December 2, 2024 – Many of you have asked me to comment on a recent phenomenon in the FI/FIRE community: Small-Cap Value (SCV) Stocks. Oh, my! Just when I thought we were making progress in the personal finance community, a new fad comes along and threatens to undo all that hard work. When you thought everyone was on board with simple, stress-free, and hands-off index investing, I sense that people feel the itch again to tinker with their portfolios – this time with “Small-Cap Value (SCV) Stocks.” To be precise, small-cap value is nothing new. I’ve written about my SCV skepticism in a post over five years ago. But the way it’s sold now is from a new angle, and it’s getting traction in the FI/FIRE community. The SCV media blitz relies on the latest narrative that, sure, broad index funds (VTI, ITOT, etc.) are a great and simple way to reach your financial goals. But SCV is an even better way—a “more optimal” way to reach your goals. Allegedly, SCV is the secret sauce for accelerated financial success for astute and enlightened investors.
But alas, much of that narrative is hype and false advertising. In today’s post, I want to reiterate the case for simple broad index investing. Let’s take a look…
Before we get started, though, I wanted to link a few recent podcast appearances. Please check out the following awesome podcasts and show them some SEO love:
- I chatted with Paula Pant on her Afford Anything podcast about the economic outlook, especially monetary policy. This episode is also on YouTube.
- My second appearance was on the Forget About Money Podcast, where I chatted about options trading.
- Finally, I appeared on the Catching Up to FI podcast to chat about my FIRE journey.
Now, back to today’s scheduled content…
Small-cap value stocks had an extraordinary performance until the early 2000s!
To give credit where credit is due, small-cap value stocks have had a fantastic run since 1926. A small-cap value fund would have significantly outperformed my go-to index, the S&P 500 (and its precursors), which I use for my Safe Withdrawal Rate research. Of course, to my knowledge, there were no low-cost SCV index funds before July 2000 when iShares launched its iShares Russell 2000 Value ETF (IWN). So, I need to rely on estimates for monthly returns for my backtest. To that end, I run a “factor model” to estimate IWM’s exposures to the S&P500 and the Fama-French SML (small minus large-cap) and HML (high minus low book/value ratio, i.e., value over growth) factors and then construct the estimated IWM monthly returns between 1926 and 2000. In my regressions, I found that the IWN ETF behaved like a portfolio with 99% S&P 500, 1% cash (i.e., 3M Treasury Bills), 95% Fama-French SMB factor, and 52% Fama-French HML factor. I also subtract a 0.24% per annum expense ratio from the estimates. To replicate the S&P 500 ETF, I use a 100% S&P index portfolio minus 0.03% annually to account for the IVV expense ratio.
In the chart below, I plot the cumulative real (CPI-adjusted) returns since June 1926. I also include the SCV outperformance vs. the S&P 500. I normalized all figures to 1.0 on June 30, 1926, and plotted this with a log scale on the y-axis to better display the exponential growth over time. No doubt, SCV had a great run over the last 98 years. It outperformed the S&P 500 index by about 2.6 percentage points annually. The final value of the SCV portfolio is about 10x the S&P 500’s.
So, it’s no surprise that SCV has won many fans in the personal finance community. Some of the proponents are Paul Merriman and several other SCV fanboys (and girls) on the interwebs, including the Risk Parity crowd; Frank Vasquez uses SCV in some of the sample portfolios at Risk Parity Radio.
A previous version of the post also attributed the recent SCV-mania to Joe Saul-Sehy (Stacking Benjamins) on a recent Afford Anything podcast (at about the 9:30 mark). However, in a comment below, Joe made clear that he’s not an SCV fanboy. Rather, he prefers a more nuanced approach to portfolio construction through the Efficient Frontier; see a recent Afford Anything episode for more info. With this approach, you may find portfolios not as simple as they are often recommended in the FIRE community.
The 4-Fund Portfolio
But let me focus on the SCV main character, Paul Merriman. His new en vogue portfolio is the 4-fund portfolio (4FP), comprised of equal 25% shares of the S&P 500, large-cap value, small-cap blend, and small-cap value. A few years ago, it was a 10-fund portfolio. I couldn’t ascertain why the preferred SCV portfolio changes over time. I’ve been around in the finance world long enough to suspect that when someone’s “Forever Portfolio” changes every five years, it’s likely due to overfitting and backward-looking bias.
In any case, I simulated the returns of a portfolio comprised of the corresponding iShares ETFs: IVV for the large-cap blend, IWD for large-cap value, IWM for the Small-Cap Blend, and, as before, IWN for the Small-Cap Value. ETF returns, the simulated returns, and the factor exposures I estimated for the funds are in the table below. I also subtract the respective expense ratios to arrive at my estimated ETF returns pre-2000: 0.03% p.a. for IVV, 0.19% p.a. for IWD and IWM, and 0.24% p.a. for IWM.
We first notice that the most recent ETF performance of all the non-IVV funds stinks. Very consistently, over all horizons from one year to twenty years, the exotic ETFs underperform the plain old IVV ETF, and often significantly. The same is true for the four funds mixed with equal weights. My simulated returns align very well with the ones Paul Merriman found. Specifically, I confirmed that during the 1928-2021 period, the 4F portfolio outperformed the IVV by 1.7 percentage points (11.7% vs. 10.0%) in my simulations, the same outperformance as Merriman found over the same horizon.
But let me first plot the same chart for the 4-Fund portfolio; please see below. If you’ve seen Paul Merriman’s presentations, say, in Paula’s podcast at around the 4:12 mark, they resemble each other remarkably well. However, note the slightly different start and end points, especially that Paul Merriman “conveniently” ignores the post-2020 underperformance.
So, what’s not to like about SCV and the 4-Fund portfolio, then? The pretty chart above insinuates that the 4F portfolio consistently outperformed the S&P, right? The orange line has been above the green one since at least the 1930s, so you would have always done better during that period, right? All the cohorts beat the S&P 500 by 4x, right? Wrong! What the chart disguises is the recent underperformance of the 4FP. The orange line may have been above the green, but the distance has narrowed noticeably. That’s also reflected in the black line, i.e., the 4FP divided by the S&P 500 falling. Of course, we already knew that by looking at the recent ETF performance data in the table above. This brings me to the next point…
Small-cap value stocks have underperformed recently!
What I found very sneaky about long-term cumulative return charts like the ones Paul Merrmiman tours around with is that they have very little relevance for investors alive today. Notably, I know nobody who has been investing in the 4FP since 1926. Every point along the chart above corresponds to a cohort that initiated a buy-and-hold investment in 1926. For example, if I go along the horizontal axis and look up the values for 1980, the lines indicate the performance from 1926 to 1980. How many investors like that do you know?
A vastly more informative chart would be to ask, among all the different investors alive today, how well would they have done with the 4FP relative to the S&P 500, depending on when they started investing? So, let me do that in the chart below. On the horizontal axis, I plot the buy-and-hold investing period’s starting month rather than the ending date. Of course, by definition, the function values at 6/1926 in this new chart coincide with those at 10/2024 in the Merriman-style chart above.
The central insight I gather from this chart is that since the late 1990s, no buy-and-hold investor would have beaten the S&P 500 with a 4-Fund Portfolio. Even for investors between 1982 and 1999, the two portfolios were about even, visible by the black line closely hugging the 1.00 value. Finally, only pre-1982 4FP investors have consistently outperformed the large-cap index. So, absolutely nobody alive today would have done four times better than the S&P 500 index with this 4-fund portfolio, in stark contrast to the false advertising you heard from SCV fans in their recent podcast media blitz.
So, is it only me who’s defending the honor of simple index investing? Not entirely; I should give kudos to Rick Ferri for a good presentation, “The Case Against Factor Investing,” at the 2023 Bogleheads conference (his excellent slides are on pages 141-151 in the pdf). At the 2024 Bogleheads conference, one segment featured Paul Merriman with Dr. James Dahle (White Coat Investor) moderating and providing some skepticism and pushback. But I’m still shocked that so many other usually enlightened FI/FIRE influencers cast aside their passive investing principles and were intrigued by this SCV/4FP mumbo-jumbo. Most disappointingly, in a recent Forget About Money podcast, JL Collins seems to endorse SCV.
Is the Small-Cap Value Stock out-/underperformance statistically significant?
Another fly in the ointment for the SCV crowd is that despite all the pretty charts and tables that Paul Merriman posts on his site, he fails to answer one central question: is any of the 4FP outperformance even statically significant? If so, at what level? So, let me do that for you. Of course, there isn’t one single way to determine that, so I will present three different ways:
1: Statistical Significance over the entire 1926-2024 time period
First, I calculate the monthly outperformance of the 4-Fund portfolio over the S&P 500. The average monthly outperformance of the 4-Fund portfolio was just about 0.10%. The monthly standard deviation was 1.85%, and the sample size was 1,180.
