s Risk Parity portfolios don’t work. Now everyone can make their own mind. I would rather people not sugarcoat when my money is at stake.
Thanks! Yes, I’m passionate about this, just like Dave Ramsey is about his field. But I use math to back my views.
You’re underselling yourself. If you’re a chimp with a calculator, then Frank would be an ameba with an abacus. I have never seen, read, or heard anything from Frank that makes sense from a quantitative, mathematical, or statistical perspective.
I am not sure why you have to be so uncivil, rude, and condescending; resorting to ad hominem attacks, while also twisting Frank’s work, his words, and his podcast. Frank, in virtually every talk about the portfolios, only refers to the Golden Butterfly and Golden Ratios as “sample” portfolios that most people should hold in retirement. He does not advocate holding the others for retirement.
The All Weather he calls a “reference” portfolio and that it is too low on stocks to be viable long term and that you would need to apply leverage to it like the hedge funds do in order for it to produce enough returns. He calls the other portfolios “experiments” and “hideous experiments” that he would not recommend people use. But you clearly have not actually listened to his podcast. He is also intentionally “abusing” them with higher withdrawal rates to learn more about how they perform.
Frank is using the Risk Parity concepts in a way that normal investors can use without needing to apply leverage. David Stein would refer to these portfolios as “Role-Based.” Maybe that is a better term for you? Frank has never claimed these are traditional academic based volatility matching. He is not interested in them on that basis. His interest is in applying the theory for average do it yourself investors to be able to spend more money in retirement with shallower and shorter drawdowns.
He doesn’t claim SCV will out perform the market, he expects it to perform similarly to the overall market but at different times. With periodic rebalancing, you get the effect of selling high, buying low, and increasing your overall returns on a risk adjusted basis. Frank says rebalancing 1x a year is plenty or do it on rebalancing bands a la the work of Michael Kitces from 10+ years ago. The fact that you are trying to argue against this basic element of portfolio allocation that is accepted by everyone in the financial world is baffling.
The number of false assumptions and straw man arguments in the piece are difficult to count. You say you have to “continuously rebalance” which Frank has never advocated. You also write that fees will be way too high for these “smaller, exotic funds, e.g., one growth and one value fund” which is laughable in 2026. VIOV is 0.1% fee and VUG is 0.03%.
AVUV has substantially outperformed VIOV by over 5% since its inception and is only a 0.25% fee.
This is 2026, there are all kinds of interesting ideas and funds to potentially use. Yes, many of them are “ETF slop” as Ben Felix would say. But no one in the financial world can deny Avantis and Dimensional funds are not at least worth looking at since they are built using Fama and French factors to improve returns.
Your 75/25 portfolio has much more risk with much lower and longer drawdowns than the 40% or 42% of GB or GR portfolios. Frank proved using your spreadsheet tool that over the last 100 years using a similar portfolio you could get a 5% SWR even with Gold essentially acting as cash until the 1970s.
“If they attack one personally, it means they have not a single political argument left.” – Margaret Thatcher
“An ad hominem attack against an intellectual, not against an idea, is highly flattering. It indicates that the person does not have anything intelligent to say about your message.” – Nassim Nicholas Taleb.
I could go on a lot more, but I hope you will consider adjusting your messaging and style. Throughout the piece you cherry picked data to make Risk Parity look worse by pushing the experimental portfolios Frank has on the site instead of a more nuanced investigation of the Golden Butterfly and Golden Ratio. That would have been interesting and much more worthwhile reading.
We need more civil discussions about ideas that people disagree with in this day and age. And those of us that want to do our own investing would benefit from that a lot more than the piece you produced.
Frank started the rudeness, and now he and his friends complain when they face the same treatment. If you can’t take the heat, get out of the oven. You are suffering from “projection,” i.e., accusing others of what you do 10x worse. Besides, he could be the nicest, most polite person; if he intentionally spreads lies about the 5% Rule in his portfolios, then he needs to be rebuffed. He’s performing financial malpractice, and I want to point that out to make sure that my readers don’t fall for this nonsense.
The rest of your comment is exactly Frank’s style: find one or two small details, latch onto them and claim that they completely invalidate my point (they don’t), and then falsely claim that there are so many more to count. There aren’t; you’re just making that up. You can’t argue on any scientific or intellectual level with the avalanche of points and evidence I provided.