Technical note for the Math Geeks: We cannot use r_4FP-r_SPX or even (1+r_4FP)/(1+r_SPX)-1 for this exercise because averages of those times series will usually not coincide with the average outperformance due to compounding effects. To account for compounding, we must use (natural) logarithmic returns, i.e., ln[(1+r_4FP)/(1+r_SPX)] as our time series.
To test whether the mean outperformance significantly differs from zero, I compute a t-statistic of 1.90. That’s quite impressive, but you still can’t reject the Null Hypotheses of a zero mean at alpha=5% significance. That critical value would have been 1.96. So, the SCV crowd sells me a story that’s only slightly less than statistically significant, even under the best circumstances, i.e., including all the outperformance from “way back when” and the smell of “hindsight bias” of regularly changing the portfolio weights of the “forever” portfolio.
And it only gets worse from here…
2: Rolling 10-year windows
I also like to see how consistent (or not!) this 4-Fund portfolio outperformance vis-a-vis the S&P 500 has been over shorter horizons, i.e., over the average complete economic and stock market cycle. Ten years would be a long enough estimate for capturing at least one recession and expansion. And one bull market and one bear market. So, let’s plot 120-month rolling t-stats for the Null Hypothesis that the mean outperformance is zero. I also include the +/-1.96 critical value bands to indicate where we can find significant outperformance (or underperformance). Sadly, there are only a few 10-year windows when this outperformance is actually significant.
So, can you increase your safe withdrawal with SCV/4FP? Due to Sequence Risk, the first market cycle determines much of your retirement success. But if the outperformance isn’t consistent enough to prevail over every single market cycle, I would not pin my hopes on SCV/4FP increasing my safe spending rate if the outperformance is so elusive and unreliable and may only benefit my portfolio too late into my retirement.
Of course, the good news for the SCV/4FP fans is that the recent underperformance isn’t statistically significant either. So, for the SCV skeptics, don’t feel overly confident either! Specifically, don’t replace your broad market index with just large-cap growth.
3: (Backward-)Expanding time windows, ending in 2024
Finally, I do the same exercise as before, i.e., expand the time intervals backward with the same endpoint in 2024. In other words, I want to determine which cohorts with a starting point between 6/1926 and 10/2014 and a common endpoint in 10/2024 would have experienced statistically significant outperformance when investing using the 4-Fund method.
I plot the t-stats in the chart below. The good news for SCV is that, as bad as its recent returns have been, they have never been significantly different from the S&P 500, at least not at a 5% significance level. But that’s where good news ends. Even for cohorts in the 1990s, the occasional slight outperformance of 4FP over the S&P 500 is essentially statistical noise. The t-stat never even reaches +0.5. Even well before 1982, the 4FP outperformance is far from statistically significant. To reach statistical significance, we’d have to find a buy-and-hold investor who’s been in the market since 1943, i.e., for 80+ years! As pointed out before, if we move back to 6/1926, we drop just slightly below the significance threshold again to 1.90.
Side note: Of course, we could also run one-sided tests. In that case, the 5% critical values would have been about -1.645 and +1.645, respectively. The most recent data would have supported rejecting the Null Hypothesis of SCV weakly outperforming the S&P 500.
Small-cap value stocks look even worse with regular investments!
In the calculations above, I studied buy-and-hold investors. We can also ask ourselves whether any investors alive today who made regular contributions along the way would have benefited from this 4F portfolio. That is even more relevant for those investors who built their nest egg over time, like most of us in the FI/FIRE community.
So, how early would a 4FP investor with regular contributions have to start investing to outperform a simple S&P 500 portfolio? I plot that information in the chart below. Specifically, for every cohort on the horizontal axis, I plot the portfolio value this cohort would have accumulated up to 10/2024 when saving $1 monthly. As always, all returns and contributions are CPI-adjusted.
The crossover point where an investor with regular monthly contributions would have done better with the 4F portfolio was in March 1977. To get a sizable boost, say 2x vis-a-vis a simple S&P500 portfolio, you’d have to move back to regular investing since 1943. Even with regular contributions since 1926, almost 100 years long, you’d stay slightly below 3x the S&P 500 portfolio. Thus, very few SCV fans alive today are richer than their S&P 500 counterparts. All the claims that you can miraculously 4x your portfolio with this simple SCV trick are false advertising.
Why is the crossover point even worse than for the buy-and-hold investors? It’s due to one of my favorite finance subjects, i.e., Sequence Risk; you benefited relatively little from the SCV outperformance before its peak when your portfolio was (relatively) small during the first few decades of your accumulation period. And then, when you had a much larger portfolio later in the accumulation phase, you got slammed with the terrible underperformance of SCV vs. the S&P 500.
Diversification or Di-WORSE-fication?
If the average returns of SCV don’t look too appetizing, especially in the last 40 years, another angle folks try to sell SCV and 4FP is that they offer better diversification. Specifically, most SCV fans claim that your average total market index funds (e.g., Vanguard’s VTI, iShares ITOT, etc.) and certainly your large-cap index funds (Vanguard’s VOO, State Street’s SPY, iShares IVV, etc.) are “not diversified enough,” and adding SCV lowers your risk (their words, not mine). That would be a desirable feature, especially in retirement when you like to avoid steep drawdowns and Sequence Risk.
But alas, the volatility reduction claim is demonstrably false on at least two levels. First, a total market fund is diversified by definition because it includes all the large-cap and small-cap stocks plus value and growth stocks. All constituents are weighted according to their market cap (often adjusted for “free float” only, i.e., excluding restricted/closely held shares). So, adding more small-cap value will cause less diversification and more concentration.
Second, in most cases, the diversification claim is also purely statistically false. Adding new assets to a portfolio can indeed increase your portfolio’s standard deviation, even if those new assets have correlations substantially below 1.00. A necessary condition for a new asset to reduce the overall portfolio volatility is that its standard deviation times the correlation between the asset and the current portfolio is less than the current portfolio’s standard deviation. For the Math wonks, here is my short derivation of this basic finance/statistics fact:
So, let’s put that formula to work. Let’s calculate the volatilities and correlations of the alternative ETFs and the Merriman 4-Fund portfolio to see which would help diversify equity risk. Notice that all exotic ETFs had volatilities greater than the IVV S&P 500 ETF. But IWD has indeed a correlation small enough to bring the sigma times rho below that of the IVV. So, a large-cap value ETF like IWD would have, at the margin, slightly reduced your volatility. However, the other ETFs and the Merriman-Style 4-Fund portfolio would not diversify your portfolio. Instead, they would di-WORSE-fy your portfolio by increasing your volatility. It’s a mathematical certainty. I also confirmed that the same is true over different horizons (i.e., over 20 years, since the ETF inceptions in 2000, and even when extending the series back to the simulated period starting in 1926).
Side note 1: By the way, the same result also holds if adding only the three funds IWD+IWM+IWN, instead of the four funds together.
Side note 2: Even a total market index, i.e., the iShares ITOT or the Vanguard VTI/VTSAX, would routinely fail this criterion. Example 10/2014-10/2024: VTI risk = 15.63% p.a., correlation w/ SPX=0.9953. 15.63%*0.9953=15.55%, which exceeds the S&P 500 risk of 15.2%. So, intriguingly, going from 500 stocks to 3,600+ tickers in the VTI does not lower your risk!
Side note 3: This is solely about volatility. Despite the slight vol reduction after including in US Large-Cap Value stocks, you still would have reduced the Sharpe Ratio!
To be clear, Paul Merriman has gone back and forth on this volatility issue. Sometimes, he concedes that the 4-Fund portfolio has higher risk; please see the return stats table screenshot from his site:
However, in a recent Afford Anything podcast (around the 4:15 mark), he contradicted himself and claimed that the 4-Fund portfolio had lower volatility.
So, I recommend that my readers avoid the 4-Factor portfolio and other variations of SCV. It would have made people’s portfolios WORSE over the last few decades. Thus, it’s been true Di-WORSE-fication, not just from an average return perspective but also from a diversification and risk minimization point of view.
Can this poor performance turn around?
I don’t want to sound too pessimistic about Small-Cap Value. As I indicated above, the statistical evidence for the S&P 500 outperforming SCV/4FP is also weak. There is also the risk that all the artificial intelligence (AI) optimism was for naught and all that large-cap growth outperformance comes to a crushing end. It’s certainly possible, but it’s hard to know if this reversal ever happens. And it’s even more difficult to time the exact turning point. I don’t want to take a side in the pro vs. cons AI fight, so with a broad index fund, I cover both scenarios: growth and value.
Conclusion
My regular readers know that I am in no way beholden to the passive indexing & simple investing crowd. Quite the opposite, from 2008 until 2018, my paycheck came from working in (gasp!) active, global tactical asset allocation (GTAA) at Bank of New York Mellon Asset Management. In retirement, I generate extra income trading options. My Safe Withdrawal Rate Series is about how personal finance isn’t as simple as often proclaimed in other corners of the FIRE movement. If there are ways to improve your financial success and optimize your portfolio, I’ll be eager to check them out with an open mind. I’d want them to work and will proudly announce the news here on my blog. I am currently doing so for my options strategy. But SCV and its derivatives like the 4FP aren’t that new secret sauce. The incremental performance of SCV and the 4-Fund portfolio above the S&P 500 is too volatile to pin your hopes on this “flavor of the month.” But for the record, I’m not forecasting that the SCV/4FP’s recent underperformance will continue. So, don’t go 100% large-cap growth, either. I recommend keeping it simple and using a broadly diversified equity index fund. A total market fund (FZROX, ITOT, VTI, VTSAX) works, though I slightly prefer large-cap funds like FNILX, IVV, VOO, etc. Happy Investing!
Thanks for stopping by today! I look forward to your comments and suggestions!
Title Picture: AI-generated with WordPress
I get really happy whenever a new ERN post drops! Thanks Karsten!
Thanks for the words of confidence! 🙂
I drop what I am doing and I read the post and then reread anything I don’t understand.
That’s a great way to extend the fun! Thanks! 🙂
Broad question. What if the Limits To Growth study is correct and around 2050 the global economy goes into permanent decline (because of overshoot, dwindling resources, food scarcity and pollution)? Do you think in that case (which I find more than plausible) the 4% or 3% rules will still hold? Or will we have other things to worry about (like holding to dear life)? Is this something that you have at the back of your mind?
The Club or Rome and other fringe groups have been preaching doom and gloom since forever. I don’t take that very seriously.
In fact, some of the “solutions” preached by these fanatics could potentially cause the doom and gloom about which they try to warn us.
Thank you BigErn. I deeply appreciate your insights and all of your blog posts on FIRE. However, I can’t help but grapple with the reality that infinite growth on a finite planet is a biophysical impossibility. Ecological challenges are becoming increasingly tangible, and I wonder how long mainstream economics can continue to sidestep these issues. Take climate change, for instance—we’re on track for 3°C of warming by the end of the century, a level that may be incompatible with advanced civilization.
I’m not looking to debate, but rather to understand the apparent disconnect between biophysical science and economics. No need to respond here—just offering this as potential food for thought for a future post. I’d be fascinated to read your in-depth analysis on this topic, especially since the FIRE community is planning for multi-decade horizons based on their investment returns.
Thank you again for all the great content. You have helped me a lot to achieve my financial goals.
Infinite growth may be impossible. But who decides when we hit that wall? 100 years ago people thought that there can be no more growth because how would we deal with all that horse manure. Technical progress took care of problems in ways not thought possible previously.
And this growth and global warming discussion distracts from the actual topic here. If there is no more growth, that would be the kiss of death to SCV stocks, so are making my point here.
I forgot to say, that I got here by pressing a link in one of Paul Merriman’s emails. So for sure he is fair and wants us to hear also other opinions! But it seems that so are you. Thanks
Yeah, he’s a good guy. Very gracious with his time. It seems that the two opposing sides on this topic actually get along well and respect their differing opinions.
Even though you do not write that often any more, every post is still gold, Karsten!
Quick question/possible topic: does using AAA/AA corporate bonds instead of Treasuries in the retirement portfolio improve SWR? My hunch is probably they do not because of (a) no inflation protection like TIPS, (b) they will probably be much more correlated with equities than Treasuries, so additional credit risk premium does not justify smaller downside protection. But would love to hear you opine on this.
Thanks!
Corporate bonds will have a slightly higher yield than Treasury bonds. But even the highest rated AAA/AA bonds will have slightly less diversification when a recession hits. So, there’s no free lunch.
AAA/AA corporate bonds have no inflation protection, true. But neither will your Treasury bond fund. TIPS are a separate and a completely different animal from nominal Treasury bonds I use in the SWR toolbox.
Nominal TIPS or funds? Does it matter?
There are no “nominal TIPS,” so you likely meant to write “Individual TIPS or funds?”
I have never traded individual TIPS bonds. For TIPS, you may be better off trading the ETF or mutual fund. The tax treatment should carry over through the issuer to you the buyer of the funds.
Thanks a ton Karsten for your super honest contributions to this space. We are all DIYers trying to do what’s best. Quick question, does regular rebalancing between SCV and S&P500 make a difference in the returns or volatility?
I haven’t run any detailed simulations. Because there seems to be momentum of SCV out/under-performance, so it’s likely better to not rebalance that often.
Big Ern, thank you for your insight (as always). I was wondering if you had opinions on the Merriman 2-fund (70% S&P, 30% SCV)? A quick back test on portfolio visualizer shows a modest 0.32% increase in CAGR (and corresponding increase in volatility) since 1987. All of the podcast recommendations aside, I can’t help but feel like the top heavy S&P 500 is extraordinarily overvalued with PE of 35. LCV PE20 and SCV PE 15 look like “bargains” right now (relatively speaking). I hesitate to use the word “hedge” but, I’ve made impressive gains over the past ~10 years, and I want to “lock” them in. Now that I write this, I’m thinking maybe I should just surrender to the 60/40.
https://www.portfoliovisualizer.com/backtest-asset-class-allocation?s=y&sl=3Jy1WRwRLn82yCh2UY0xfX
The 2-fund portfolio wouldn’t have done very well during the Great Depression, so, I suspect that’s why Merriman declawed the SCV a bit and replaced it with the 4FP. But you’re right, the 2FP looks better 1987-now.
I’m a bit worried about the nose-bleed-high valuations, too. But it’s impossible to time this, just like it was in the dot-com bubble.
That said, I don’t think today’s “bubble” is as bad as the tech bubble from 25 years ago. NVDA is probably more solid than pets.com.
Also note that a lot of old-style industries are stagnant and cheap for a reason. So, yes, if someone could pick out an NVDA-equivalent in SCV with a lower PE, I’d be all over it. But there are none.
I’m about to surrender too, but I think Karsten has me convinced that 70/30 will likely suffice for our purposes.
Anything between 60/40 and 80/20 seems great in retirement. While accumulating, it’s OK to go 100% equities, though.
Right, and based on your articles, I’m going to sort of split the difference! We were nearly 100% equities until recently, but given that I might retire from academia in the next 2.5 years I’ve started shifting some money into treasuries. I very much appreciate your work Karsten. It has been priceless! Thank you.
Thanks! Good luck!
I think the key thing is finding an asset allocation that one can hold to. Especially if coming from 100% equities in a rising market. Personally I went to 75/25 in retirement and sometimes feel the allure of 80/20 but then decide 75/25 is right for me. The trick is having an allocation that can be stuck with when “missing out on growth” and when “the sky is falling in”.
Yes, good point! Some retirees feel resistance to 20-40% bonds when they see the continued rally in equities. I always tell myself: why keep gambling if you won the game already?
Thank you so much for writing this. I have been a committed Simple Path investor for almost 10 years. This 4 Fund Strategy campaign almost threw me off and I started to believe that maybe I was missing something about accumulation. I was suspicious when Mr Merriman said that there isn’t a specific index(es) that he could recommend to achieve the strategy. This post settled it for me. I’m going to stick with the Simple Path until decumulation forces me to get more sophisticated.
Thanks! Kudos to you! I can see that at first glance this SCV seems attractive, but very quickly people will notice inconsistencies, as you did. I never knew Merriman was so vague on the index. At least recently, there are very well-regarded broad indexes like Russell 2000 (small), R2K-Value, etc.
michael green [who is very skeptical of indexation and its effects on the market, btw] has written that value indexes are equivalent to selling both a put and a call. i.e. they sell when a stock goes up and is no longer “value,” and they buy when a stock goes down and becomes “value.” i.e. you’re selling vol. given that there is a lot of vol selling going on, [e.g. including you and many others selling options for income] in general vol is low, and a strategy of automatically selling vol is not likely to produce great returns.
Thanks for sharing. That’s an interesting concept. While I’m selling options every day, this similarity has not enticed me to find SCV attractive. That’s because poor returns and SCV becoming even cheaper can last for years, even decades, while my options trading resets every day.
I am curious to see how 50/50 LCB/SCV would have done compared to 4FF and if the returns would have been statistically significant. At least on backtesting, 50/50 looks better.
https://www.portfoliovisualizer.com/backtest-asset-class-allocation?s=y&sl=MIJpaJjA93Dxr0ERxu8NF
You can look at my chart with 100% SCV and 100% LCB and take the midpoint. It would be better on average than the 4FP, but it had some nasty drawdowns. The 4FP clamped down on the volatility much better. Hence, the 4FP, not the 2FP.
Thanks for writing this. I’m invested in a good amount of SCV now, but only because I think it’s a 3-5 year allocation, not a forever trade.
The backstudies in SCV omit the impact of Sarbanes Oxley, which causes companies that would historically have been small caps to stay private longer. Then the index gets overweighted with smaller financial and insurance stocks, which can afford to go public because they already deal with reporting / governmental compliance in their day to day business.
I’m bullish because of valuations on small financials, but also agree long term it would not be logical to believe in SCV’s outperformance unless SarBox is rolled back.