Your two ludicrous complaints are:
1: Monthly vs. Annual rebalancing: I showed in SWR part 39 that if I change the rebalance frequency from 1M, to 2,3,4,5,6,9,12,24 months and never, there is relatively little impact on my results. Though I can see how long-term momentum draws you to rebalance less frequently. If you think that’s the case, you can also rebalance the 75/25 less frequently and get the same result, all without Risk Parity. Also, quote ironic: the most recent shtick in Frank’s arsenal is Shannon’s Alpha I mentioned above in the new point #9, which – you can’t make this up – Frank claims raises your expected return from frequent rebalancing. You guys need to keep your lies consistent and agree on one: Either annual rebalancing or Shannon’s Alpha. Which one is it?
2: VIOV vs AVUV is not my idea; it’s Frank’s assumption.
So, you haven’t really shown anything. If these are the two strongest points of yours, then the #3, #4, and #5 “problems” you found in my work are probably that I forgot the Oxford comma in paragraphs 12, 17, and 23.
Besides, you still ignore one of my main points. My post here is not about SCV. If you like SCV, you could do better than Risk Parity with a 75/25 portfolio + SCV bias. My two posts here this week are about how terrible Risk Parity is. Please read my posts again, understand it, and then weigh in.
One more issue I wanted to address:
First, Frank is very much a fan of SCV outperformance. He seemed really pissed when I wrote my relevant SCV posts in 2024 and 2025. Moreover, as I showed with my simulations, RP absolutely needs the SCV premium to produce better results in the historical SWR simulations since 1926. In my simulations, if I set the average SCV premium to zero but keep the business cycle fluctuations intact, i.e., SCV performs “similarly to the overall market but at different times,” then Risk Parity stinks. Even if I give SCV a healthy outperformance of about 1.1% p.a., the Golden Ratio looks lackluster.
I wrote that, you read that, and you still claim the 180-degree opposite. It’s pure gaslighting.
oh wow…did I really read that?
“You’re underselling yourself. If you’re a chimp with a calculator, then Frank would be an ameba with an abacus. I have never seen, read, or heard anything from Frank that makes sense from a quantitative, mathematical, or statistical perspective. – Karsten”
OMG! This is going downhill fast. You guys have to stop this. It’s not contributing to anything but to create animosity and divide in the FIRE community. Shame on you!
I disagree. Frank is doing real harm to the FI community by pushing these portfolios and a 5% withdrawal rate. I’m grateful Big ERN is calling him out.
Agreed. This post was just bizarre. No idea why we all can’t just take a chill pill, enjoy a Windhoek around a braai and take life a little less seriously!
Well, we are talking about our money; we want to be sure the math is correct.
The takeaway for me is that there is no good reason for the Golden Butterfly or Golden Ratio portfolios to outperform the 75/25 portfolio in retirement. To be sure, these “golden portfolios” did outperform for certain starting dates and were not a disaster overall. If you believe in SCV and Gold, the “golden portfolios” is a reasonable option. There is no right answer here. It is an article of faith. Paul Merriman believes. Karsten doesn’t. I think the evidence for 75/25 is stronger. Regardless of the portfolio, 5% withdrawal rate is risky, especially for the 50-year retirement. But then any 50-year plan will have to be modified. Those who starts with 5% must know they may have to cut expenses or go back to work.
I would argue that even if you like SCV and gold, the evidence suggests you would be better off with a higher percentage in equities than the golden portfolios, especially for horizons greater than 30 years.
Exactly! Thanks for weighing in and replying exactly as I would have!
Agree 99%. The only slight adjustment I would add is the comment that “Prof. Vossman” already made, i.e., it’s not really about whether you like gold or SCV. If you like gold and SCV, you’d still implement 75/25 plus some gold and SCV and stay away from Risp Parity.
Thank you, great post! If you ever have a chance, a post on a related (sort of) topic of “Return Stacking” along what you did with this one would be awesome.
Return stacking comes in different flavors. NTSX takes a 60/40 portfolio and scales it up 1.5x to allow for higher expected returns (but also higher risk). I don’t think you use that in retirement to raise your SWR. But it’s not as bad as the Risk Parity idea, which is why I don’t think I need to write a full-feature article about this methodology.