Thanks for sharing. I never made the Sarbanes Oxley connection. Very intriguing.
Agree, SCV could be a market-timing play: bet on the next short spurt of SCV outperformance and then get out. But that event may come in 20 years or never.
thanks! I also follow Bill Bernstein and like his practical approaches. IIRC he said one could squeeze out a little more return by adding tilt to small and value based bunds, if one finds it worth the (to me relatively minor) hassle. So I have.
As I interpret your posts in the past, this comes down to is a SWR. At least it does to me, so that is what I looked at. And of course it raised some questions.
I use your SWR spreadsheet and I am always evolving on learning on this. So I admit I may not be using it correctly. But I went back and substituted my fama-french factors of 18%/9% with 0%/0%. At the original values my calculated results >0% failure rate was at 4.41% withdraw level. When substituting the 0%/0% factors, the failure rates at 4.5% rate stayed about the same, very slightly favoring my original fame-french factors.
I had guesstimated my blended expense ratio as 0.07%. I went back and re-evaluated my expense ratio, and perhaps underestimated the higher expense of SCV, I re guesstimated the ratio to 0.10, and that changes it to being slightly in favor of of the 0% fama-french factors. But essentially even
I have a few questions:
1. is this an appropriate use of the SWR sheet?
2. does changing these factors pull through to any other data points on the spreadsheet that I should look at?
3. Does the expense ratio get multiplied against all parameters of the portfolio ? I ask because I may go back and figure out my amount-weighted true expense ratio of equity funds, but my treasuries/cash have an essentially zero expense ratio, so I would correct based on my asset class, but if you have already done this…. I admit this would be easier if I had one broad spectrum fund I used, but still way easier than writing a blog 🙂
I realize I am exposing my ignorance, but if I dont ask I wont learn!
Thanks much
In the SWR sheet, the FF factors will help with the SWR. That’s because SCV did well during some of the past worst-case scenarios. I would not want to rely on that to be true going forward, though.
So, you applied the FF factors in the SWR correctly.
The expense ratio is simply subtracted from the monthly returns (1/12 of the annual number). It’s not “multiplied”
Karsten, Thanks again for a thorough analysis of a topic of concern to many in the FIRE community (I’ve been wondering whether to add an SCV tilt for my Roth, for instance, due to Merriman’s 2 funds for life and now Bengen’s new research).
Bill Bengen has recently been touting 5% as the new SWR instead of 4%, and I’m not convinced. I’m wondering to what degree your above SCV analysis debunks his 5% SWR conclusion. Part of my skepticism is regarding reliance on SCV premium, but also because Bengen’s original 4% SWR research was replicated by the Trinity study (for ~30 year time horizons only!), but I am not aware of anyone replicating this new 5% claim.
He claimed (on Afford Anything, episode #560, among other places, I believe from his recent book), that with a few tweaks, he was able to juice the SWR up to 5%. To be fair, he does say that planning horizon matters, and that for the FIRE crowd, it pushes that 5% down to 4.2%, so he continues to acknowledge a significant reduction needed for long time horizons (which most FIRE adherents forget).
But, he is claiming he bumped it up to 5% with 4 boosts (I think he called them free lunches):
1) Rising equity glide path post retirement (reduced before retirement – ala bond tent). I know you’re a fan of this one too – as am I.
2) Diversification of asset classes beyond stocks and bonds.
3) Regular portfolio rebalancing – this one seems bogus, as it is already an assumption of all SWR models, not some extra return source.
4) Overweighting SCV – this is why your above analysis may partially debunk his new thesis.
What are your thoughts on Bengen’s 5% SWR claim and especially as it relates to SCV?
Personally, I trust your SWR toolbox a lot more than a fixed SWR rule, whether 4% or 5%. Does your toolbox have a way of modeling a rising equity glidepath?
Thanks again!
Yeah, Bill Bengen has jumped on the SCV bandwagon. He’s correct in that historically, the SCV flavor would have improved the SWR because SCV outperformance was early enough. But since that “alpha” source has run its course, I believe we should no longer apply that generous SCV dividend. Certainly not 1.7% p.a.
Thanks for confirming what I suspected.
To be fair, he isn’t attributing the full SWR increase to SCV tilt. He claimed the 4 “free lunches” above all contributed (though he appears to double count rebalancing, since it was always assumed in all SWR models).
Also, is there any way to model rising equity glidepath in your SWR toolkit spreadsheet? If so, I missed it.
Thanks again.
Currently, you can only run a case study, one cohort at a time. But I’m working on updating the toolkit to also allow glidepaths.
Don’t get your hopes too high. The GP will maybe raise the rate by 0.25 %-points. Bengen certainly claims that most of the rise to 5+% comes from SCV. Which is bogus, because that alpha train has left the station.
But if there is (at least ) a bit of “alpha” in SCV, why don’t you like total market over the SP500? Personally, I tend to want to add 10% of mid and 5-10% of SCV just to spice things up without pinning my hopes on any one part.
That was in addition to the total market.
Noted.
(that was in addition to a total market fund)
I can’t speak for Karsten, but I would guess because the difference would be negligible, as SCV likely represents only 1%-3% of the total market index, since both types of index funds are market cap weighted.
Correct!
That said, I believe Small+Medium is about 15% of the Total index, so half of that is value. so we’re looking at about 7.5%. But your logic still applies.
Thanks. Just as a point of clarification, I believe your above estimate of 7.5% is small+medium cap value, but I was attempting to estimate only small cap value, so it should be at least somewhat lower (I’m not sure by how much).
There are splits that take into account medium as well, sometimes only large vs. small.
I simply remember the rough estimate that S&P 500 is 85% of the total. So, I estimated that 15% is below and half of that is value. But depending on what SCV index you use it could be off by a few points.
Thanks, that was interesting and informative. Gave me a different perspective on the chart.
Glad you enjoyed this. Thanks for your feedback.
Have you seen the ITOT vs. IVV cumulative return over the last 20 years? It’s essentially the same. But there are very tiny differences: IVV outperformed the ITOT slightly. And IVV had slightly lower volatility. Between the two, I’d pich the S&P 500.
Thank you for this. I was very skeptical of all the hype around this that has been coming out of Afford Anything and Joe in particular.
I honestly don’t care about getting the absolute best returns because at the end of the day, I’m not leaving my money invested forever. I’m trying to retire and live off of it. What I am interested in is having an allocation that allows for the highest possible SWR.
For me the biggest question is whether a SCV tilt can increase your SWR even if it doesn’t increase the overall performance (does it reduce the amount or length of the downturns). It seems like Joe Saul-Sehy has been arguing on Afford Anything that it does, but from what I’m reading in your last section “Diversification or Di-WORSE-fication,” your answer is no, it doesn’t, despite what we may be hearing on these podcasts. Am I reading that correctly?
What about diversification into other asset classes like REITs or Gold? I’ve heard from some podcasts that diversification into these other asset classes can increase the SWR by reducing the size and length of the downturn based on historical data. Do you think that the data supports this argument?
Thanks!
As written in earlier comments, SCV would indeed increase your SWR historically. But we should not extrapolate that 1.7% outperformance since that opportunity has been arbitraged away. So, the SCV proponents (Bengen, Saul-Sehy, etc.) are technically correct, but extrapolate those results in a faulty way into the future.
REITs: I don’t have enough return data.
Gold: See part 34 of the series.
In 1970s, stocks performed badly (in real terms), partially due to significant PE compression (PEs has to drop following bond yields going up). Having SCV tilt was helpful with achieving higher SWR because SCV had substantially lower PE at the beginning (I think, need to double check), so there was less room for their PE compression.
Similar to the logic used to substantiate calculating SWR based on current CAPE ratio, I think it’s useful to look how mixing SCV helps with managing sequence of return risk when there is large disparity in PE values between large caps and SCV. I.e. take “Rolling 10-year windows” section and overlay PE ratio of large vs. small stocks to be able to see correlation / impact of PE differences (vs. looking at all the periods).
There could be not enough historical data to make any claims, but I wonder – if we were to enter similar phase to what market experienced in 1970s, then how likely are we to see similar SWR impact from tilting SCV as we saw back then?
To some extent, SWR is a function of portfolio performance through the worst known (at the moment) historic timeframe, and currently 1965+ is such timeframe, and having SCV tilt helped there. And some of today’s conditions somewhat resemble that timeframe.
I see your point. There may be a reversal again, like in the early 2000s. But betting on it and timing that reversal can prove very frustrating. Rick Ferri’s presentation is along those lines.
Also, this current post is not about SWRs. I did study the effect of SCV in part 34. SCV improves the SWR. But I don’t believe that we can extrapolate that SCV outperformance so easily into the future.
A bit late to the reply party but…
Of course I agree with your biggest picture take, and the idea of getting 1.7 points p.a. From SCV is ridiculous.