Agree (and I read your leverage article a while back.) I was thinking more along the lines of how is SWR affected if you add the diversifying asset(s)? At the end of the day in terms of SWR, does the cost of leverage outweigh any drawdown mitigation from more diversification? The managed futures space is very interesting in this respect but I also agree with what you’ve written on that point (none of the publicly available ETFs look that compelling). In any event, thank you for responding.
This is a gem of a web resource. I’m very happy I found it (love the options too BTW!) and I hope it’s something you want to do for a long time! x2 on the thanks.
Thanks!
One idea, of course, would be to find a “better” allocation through the tangency portfolio and scale that up. In the standard efficient frontier exercise, that looks super exciting because you can go well past the non-leveraged efficient frontier. I played with that in the SWR sheet, and it improves the results somewhat. I might write another post about those results. Stay tuned!
Good luck with the options trading!
This is not how any of this works. It was NEVER the goal for the Golden Butterfly or Golden Ratio portfolios to outperform the 75/25 portfolio in retirement. Frank always made it clear that the goal of a Risk Parity (which I don’t call risk parity because they aren’t- I call it Bruce Lee Portfolios) is to Maximize SWR and Minimize Drawdowns.
If you’re looking to outperform then 75/25 is probably the way to go, but we aren’t looking to outperform, just maximize SWR.
And those who say they don’t maximize it, just run a backtest on Testfolio or any other backtesting tool and stop the drama
Tyler Kilmer: That was the whole point of the previous Karsten’s post “Can we increase the Safe Withdrawal Rate with Risk Parity? – SWR Series Part 64”.
yeah cherry-picking periods to prove a pre-conceived point! That’s not research.
Igor already replied to this, and I couldn’t have written a better response.
One problem with backtesting static AAs in general is the assumption that an independently thinking person would be willing to buy or hold certain assets at any time throughout history, no matter how bad they looked.
At least some of these so-called RP portfolios were proposed in a time when bonds offered a decent real return. The All Weather Portfolio was first publicized in 1996, when 10-year treasuries were offering about a 3% expected REAL return (vs 2% now), had delivered strong returns for year, and were destined to deliver strong returns for years to come. Maybe it made sense in that context?
In that world, a bond allocation was not so much of a drag as it would be in, say, 2021.
I question who among us managed to be so dogmatic and price insensitive to look at intermediate or long term treasuries at 1.5% nominal yield and with all that convexity and interest rate risk, and with inflation already rising, decided to stay committed to the plan of holding a 55% allocation because it worked in the past. At some point of low yields, the entire premise of diversification breaks down, doesn’t it?
I shorted those bonds and made myself $100k.
Similarly, how many of us would fail to back up the truck if TIPS ever again offered 4% real yields? (Unfortunately I did not have the required “truck” in ‘99)
At least with bonds, you can see what your nominal return is going to be, but just a few years ago people were still buying German Eurobonds at negative yields. Insane. I guess they were following a plan religiously?
I cannot make sense of such behavior. I also cannot imagine selecting a portfolio without knowing something about interest rates, inflation, and stock valuations. Maybe I bargain shop or change plans too often, and maybe US stocks at these valuations are offering a similar deal (I’m hedged), but the whole concept of buying known-bad investments in order to conform to some plan is just odd behavior, IMO.
Good points. I have to shake my head at how anyone could have been so reckless as to implement Risk Parity in 2020 at those prevailing interest rates and then shift massive % into TLT and GOVZ. It’s asking for trouble.
Static asset allocations are benchmarks. As you say, it is hard not to make changes to your portfolio given that the world is constantly changing. However, it is a useful to ask yourself now and then whether all the tinkering has been counterproductive, and you would be better off with a static allocation. Besides, constant tinkering has an opportunity cost: the time can be spent elsewhere.
Constant tinkering would be bad. But I think changing course from something like the bond-heavy All Weather Portfolio during the ZIRP era was appropriate. The alternative would have been to watch one’s portfolio whither, know exactly why it was whithering, and still choose to do nothing.
Similarly it would be appropriate to switch back to a bond-heavy AA in an environment where treasuries were again offering 3-4% expected real returns (or actual real returns, in the case of TIPS). That environment would represent a chance to lock in a near-100% chance of FIRE success, as it probably did in 1999.