But the flip side of the excellent point VLADs makes is that large cap growth has outperformed for the last 15 years, and essentially the last 30, because of Big Tech.
So slightly preferring the larger index, and on the margin pooh-poohing SCV is mostly betting that big tech outperformance will continue forever.
Where relative overweighting in SCV is a hedge against a market down turn and the likely greater PE compression for big tech than for most of the rest of the market.
So since all your analysis in this piece was from a buy and hold investor perspective rather than from the perspective of an investor withdrawing from their account for retirement, it seems quite likely for me that it makes even more sense for a retired investor to overweight SCV in terms of optimizing for SWR, no?
Maybe the topic of a future post?
Welcome to the party, Andy!
I don’t require LCG to outperform SCV forever. For me, the expected returns are the same, which is why I hedge it with my total market portfolio (maybe slightly tilted toward large-cap), but I take no view on growth vs. value. I have only the blended varieties in all portfolios.
With my portfolio I have a hedge against both extreme outcomes, i.e., continued rally in tech when AI becomes the next big thing. Or AI was just a big fraud and comes crashing down like the dot-com bubble. Either way, I got half of my portfolio in the loser and the other half in the winner portfolio.
So, again: I’m not a tech fan-boy. I like to stay in the broad index to cover all my bases.
The thing that bugs me is that investing money into VOO today is very hard to swallow with near peak CAPE. I think mid and small caps are historically cheaper by comparison. As others have commented – it would be cool to see this type of analysis convolved with CAPE ratio for each of the asset types. Just assume in accumulation, you invest more in classes that are significancy lower relative to their own history and less in classes that are significantly higher. For drawdown assume you sell from the highly valued classes first. This is why I like to hold several classes in representative etfs, so I can use a diversity of historical valuations to help guide buying and selling decisions. You feel alot better selling assets that are at big gains than at big losses. Karsten – have you thought about this type of analysis? You use cape for SWR, but not differentially across the asset classes.
During accumulation, you should not time the market like that. If you’re close to retirement, you should be bothered by the high CAPE. The only reliable remedy is to reduce the SWR in that case. I don’t think that SCV is any guarantee.
Also, regular rebalancing will perform some of the tasks you mentioned. In retirement, if you sell during a bear market, you would likely sell out of bonds and not stocks.
Your post is timely, Karsten, despite all the heavy PhD talk which I’m clueless about LOL.
Yep, I read Paul’s little book about small caps a few years ago and I was inspired to change our AA in our Roth IRA’s. We didn’t though, but the media blitz as you call it refreshed my memory last year. The bigger reason why I directed 2023 and 2024 Roth IRA contributions to small caps was not because of tilting ($50k is a pittance for the size of our portfolio) but because of SP500 and Total Stock Index going bonkers and small caps were at 15-16 PE. In the grand scheme of our AA, having such a small amount in small stocks is more of a nuisance though, but I don’t want to sell and buy VTSAX or ITOT today either. Would you say this fear is a tad too irrational and I should just sell and add to VTSAX and call the day? We are also very underweight in International Stocks, but the fund we have had for 15+ years shows such a miserable performance that it feels as if I’d be throwing good money after bad…
It sounds like you would support J.Collins’s portfolio than Paul Merriman’s in this case. Choose one Total stock market index fund and you are done. That’s pretty much what I did in my teenager’s Roth IRA so I don’t need to overthink it, and it’s very easy to explain to a teenager if s/he is willing to listen LOL to the rambling about investing for the long term.
I have a different question though and hopefully you can opine on the below quote that I read yesterday.
“The market capitalization of US stocks is approaching $62 trillion — more than twice the size of the economy, and at a ratio that Warren Buffett explicitly warned about more than two decades ago… Buffett commented back in 2001 that ‘if the percentage relationship [between market cap and GDP]… approaches 200%… you are playing with fire.’”
I didn’t fully grasp it, so could you please share what you think the above quote tries to purport and what do you think about the situation? Is it an empire on the way of its own implosion?
TY!
I follow the CAPE ratio closely. It’s very high and would lead me to lower the SWR recommendation, as I have written elsewhere on the blog.
I think that your valuation comparisons between different market segments make it sound like SCV is really attractive. But this relative valuation is historically horrible at timing different stock market sectors. Tech will always demand a higher PE than utilities. Some of the low PE industries look cheap but they are cheap for a reason: they are stagnant and even dying industries, like much of retail.
So, I’m much less concerned about the different valuations between Tech and non-Tech.
“I recommend keeping it simple and using a broadly diversified equity index fund. A total market fund (FZROX, ITOT, VTI, VTSAX) works, though I slightly prefer large-cap funds like FNILX, IVV, VOO, etc. Happy Investing!”
is this based upon anything data driven or gut feel? did you write a post on this? I personally do the same.
I find ITOT and IVV almost interchangeable. I have S&P 500 index funds (or equivalent) everywhere I can but I also keep a few total market funds when those are the cheapest index fund (e.g., in one 401k plan I still have).
As I indicated in the post: total market actually has a slightly higher risk than the S&P 500. So, all else eual I pick SP500, but if the TMI fund has a lower expense ratio I am happy with that one too.
This is a fascinating exploration of small-cap value stocks and their role in portfolio diversification! The distinction between genuine diversification and diworsification is an important consideration for any investor aiming to optimize returns without adding unnecessary risk.
One aspect worth noting is how peer review services can enhance the credibility of financial research. Thoroughly reviewed studies on asset performance and diversification strategies often offer deeper insights and minimize the chances of bias or oversights. It would be intriguing to see how the conclusions here might align with or differ from peer-reviewed analyses on small-cap value stocks.
Thanks for this detailed breakdown—definitely food for thought for anyone reconsidering their investment strategy!
If you need further information, visit us at: https://www.journalpublisher.co.uk/peer-review-services/
This is an incredibly insightful analysis of small-cap value stocks and their role in diversification strategies! The detailed breakdown of risk vs. reward is a must-read for anyone navigating portfolio management. Posts like this highlight the importance of well-researched content that delivers value to readers.
If you’re looking to maintain such high-quality content while expanding your reach, investing in blog writing services can be a game-changer. They help keep the tone engaging and professional while ensuring consistent updates for your audience. Keep up the excellent work—this post is a gem for both new and seasoned investors! If you need any kind of help to reduce your stress for educational writing related work, visit us at https://www.authorsbreeze.com/blog-writing/
Hey, man! Great post. But I had to come here and defend my honor :-). I am NOT a small cap value fanboy. I believe that an approach closer to the efficient frontier ALWAYS HAS and WILL beat one “simple” ETF. There’s a reason why Dimensional Funds win. It’s NOT small cap value. It’s a more meaningful diversified approach based on history, not on “betting.”
Thanks for weighing in, Joe! In my old job, we applied mean-variance optimization, so our portfolio recommendation would be one point on the EF. For different preference parameters we would span the rest of the EF. So, this approach certainly resonates with me.
The major headache with EF analysis is the expected return part. Sure, we can take past returns, but the better approach is using future expected returns and variance matrix. Which brings us back to square one in the VTSAX vs. SCV discussion because nobody can agree on how much (if any) SCV “alpha” we should use. The same problem also shows up in past returns: If you use the 0-20 years of trailing returns, the realized SCV return is so low, it will never find a spot along the EF (lower return + higher risk + high correlation => zero weight). Even going back to the 1970s, the outperformance is small enough that it’s often penalized too much to get a significant weight. Using returns 1926-2024, one would certainly put a lot of weight on SCV.
Anyway, I have changed the text in the post to reflect your views on SCV and the Efficient Frontier!
The current 10 year record is: DFSVX 8.27 VSIAX 8.40.
Based on history why not stick with an index fund and eliminate the uncertainty of an active fund and the layers of additional expenses. Jack Bogle wins again!
Or just go with VTSAX and get even better returns. But yes, you make a good point: DFA funds are not special. They are actively-managed funds and have a hard time beating their benchmark.
Can you give your thoughts on Buffer Funds/ETFs? Are these funds easily replicable for the ordinary investor through the use of options? Have you considered using collars to protect your portfolio? Would love to get some thoughts on portfolio protection and whether it would reduce SORR risk.
They often have high expense ratios for very little work. One could replicate that easily. I haven’t studies it carefully enough and can’t endorse it either way as of now.
If replicated it would make the most sense to buy and sell the different parts of the collar depending on how high/low vol is. So, it would take some skill. But it looks promising.
It would be interesting to hear your views on TAA. There are TAA strategies, or mixes of them, that offer diversification from the standard buy and hold VTSAX/S&P approach, at least in how they offer non correlated returns.
Given market volatility and high valuations, I think adding some degree of TAA could provide not alpha but risk management that is difficult to get from bonds in a potential higher inflation environment.
Since you have experience in this area it would be great to hear your perspective on this, and which types of TAA strategies you think could be helpful.
I like the idea in theory. But doing this profitably will be hard, because that’s the holy grail in the hedge fund world. I’ve written about momentum here:
https://earlyretirementnow.com/2018/04/25/market-timing-and-risk-management-part-2-momentum/
Ideally, this should be mixed with other TAA signals, i.e., valuation, risk model, etc. It’s not easy for the average retail investor to get this right!
So…all good stuff, great analysis and an interesting read. However I do think you are directing the comments at Merriman / Risk Parity / Afford anything which is not really right. The theory comes from Farma and French, nobel prize winners, that would still argue the case for SCV going forward – its also 30 years old so not sure where the shinny/new comments come from as all the people mentioned have been saying this for years and years. People like Merriman and others are simply passing on information and he also typically always says that SCV will likely have long periods of underperformance which has always been the case – their numbers are also updated frequently on the website so cherry picking data is also not correct in my view. The article reads like they are the ones doing that have uncovered all this stuff – I think more transparent to clearly say your argument is with these other also very well qualified people rather then just the online FI community/you tubers etc.
A lot of confusion here:
1: Fama won the Nobel, French didn’t.
2: Fama won the Nobel but not for his Fama-French work.
3: Neither Fama nor French will ever claim that their various factors are guaranteed money making machines for all eternity. We will always find factor exposures to SMB and HML. The question is, are the SMB and HML factor still factors that pay a premium? And how big is that premium?
In summary: My blog post is not a hit piece in Fama-French. It’s a hit piece on folks misunderstanding Fama-French.
A lot of confusion here also from you.
1. Where did I say why they won the Nobel prize? My point was not to do with that; it is just to say they are renowned academics that would still support say there is a SCV premium.
2. Literally no one says they will guarantee a return going forward (certainly not Paul merriman who always re iterates there is no guarantee on anything) who is you comment addressed at here?
3. Farma and French would still say tilt to SCV. Dimensional would still say tilt to SCV. You argument is with these guys, confused how you think it’s not and aim your hit job elsewhere.
Maybe you should get this article peer reviewed, ask the guys at rational reminder to organise a proper debate with other academics that still believes there is SCV premium. I think that’s the arena to debate rather then target others who are effectively just promoting the DFA type strategies.
PS love your writings and followed for a long time and actually don’t believe you are wrong here either (I don’t know).
I just believe some of this article mis represents the people that are targeted. (Joe already said as much and pretty sure the others will point out errors in assumptions/accusations – eg: no one guarantees anything, merriman provides multiple options for people to make there own and doesn’t just say 4 fund/or change portfolio every 5 years, they update data frequently so don’t cherry pick, they have best in class recommendations for ETFs, this is not a shiny new topic for any of the creators involved)
1: “Where did I say why they won the Nobel prize? ”
You wrote that in your earlier comment:
“The theory comes from Farma and French, nobel prize [sic] winners, that would still argue the case for SCV going forward”
Did you forget what you wrote in your own comment? Besides that, their 1992 is an empirical paper, not a theory. They conjecture all sorts of theories that may account for this empirical observation, but nothing is settled. But that’s a whole other topic.
2: Merriman and most SCV disciples indeed come across as though they claim that. That’s my reading (hearing) of the Afford Anything podcast and many other appearances, like ChooseFI in 2019 (I believe). Just look at the title picture of that AA podcast appearance “Beat the S&P 500.” Many folks who hear about this come away with exactly that false narrative. Maybe Merriman is an innocent victim of the podcasters who make these bombastic claims. But he’s certainly not fighting back very aggressively against that narrative.
3: Do you have a source for that? Where do they endorse SCV? Where do they endorse a specific estimated average annual outperformance of SCV?
PS:
I will debate Paul Merriman later this month. Podcast to be released early in 2025.
Hi Big Ern,
I just listened to this podcast and near the end, Paul brings up DFSVX’s higher return than SPY over the past ~30 years. You said you’d need to do more research before commenting. Curious if you’ve had time to look into this fund further?
Also, I’ve recently been listening to Risk Parity Radio and I agree with some of your reservations about Frank’s analyses on various topics, particularly the lack of statistical rigor, limited historical data when backtesting, etc. (all areas where you shine!).
While I have not drank the “risk-parity-kool-aid” so to speak, he makes some points which sound very compelling, two of which I might actually act on. The first one is having a little bit of gold (thank you for part 34). The second one I’m still on the fence with. He recommends using long term treasuries for the fixed income portion of the portfolio instead of intermediate treasuries, corporate bonds, or total bond funds. His main reasoning is that, yes, the volatility is higher, but this is actually desirable because long term treasuries are more negatively correlated to equities than intermediate, and especially, corporate bonds so in market crashes this asset will likely rise and cushion the blow. I haven’t pulled the trigger yet on moving out of BND into something like TLT. What are your thoughts on this?
It appears long term treasuries perform better than intermediate and total bond funds, at least since 1986:
60/40 (BND vs IEF vs TLT)
https://testfol.io/?s=clNpSrq3Swo
But I’ve been a reader of this series long enough to know the devil is in the details and frankly I’m not an expert. Would love to hear your thoughts.
Thanks for the series,
Ryan
DFSVX would have been slightly better. But for the longest time, you’d have paid 1% extra expenses for a DFA-approved investment adviser. A lot of the advantage went out the window and into your DFA-adviser’s pocket, unfortunately.
Frank is a nice guy, but he’s wrong on pretty much all major issues.
He claims valuation doesn’t work and uses one single observation, a high CAPE in 2011 and high returns 2011-2021. He touts the SWR simulations strating in 1970, which is also useless and lacks statistical rigor because you have a sample size 1 to 1.5 retirement windows and you miss all the historical bad scenarios (1929, 1964-1968). So, I take anything from Frank with a grain of salt.
BND has corporate bonds, so there is not as much diversification against equities as with Treasury ETFs.
TLT with a longer duration should offer better diversification than IEF. You can check the different performance of 10y vs. 30y with my SWR toolkit. Intriguingly, 30y bonds don’t get you any improvement in historical simulations. The 1970s would have killed your withdrawal rate.
But if you cross your fingers and hope that the next recession is a demand shock recession again, you will do better with 30y bonds.
Haha I knew what I wrote and I didn’t say a Nobel prize was won specifically for their research on factors (just that one had been won!). I repeat my point was that these renowned academics are the people the debate should be had with rather then those targeted.
Eugene Farma, Ken French and David Booth sit on the board of Dimensional. DFA use their research to produce portfolios that implemented thier research to try and capture the 3/5 factor models premium.
Farma will always say he can’t confirm if the premium will persist and we won’t have the data for years and it could be arbitraged away (in also says stock premium could also be the same if enough people decide stocks are less risky). But the very fact he sits on the board I believe answers the question
Looking forward to the future debates!
Yes, thanks for confirming. We agree on essentially all major issues.
Maybe on the principles, but still not on way the article was written which I think mis represents some views so will be interesting to hear the debate, even better if can discuss with the academics or DFA researchers via rational reminder.
Turning to shiny new things, would be interesting to have your views on managed futures etf (DBMF) and impact on SWRs, maybe a future article?
What would rational reminder contribute?
Managed futures:
I looked at four of these funds:
First Trust Managed Futures Strategy Fund (FMF)
iMGP DBi Managed Futures Strategy ETF (DBMF)
KraneShares Mount Lucas Managed Futures Index Strategy ETF (KMLM)
WisdomTree Managed Futures Strategy Fund (WTMF)
One did well, but the other three not so. Not sure I would buy an ETF in that space.
But I can see the case for momentum strategies, see here: https://earlyretirementnow.com/2018/04/25/market-timing-and-risk-management-part-2-momentum/
Thanks re managed futures.
Re Rational reminder they had people from Dimensional and Avantis on previously. Also Fama, French, Booth. I’d be interested to see if they could broker a discussion on this topic, especially focusing on value premium and all recent research on this including yours. Maybe all would come to same conclusions but could be interesting and maybe useful to the general community as I don’t think people are aware of some of this detail. (Maybe rational reminder aren’t needed but best podcast I’m aware of to get something like this sorted)
Thanks. That’s useful.
I have seen the discussion Rick Ferri vs. Merriman on the WCI podcast. But this topic needs more discussion for sure.
Following up on this.
I would be glad to hear your take on Managed futures. The leveraged ETF community on reddit uses a lot as a diversifier, and it did really well in 2022 when the equity markets were tanking.
While I understand this is can of worms given there is no one homogeneous “recipe” to these ETFs as you pointed out in your post, I am wondering whether adding 5-10% of them to my asset allocation might not be a good idea, especially given how high equities currently are and considering the level of uncertainty on yields direction going forward.
5-10% would not be enough to make a meaningful difference. Either we run this momentum strategy on essentially all of the equities, or it’s not enough to deal with Sequence Risk.
I also looked at four ETFs in that space.
First Trust Long/Short Equity ETF (FTLS)
First Trust Managed Futures Strategy Fund (FMF)
iMGP DBi Managed Futures Strategy ETF (DBMF)
KraneShares Mount Lucas Managed Futures Index Strategy ETF (KMLM)
Only the DBMF has attractive returns.