We can talk about the performance of various asset mixes since the 1880s, but also acknowledge that these long-term stats do not raise the yield of bonds right now, lower inflation right now, or lower stock valuations right now. Regardless of what happened in the past, the deals offered to you by markets today are the only ones you can take.
Opportunity cost can be a confusing topic because we have to define what opportunity we’re talking about: maximizing returns or minimizing SORR?
If I’ve learned anything from this blog, it’s that retirement success depends upon minimizing SORR rather than swinging for the fences. Also, when the Shiller PE is over 40, and ERN’s own backtesting confirms we get much higher failure rates and much lower forward equity returns at high valuations like these, SORR avoidance is the name of the game. The fat pitches will come for those still able to swing.
Chris B.: Allan Roth in the article “I Love TIPS—Just Don’t Go All-In” discusses what can go wrong when your retirement is based on TIPS. By “opportunity cost” I meant the extra time spent relative to the “set and forget portfolio”. The time can be used elsewhere: the primary job, side hassles, leisure.
Good point: today’s rates are good. But use this sparingly.
Interestingly, today’s environment again supports higher bond weights: high CAPE and relatively high real bond yields. But I wouldn’t go into Risk Parity. Gold has negative momentum now. I’d simply do a 60/40 portfolio with the option to shift into 80/20 later in retirement through a glidepath.
And whether it’s risk parody or 60/40, I’d stay away from a 5% withdrawal rate.
Yes, that’s a similar point to what Joe Saul-Sehy made: more complicated allocations are harder to maintain, especially when markets go crazy. Forget about the time; people may not have the stomach to stick to their complicated allocation.
Excellent as always, Karsten!
– The real tasacworth!
Thanks, Thomas! Greetings back to Georgia!
I listen to Frank’s podcast frequently and I always get the impression he was more about pushing his life philosophy of getting people spend their money. He’s just uses risk parity as a tool to get what he calls “hoarders” to change their habits. The problem is people like him don’t understand that some of us actually get peace out of seeing our nest egg continue to grow and spending money just to spend money won’t make us any happier. To each their own.
I was always suspicious of the 5% withdrawal rate based on the way those portfolios were constructed.
Then he should shift his podcast toward FIRE lifestyle content if he wants to encourage more spending. Talk about how to spend money lavishly: Rolex, Audemars Piguet, ultra-luxury cruises, etc., and how much fun you have with that. Why recommend something stupid and reckless like Risk Parity in combination with a 5% Rule? Why talk about quantitative topics that involve mathematics, statistics, and finance when he has no clue about those? I agree that we should encourage people to spend more, but if we do so in combination with something so dumb, what’s good about it?
I fail to see the supposed benefits of a “lifestyle” that involves spending potentially unsustainable amounts of money on stuff that is supposed to make you happy, when all life experience points to that behavior as a hedonic treadmill to dissatisfaction and eventual ruin.
I know this blog tends to stick to just the numbers, but there’s a philosophy of life at stake too. Consumerists believe that going to the market and buying things to experience is the route to happiness. We’re “taught” this cultural default philosophy our whole lives through the consumption of advertisements.
For most people who subscribe to consumerism, FIRE is impossible. They’d ask why they should give up happiness itself in order to save up any kind of surplus. With this attitude, they’d never build up a surplus to begin with. The mindset inherently leads to a life of work, debt, and financial instability.
Where there’s a market, there’s somebody selling it. So there are influencers who promise you can have it all: Keep your consumerist beliefs and tendencies while not having to work to fund your firehose of wasted consumption and inflation vulnerability. They’re selling books and clicks to people who missed the point of it all: You can’t free yourself from years of wasted work if you don’t free yourself from the attitude that makes such a lifestyle necessary.
Well, if the quant-finance part isn’t his thing, I simply proposed another route. I’m not endorsing consumerism, of course.
Thank you for the detailed breakdown of the flaws of risk parity. It’s amazing how much traction this concept has gained over the last few years, and I suspect part of the appeal is that it appears to promise a much higher (5%) safe withdrawal rate than traditional approaches.
I appreciate the backtesting and analysis, but for me the biggest red flag was seeing models that rely on “magic” ratios and concepts tied to things like Fibonacci sequences. When a portfolio strategy starts sounding like it depends on uncovering ancient mathematical secrets rather than explaining a clear economic rationale, I’m out!