Another great article! Can you elaborate why you prefer S&P 500 funds over total market funds? Is it simply for liquidity and options trading or do you find the additional diversification in a total market fund de minimis.
Additionally can you provide a new update to your asset allocation? Last update you provided in 2016 stated you had 7.2% in total market funds vs 37% in S&P 500. Has this changed?
It refers to the side note above: the two are almost interchangeable. But VTI has slightly higher volatility than the VOO.
I have 70% in stocks, 30% fixed income in retirement. The equities are about evenly split between total market and S&P 500.
Hello Karsten, thank you for the interesting article. In Europe, the VWRL or VT is often recommended as a choice for simple index strategies in the FI blogs. Looking from your old home country, would you also recommend the more international orientation of the index fund or would you also rely on VOO?
I’m a bit pessimistic on non-US stocks. Sure, they got better PE ratios, but also much less growth. Just look at the anti-business government in Germany.
So, yes, I’d rely on the VOO more than the VT.
Question on value vs growth more generally speaking – not just small cap value.
In your table you have dates back to 1926? with small cap and large cap value. It has higher returns than the sp500. Of course in this post you mention that more recently and for a while, small cap/regular value has underperformed vs the regular sp500.
1. What will it take before this mindset of ‘value’ which inherently makes ‘sense’ goes away and shifts towards just ‘good’ companies with growth at ‘reasonable’ valuations? Like when will the ‘research’ and grey/white haired finance experts teaching in universities start saying to have a growth tilt due to historical outperformance?
Does growth perform better say in a post WW2 on rolling start periods?
2. My initial guess is that the past had more physical goods/high labor companies and the internet/lack of computers enabled information to be ‘hidden’ while now valuations are arbitraged away and now the really hidden information is luck / deep knowledge of an industry and what products are good.
I’m afraid that the value crowd will always be there. There’s money to be made, and a lot of financial advisers push this as their shtick to convince investors that they need professional advice.
But I agree with you: there could be better definitions of value. As I pointed out in my old post in 2019, value has a terrible sector bias. Maybe come up with a sector-neutral value fund. Smart investors are working on that already. But it will take time for better value/growth distinction to arrive in the retail sector. Until then, I’d stay in the overall market! 🙂
I’ve followed Paul Merriman over the years and I can tell you why Merriman went from touting a 10-fund “ultimate” portfolio to a 4-fund version. It is not nefarious. He was criticised, including by John Bogle (Merriman cites this often), on the grounds that 10 funds was unnecessarily complicated for a DIY investor and thus counterproductive. A Boglehead on the forum years ago came up with a 4-fund alternative to Merriman’s portfolio with approximately equivalent risk, volatility and expected return. Merriman later started his foundation and brought in some retired-engineer sidekicks who back-tested that 4-fund scheme and found that it was indeed a simpler, near-exact replacement for the 10-fund version. Four funds slowly became the main portfolio he’s touted, alongside a 2-fund approach that involves a target date fund plus SCV.
Merriman seems to have had a quasi-religious conversion when he attended Dimensional’s seminar for financial advisors in the 90s. That’s where he learned about Fama and French and small and value. He has been a vigorous proselytizer for that idea, but his understanding of academic literature seems pretty superficial, as I think he would agree.
So I don’t think Merriman will be a match for you in an argument–he really hasn’t responded at all to the points raised by Rick Ferri in that “case against” presentation you mention.
I would love to hear it if you could arrange to debate the viability of “factor investing” with a proponent who has familiarity with the literature and some relevant academic training. I think that could be really clarifying for those of us trying to design our retirement portfolios. Larry Swedroe, maybe, or one of the principals from DFA or Avantis? Or an academic like Fama himself?
Thanks for all of your work here.
Thanks for the clarification. That’s most helpful.
My suspicion was that because the 10-fund portfolio also included some REITs and international funds that all lagged terribly behind the S&P500 recently, they did the switcharoo to make the results look more palatable post-2009.
I will debate Paul Merriman in late December on David Baughier’s “forget About Money” podcast. I’m open to discussion with anyone, but i would never invite myself on any podcast. I usually get invited. 🙂
There is a hint of the switcheroo you suspected: the original ten-fund (and four-fund reduction) was half international, but they now often point to the option of an all-US variant. I think Merriman still prefers his original concept of 50/50 us/international, 50/50 small and “blend.” Beyond including small value, Merriman really doesn’t have a clear portfolio recommendation: he offers charts describing many possible portfolios and urges listeners to find one they will feel comfortable sticking with.
Understood. Thanks for providing further background!
Hi Karsten,
Thanks for your work and sharing the findings! I may not fully capable of understanding the PHD level mathematics. But I like the conclusion.
For average retirees without financial or investment background, keeping it simple would be best approach. People can have large cap, small cap, bonds, TIPS etc in portfolio hopping to achieve the last 0.05% withdraw gains. Would it be just having 75% SPY and 25% BND that could achieve almost the same withdrawal rate? I would like to hear your comments.
Thanks,
John
I recommend 60-80% S&P500 (or total market index like VTSAX, not much difference) and the rest in 10y US Treasuries or equivalent. BND has some corporate bonds in there, so 75% stocks + 25% BND will likely behave like my 80% stocks +20% 10Y Bonds. But it will still hedge against the worst.
Guten Tag aus Deutschland Carsten! Would you recommend part of that 60-80% equities allocation, about 10-15% in a gold etf like you mentioned in part 34 if they were transitioning to decumulation and could do so without tax penalties etc?
Yes. Historically, a 10-15% allocation to gold has smoothed the volatility both in inflationary and disinflationary recessions. The rest in bonds.
Viele Gruesse zurueck nach Deutschland!!!
I get more confuse everyday in this FIRE movement. One say you can’t look only the past 10 or 20 years but the whole history. Now comes Big Ern saying that SCV is underperforming lately so we shouldn’t use it.
Man, what is it. long term or short term? I’ve gone from 100% VTSAX to 35% LCG, 35% SCV, 25% 10YBonds, 5% Gold following the “latest” trend. Did I go wrong?
Most FIREEs aren’t financial guys, we have regular jobs. We need to rely on people like Ern and Vasquez who actually run scenarios and tell us what to do. Otherwise I’ll use a TDF or Robo. Please help make sense of things
Sarah, I feel the same way (and get confused) sometimes. I also follow Frank Vasquez from Risk Parity Radio, who, for me, is one of the most brilliant and down-to-earth financial minds in the podcast world today. His insights on building better portfolios using modern tools and funds really resonate with me. The principles behind the “Holy Grail” approach are particularly useful.
I know Kersten and Frank had a bit of a disagreement a few months ago, but I’d love to see both of them—along with Joe and JL Collins—in a podcast together to debate and defend their ideas. It seems like their approaches have diverged recently, which would make for a great discussion.
For me, the most important goals are maximizing my safe withdrawal rate (SWR), while minimizing drawdowns and the time my portfolio spends underwater. For others, it might be all about maximizing compound annual growth rate (CAGR). It’s such an interesting balancing act and who knows, in the end they might all be right, it just depends on what you’re looking to maximize and minimize in your portfolio construction
Frank is a funny guy, but unfortunately confused about a lot of the mathematical details, such as:
1: he keeps claiming that equity valuations don’t matter for equity returns. But they do, and especially for SWR calualtions. See Part 54. Actually, almost all parts of my series.
2: he claims that owning a house reduces your inflation rate by half. That’s ludicrous. See part 57.
3: he spreads this nonsense about risk parity and how it raise your SWR to 5%. All without any simulations. Or, correct that: he has live returns since 2018 or so but they look atrocious. They lag behind a simple 75/25 portfolio.
Did you not have a discussion about gold with Frank and then once you did the analysis accepted adding a bit of gold had helped safe withdrawal rates? What was your starting position in that debate? He has done loads of simulations via portolio visualiser and portfolio charts (and your own spreadsheet 🙂 the house inflation thing I’ve not heard him say, sure maybe he has but I’ve listened to his podcast for several years now and not once remembered him saying it. Any person would say simulation from 2018 doesn’t count for anything, why would you bother making and comment on that around 75/25 over such a small period? Odd. (He is simply tracking portfolios in real time out of interest)
I did not have any discussion with Frank. I wrote my Gold post (SWR Series 34) on my own initiative.
Portolio Visualiser and Portfolio Charts are junk simulation tools for SWR analysis because they start only in 1970.
Yes he has. Here in ep. 209, between the 34:00 to 35:00 time stamps:
https://www.riskparityradio.com/podcast/episode/75658718/episode-209-popular-swr-delusions-and-the-madness-of-gurus-and-portfolio-reviews-as-of-september-30-2022
As always in statistics, the number of observations necessary before you can draw useful conclusions is situation-dependent:
For SWR analysis you need 100+ years, correct.
For determining if an alpha source is worthwhile you need much less data.