The Fibonacci thing is so silly lol
Name dropping. There is obviously a connection between the Fibonacci series and phi = (1+sqrt(5))/2, but what that has to do with the variance-covariance matrix of asset returns and risk parity remains a mystery.
So, like you, I had to lol. But this nonsense seems to work and strike a chord with a lot of folks. Not with the educated folks who read my blog, but in the other corners of the interwebs.
Thanks! We think very much alike. Like you, I also rolled my eyes at the phi, “Golden Ratio,” and Fibonacci magic that has really no bearing on financial data. 100% charlatanry! Amazingly, he never got any pushback in the various podcast appearances.
Thanks for the post. I do think frank posts some dangerous misinformation. Some just silly. He’ll say “in retirement you want your stocks to be half growth, half value”
That’s just the market! LOL
My thinking exactly: half-half is blended. What’s the point? I would lol, too, if the 5% SWR wasn’t such a scam.
for rebalancing purposes of course!! just wow
I’ve spent almost all of my working life in academia, and find ERN’s tone and content appalling.
[Sorry. I didn’t mean to send only the above.]
I thought I should just compare the Golden Ratio with 75/25 on testfol.io’s Backtester and Monte Carlo simulator. Unfortunately, I don’t know how to attach a screenshot, and one must have a paid subscription for the MC I ran. The backtest shows what while the CAGR of the GR is lower than that of the 75/25 (as expected because of its higher stock allocation), the drawdowns (both in magnitude and length of time) strongly point to the GR as superior. All risk-adjusted measures also point to the GR as superior.
The MC was run with a 5-year block bootstrapped approach and 5000 runs. I like this for computing SWRs. If one focuses on the lowest 10%, he will see that the GR is again far superior: 6.16% vs. 5.21% for the 75/25.
I have focused on relevant measures for decumulation. These do not include maximizing CAGR. ERN could easily have done this himself.
[Backtest] https://testfol.io/?s=7xKj6h4nLya
[MC] https://testfol.io/monte-carlo?s=kSu4GoDby0P
You didn’t even model in withdrawals. We are talking about decumulation portfolios. Try with a $-400 cashflow (4% of initial withdrawal).
Volatility is reduced, so is your return and portfolio balance. Trade offs. Though you could probable just add more bonds for the volatility reduction.
Noted, but the SWR is already what it is; see Part 64, and GR doesn’t help you much or even hurts your results, especially over 50y.
Spencer, as Big ERN has written, Testfol.io’s MC SWR does not depend on withdrawals; it just tells you what they need to be to be “safe.”
Why Karsten continues to claim that the GR is inferior in decumulation is an issue that will never apparently be resolved.
“Why Karsten continues to claim that the GR is inferior in decumulation is an issue that will never apparently be resolved.”
What’s with that comment? I laid this out multiple times, here in this post and in Part 64 of the SWR Series. For your benefit, I’ll do it once again in the comments section:
We need more than this charts, links to Testfolio, Portfolio Visualizer or other independent tools we can play with and run ourselves will be more appreciated.
I used my SWR Series Google kit, not Testfolio.
To my knowledge, my toolkit is the only one that accommodates these asset classes, going back to the 1920s.
There’s no question that Golden Ratio and similar multi-asset portfolios containing trend following perform better for decumulators than 75/25 or other two-asset portfolios, over the time period from 1988 when Testfolio’s KMLMSIM starts. Even without trend following, portfolios containing gold generally also perform better in the only time period that they can properly be measured (1968 onwards).
There are reasonable arguments, though, that both these periods are comparatively short for testing decumulation portfolios; and in relation to trend, that caution should be used with simulated assets.
I proved it with gold and SCV: the observed patterns in SWR analysis don’t support Risk Parity. If you like those styles, you can do better by simply starting with 75/25 and shifting slightly into SCV and/or gold. The same is true with momentum strategies.
And I agree: we should always apply a big grain of salt in projecting those attractive return patterns from gold, SCV, etc. into the future. So, either way you’re better off with 75/25 and worse off with Risk Parity.
That performance looks very different from Frank’s in 2022: In 2022, when looking at drawdowns, Frank’s live returns were worse than the 75/25. Why is that?
1 yr? does it matter?
Yes, drawdowns matter from a sequence-of-returns perspective. See Parts 14+15 of my series. Actually, see all parts 1-64 of my series if you want.
What field? Probably not economics, where the tone is much rougher.
Biostatistics.
The SWR-relevant comparison is:
1966 annual-average ECY: 1.59%
Jan 1966: 1.27%
2025 annual-average: 1.53%
2026 average through June: 1.57%
June 2026: 1.32%
So on ECY, today is essentially in the same valuation regime as 1965–1967, rather than looking radically worse as raw CAPE suggests.
What do you think Karsten (And I want Karsten’s answer, not his dog Spender)
I am obviously not Karsten, but I think it is illuminating to look at the tracking error between the ECY predicted returns and the actual returns. With AI you can easily get these charts and see for yourself that the ECY prediction can be way off. Be sure to choose sufficiently many data points, say, compare the predicted and actual returns annually from 1950 to present. Taking only one data point per decade will give a misleading chart, which may make you believe that ECY predictions work short term.
Exactly: as I stated in my reply, the relative valuation is more relevant for relative performance. For SWR, I am interested in the total return.
The ECY (excess of 1/CAPE minus 10-year real bond yield) is a valuation metric for stock vs. bond returns and could theoretically be useful for stock vs. bond percentage allocations (but with a big caveat: this has frustratingly long lags before any equilibrium is reached!!!). I don’t view this as a useful measure for the real total return, so I wouldn’t use it to educate me about the SWR.
But looking at that ECY, it feels like 60/40 to 75/25 is likely the “correct” asset allocation.
I am italian and I’ve been retired since 2021, living off my financial portfolio. My personal withdrawal rate is low because of my personal frugal lifestyle, and in my case, let’s say any portfolio allocation will be enough “to guarantee” portfolio longevity. Additionally, I’m going to receive more money within 15 years thanks to my pension.
I’m intrigued about Shannon’s Alpha, but unfortunately the friction costs, mainly taxes, reduce the benefit, as you’ve underlined.
My portfolio could be simpler, but I’ve chosen to make it more spicy… and I use exchanged traded products that are UCITS compliant, not US products:
30% euro area inflation-linked bonds average duration 4.5 years (etf ticker INFL1)
60% equity FTSE All-World (etf ticker VWCE)
15% gold (etc ticker SGLD)
5% Bitcoin (etn ticker WBIT)
I’m using a 10% leverage: I have a credit lombard that is similar to SBLOC with LTV 60% and passive interest based on Euribor 3 months + spread 0.65%.
As time goes by, I’m not so passionate about safe withdrawal rate and discussion about decimal point of difference. The main reason is that the worst case is completely different from the median case, so focusing too much attention on this parameter generally induces FORO (fear of running out) and a high risk of underspending/underliving. In case of the worst case due to a bad sequence of return, I’m convinced that flexible spending is a more powerful tool than asset allocation. Do you agree?
At the same time I’m not so confident about backtesting. Learning from financial market history is important (cum grano salis), because the friction costs are not applied and the results are distorted.
You have 60% equities plus 5% Bitcoin, which is essentially a proxy for about 10% Nasdaq (though underperforming dramatically recently). So, you’re well. I would probably get rid of the Bitcoin and just invest in equities directly and replace the gold with a trend-following ETF. But you’re generally on the right path.
With all that, I don’t think you want to apply a 5% WR here.
Thanks for your reply!
It is difficult to talk about safe withdrawal rate because I’m using euro not US dollars for living. In Europe we have many problems finding reliable backtesting and historical analysis given the different currencies before the euro introduction.
I suppose that the EUR safe withdrawal rate is inferior to the USD safe withdrawal rate, with a similar conceptual portfolio.
I want to avoid too much exposure to currency risk. Coming to my asset allocation, in my mind the gold is a useful way for partial but not perfect tail risk hedging to equity during serious drawdown and generally useful for negative correlation to US dollar during “normal” time. With a 60% global equity allocation, I am mainly exposed to the US dollar currency risk (in my case a foreign currency) to more than 1/3 of the total portfolio.
The bitcoin allocation is a speculation just for fun, to calm down my bad idea to play active investing. I wold like to discover if bitcoin’s (past, recently not) high volatility will reward during periodic portfolio rebalancing. I don’t think the past stellar return of bitcoin will be repeatable in the future.
This is a backtesting for EUR investors, with my allocation rebased to 100% without leverage, with the limitation we know (no tax, trading commissions, and market spread). This portfolio is a little bit different because the duration of the chosen inflation-linked etf is longer than mine:
https://curvo.eu/backtest/it/portafoglio/spicy-60-40–NoIgygDglgxgngAgGwAYD0AWFIA0xQCSAoiigEIDSArGQIoAqAHNjigHRVUC6eIxpZFAFkkAJgAaopLnaiA7D0IlyVOQE4AYgGZRVGWwCMGRSDL0AwvpRZStu3a1IuzoA
Andrea, I’m not sure it’s clear that I replied to you.
Andrea, see my comment beginning with “Frank’s motivation.”
Frank’s motivation for his portfolio design work was to spend more money in retirement. You can hold most anything if your strategy is to spend very little. He also wants to give money to family while he is alive and they can benefit most from it.
Just so you know, if you use Monte Carlo methods, you will get a set of SWRs. Do not use the worst, but go to the lowest decile. That is a plausible bad performance.
Thanks for your reply!
My spending is low because I’m “frugal inside” and still in the accumulation mindset 5 years after leaving my job, aided by good market returns. I’m single, have no parents, and I’m childfree. I’m working to improve my approach to discretionary spending without inner regret, i.e., paying someone else for household activities I don’t like. I will probably give most of my money to charity in my will.
Monte Carlo simulations are a useful tool to experiment with. The definition of capital market assumptions, expected inflation, withdrawal rate, and expected longevity, just to name the main parameters, can lead to vastly different results. And we know that both personal portfolio returns and personal longevity are non-ergodic. If you play with Monte Carlo simulations to maximize the probability of success, you also maximize the probability of underspending. We should change the terminology “probability of failure” into “probability of adjustments”.
We often fail to recognize that later in life, the probability of death within a few years is higher than the probability of portfolio depletion. As we age, the ratio between the standard deviation of the expected life and the expected life itself increases. Epigenetics is more important than family genetics, and healthspan is more relevant than lifespan.
When we see numbers, we are rationally inclined to believe numbers are telling the truth. But is it possible to measure the future? When we look to the past, it seems everything is clear and consequential. Yet we forget that, given the same starting point, many different futures could have come to pass. I suppose all of us will benefit from adopting a stoic approach to life, accepting that many things just happen and are out of our control. But we can (or can try to) control how we react.
Sorry for the philosophical rant!
I’d use MC simulation on randomized blocks of years in the historical record. I would not be specifying parameters ahead of time. The biggest reason is that the covariance of parameters is complex and generally uncertain. Blocks of time preserve their historical relationship. For example, it would be HIGHLY unusual to have high and increasing inflation coupled with a boom in long-term treasuries.
RP portfolios allow you to have the reward of donating much while you are alive.
I am dismayed that an Italian is so committed to frugality. 🙂
To wrap up my discussion on Monte Carlo simulations, here are some notes I jotted down in the past.
The success rate of a Monte Carlo simulation does not tell us what our safety margin is – that is, what portion of our essential expenses is covered by “guaranteed” income (Social Security, pension, annuity).
One person with a high degree of flexibility in his spending may be able to accept a much lower success rate than another person who, despite having an identical financial situation and investment profile, cannot accept any reduction in his spending.
A simulation is useful for evaluating the feasibility of a plan through a stress test, but it should not be used as the sole decision-making tool. A plan that is feasible does not necessarily mean it is resilient: knowing how to steer a boat (feasibility) does not mean you are capable of weathering a storm (resilience).
Regarding CMAs, I found this article helpful:
https://www.advisorperspectives.com/articles/2023/12/19/questioning-the-accuracy-of-capital-market-assumptions
When developing a financial plan for decumulation, the sensible approach is not to apply a safety margin to each parameter – to avoid the multiplicative effect of the safety margin – but rather to use a realistic estimate for each parameter without a margin and to apply a 25% safety margin only to the final result. I found this article helpful:
https://abundowealth.com/blog/are-you-making-too-many-conservative-assumptions
In my opinion an intelligent alternative approach is to assess your funded ratio to gauge the sustainability of your withdrawal plan. It’s much simpler than a Monte Carlo simulation, but I don’t think it’s any less effective. I found this article helpful:
https://obliviousinvestor.com/whats-your-funded-ratio/
As you show on your figure the 60/40 and 75/25 are both efficient portfolios with a corresponding reduction in return for a a reduction in risk with more bonds. Personally I’m more comfortable with the volatility characteristic of a 60% equity weight vs 75% which is just my own personal risk tolerance having experienced multiple times went stocks went down – a lot. Instead of “40% bonds” (and consistent with your previous piece on gold) I like 15% gold, 20% bonds and 5% cash for the “lower risk” portion. It back tests OK, Monte Carlo OK, and your research on 15% gold aligns and supports. Thank you for cutting through a lot of the noise on this topic. Your work has (a number of years ago) moved me out of the permanent portfolio type approach – to something that still has those elements but with a more optimized weighting to each – and it’s worked out a lot better for me personally. Thank you again.
I can’t blame folks going to 40% bonds with today’s equity multiples and high bond yields.
With gold, we’re taking a bit of a gamble here, due to the occasional deep and prolonged drawdowns. But gold certainly had an intriguing correlation historically.
Thanks as always Karsten! I always appreciate how you get your hands dirty and do the math I’m not willing to.
I sent an email to Scott at Biggerpockets and asked him to make the interview with Frank available in some form since I think it would be valuable to the community. If anyone else is interested, please consider sending them an email and making a polite request.
I did the same asking them to please reconsider. Release the tapes!
Thanks!
Yeah, it would be nice to get at least a declawed version of this (hopefully with them muting Frank’s mic while screaming).
Thanks Karsten for your thoughtful and data based comprehensive analysis on personal finance issues. Your blog has kept me on a rational path even if the math sometimes eludes me!
Thanks for your kind words! That makes the effort here worth it! 🙂
ERN – this probably doesn’t belong here and only impacts a small percentage of your readers, but I am curious if you have ever thought about doing the analysis on how access to the TSP G Fund might impact SWR? G Fund is fundamentally different than other bond funds as it NEVER goes down in value and historically has beaten inflation and even inflation protected bonds such as TIPS and I Bonds (see https://www.morningstar.com/columns/rekenthaler-report/how-good-is-federal-governments-g-fund). Just curious what your thoughts are on how it could impact SWR and the percentage of bonds that make sense for an early retiree at retirement. My instinct is that G Fund allocation at retirement may have a bigger impact on SWR than gold exposure. While the actual historical returns for G Fund only go back to 1988, I suspect you could probably easily approximate it given historical medium term bond rates and the known information on how G Fund operates.
I bring this up on this article because Frank fundamentally misunderstands G Fund (his podcasts discussing it treat it like it is a short term bond that won’t keep up with inflation, but as that Morningstar article demonstrates, it functions more like a medium term bond but with a guarantee of never losing nominal value and consistently beating inflation in most years). Frank’s misunderstanding of G Fund was what made me first very skeptical of him. Frank discusses G Fund in Episode 412 of Risk Parity Radio and characterizes G Fund as short term government bonds, but that’s not really what the G Fund actually is. Just curious if this is another area where you disagree with him and also if you think it could have a meaningful impact on a retiree’s SWR if they have access to TSP.
I used to have a TSP myself and was always amazed by the G Fund performance. It’s better than a simple money market or T-bill fund. Historically, it has outperformed the T-bill by more than 150 bps. Recently, this outperformance has shrunk (no wonder, given how bonds with even a little duration got hammered), but it’s still intact over 12m and 120m. So, you can probably assume that the G-Fund is a T-Bill allocation plus about a 0.50% annual boost.
Months T-bill G Fund Diff
12 3.83% 4.42% 0.59%
36 4.59% 4.46% -0.13%
60 3.65% 3.85% 0.20%
120 2.37% 2.92% 0.55%
472 3.07% 4.64% 1.57% (since inception)
TLDR: If you have a TSP, you can model the G Fund as a T-bill allocation with about a -0.50% annual expense ratio (i.e., a 50bps annual boost to returns). I still wouldn’t allocate massive amounts to this fund, but if you want a cash bucket, this would be the right candidate to hold it.
And yes, this is another reason to be cautious about Frank’s “advice.”