Thank you, as always, Karsten for your priceless analysis and point of view. I have to admit that I drank the Kool-Aid on SCV based on the analysis and statistics cited on Merriman’s presentations and had about 10% of my asset allocation allotted to SCV until your analysis. Thank you for presenting a more thorough recent analysis (past 50 years) of what one could expect by adding SCV to one’s portfolio. I have seen the error of my ways and will be reversing course.
Mark Twain is rumored to have said “There are three kinds of lies: Lies, Damned Lies, and Statistics.” I appreciate a more thorough statistical analysis of SCV. I certainly don’t attribute any malice to Mr. Merriman’s recommendations. I met him (as well as yourself) at the last Bogleheads conference. He is a very kind man whose heart is in the right place.
Thanks for the feedback! Glad you got something out of the post and seeing me at the Bogleheads conference. Yes, Paul is a great guy. Which makes it particularly painful for me to point out some of the SCV shortcomings! 🙂
I went from 100% S&P 500 to an equity split of 70% S&P 500 and 30% “extended market index” (mid and small caps) in November last year to allow for the possibility of US-focused companies (small and mid cap) outperformance over the next few years, especially with S&P CAPE so high meaning potentially lower going-forward returns. This is a slight tilt over Total Market in terms of mid and small cap. If the “small cap crowd” ends up being right this will pay off (modestly) and even if not it will probably won’t be much different than S&P only. It also let’s me rebalance any excess returns of small/mid back into the large cap allocation, especially if there is a correction in the Tech sector. Just my thoughts. Completely different view than “small cap value for historical reasons” though.
As an update to my comment above I’ve re-read your analysis a few times and have changed back to a simple indexing approach and will stick with that as that is what is suitable for me. Thank you so much for your articles on this site.
Good! Congrats!
Using different backtesters and considering your Part 34 I have settled out on an AA of 60% large cap Stocks, 15% Bonds/10% Cash, and 15% gold. For me personally this lets me sleep better and not try to tinker (which has been a problem for me). Your research helped me understand how this type of AA had behaved in a lot of different market conditions and I’m comfortable personally with that (and the trade-offs versus other approaches). Thank you!!
Awesome! Thanks for sharing! It does help if you run backtests and see that the plan would have survived the historical worst-case scenarios! That helped me get over the fear of leaving my job behind! 🙂
Yes, that’s a bit more than the ~15% market cap outside of large-cap. Good luck. I’m not saying SC will underperform. You will likely be just all right and in line with the overall market.
This is my first time reading this blog. Paul Merriman believes that SCV produces diversification and your response to that was inaccurate. You wrote that because there is small cap value in the S&P 500, that diversification is accounted for. The median market capitalization of the S&P 500 is 29.4 billion (vs. less than 8 billion for small caps), so the amount of small cap value in the S&P 500 is way too small to add diversification. I love Paul’s 2 funds for life portfolios and will stick with those, as they do offer more diversification than the S&P 500.
My bigger beef with the diversification claim is the pure statistical story: that’s something you can prove mathematically and small-cap routinely fails that criterion.
But I concede that different weights than the market cap can reduce risk, even though you may have concentration in certain parts of the market. But you’d want the concentration in the low-vol parts (say with min-vol ETF). But Small-cap and small-cap value are much too volatile.
Hi Karsten –
Many thanks for your comprehensive analysis. I just watched the “Forget About Money” podcast with the discussion between you and Paul Merriman. Paul Merriman seemed to be proposing that at least recently there was a SCV premium vs S&P500 provided you invested in the “right” type of (actively managed?) SCV fund/ETF and that e.g. Vanguard VSIAX or VBR wasn’t it. There seems to be differences in the SCV indexes followed by different funds /ETFs, so is this cherry-picking based on historical data and a concern given that almost all active funds underperform over longer durations.
SCV vastly underperformed during the last few years, even when using the “best” SCV funds/ETFs. Only when going back to the 1990s you can find funds with slightly better performance than the S&P500.
This is exactly my big concern: hindsight bias and also survivor bias (only the successful SCV survived, while the awful ones available in the 1990s shut down an nobody knows about them anymore).
You make the choice seem arbitrary and indicate survivorship bias, but different indices and funds have different factor loadings. This is apparent if you compare VIOV to VBR. Both are passive, so no active management idiosyncrasy to find.
Yes, they may have different factor loadings. So, it makes it easier for unserious financial experts to cherry-pick their preferred comparison index.
I just listened to Paul Merriman’s podcast rebuttal to this article. He has some pretty good points, so I encourage you to check it out. Striking to me, is that you agree SCV performed much better than the S and P 500 for the majority of the 1900s. Also, SCV beat the S and P the 1st decade of the 2000s. The next decade plus a few years the S and P outperformed. Why does it follow then, that the S and P will win out again for the next decade? Sounds like recency bias to me. EVERYBODY and their brother says these days the S and P is it, or VTI (which is mostly S and P 500) is it, based solely on what has happened for the past decade. Very short sighted. Maybe Internationals or emerging markets will be top dog the next 10-20 years. Does anyone know for sure?
I never claimed the SPX will win against the SCV funds. I merely fight against this fairy tale that SCV will return to its old grandeur and outperform the SPX by 1.7% per year.
SCV has now underperformed the SPX for 3 full market cycles. Statistically, the SCV funds are not decisively different from the SPX. That’s all I’m saying.
This is on point. Thank you for bringing the focus back to what the argument is about.
I think we’re on the same page! Whew!
Also want to say the writing style of this article is pretty angry and attacking (and maybe even arrogant). It is difficult to read, and it feels like much of it is there just to show you know (or claim to know) a lot about statistical analysis. The level you go to is almost like you are wanting to write for a peer-reviewed scientific journal. Any journal would reject this piece based on pot shots and exclamation points alone. You have not convinced me to believe anything with this article. Got a punchline for you: 5yr return S and P 500 is 81%. 5yr return of AVUV (scv) is 82%. Didn’t need much to get that point across.
As someone who’s actually published in some of the top journals in economics (JPE, JME, JMCB, etc.), I don’t need your advice on writing style and publication strategies. I know that this is a blog and not a journal.
Got a punchline right back for you: Your AVUV vs. SPX comparison is apples to oranges. I could have also picked a more actively run large-cap fund that would have mopped the floor with AVUV.
Points taken. I realize AVUV performance past 5 years kept pace with the S&P 500, yet did not outperform at a premium, which is what the whole discussion is about. So, not great for me to call that a punchline. Yesterday I watched the new YouTube Video where you debate SCV premium with Paul Merriman. I have to say I was very impressed by your thoughts and honest take on how you feel this topic should be seen from a scientific point of view. Paul admits he has faith and not trending proof that the SCV premium still exists. Your view seems more skeptical, requiring proof SCV premium is still valid. Have you done additional research after the video and have you come to any new conclusions? Are you aware of any research being done by others to prove/disprove whether SCV premium still exists or will exist in the future? I am genuinely curious as I have 33% of my portfolio in AVUV. I am 47 and plan to glide down to 25% AVUV over the next 10 years. If the SCV really is dead, maybe I should glide it down further to 10%? I do plan to be a scientist my self and will be watching for more evidence as the years go by.
Yeah, you put the finger on the wound there: I’m faithful, too. At church. I’m an evangelical Lutheran and I believe in Christ, my Redeemer, no questions or doubts allowed. But in finance, we should constantly check our assumptions. I wanted to press Paul on what it would take for him ta change his views and he dodged that question.
Again: I don’t want to talk people out of AVUV or the DFA SCV funds. My working assumption is that they will perform in line with the broad market. Maybe a bit above to compensate for the additional risk. We will know after the next market cycle.
Not all indexes are made the same. I really like the S&P 600, S&P 400 for small and mid cap. These indexes also have a profitably requirement like the S&P 500, which should cause it to have some of the quality factor. These indexes routinely outperform others in their respective categories like the Russel 2000. Likewise, the S&P 1500 index, SPTM ETF, I believe is the best total market index. I’ve seen the historical performance since the indexes were created around 1992, (which outperformed the S&P 500 but with more volatility) but I’ve never seen the S&P 600 and 400 back tested back to like the 1920s. I wonder if you could perform that backtest and see if these higher quality small and mid cap indexes made a significant difference in returns and sequence of returns.
Yes, quality would have helped recently. But that’s also hindsight bias. Part of the appeal of Value is the the junk stocks rally very nicely in the next expansion. By doing too much quality control we might miss the boat.
Because I want to cover all my bases, I still prefer the blended index for that reason!
What if one’s objective was rather to minimize volatility in an all-equity portfolio? Could including some VBR or other specific ETFs in the portfolio reduce standard deviation compared to 100% VTI, as per MPT?
If that’s your objective there might be better alternatives, like Min-Vol ETFs, e.g., USMV.
I’m concentrated in small and mid caps (not SCV) for the much lower PE ratios, not for the historical performance. As noted in other ERN studies, valuations make a big difference in SWRs.
Yes and no. It’s true that S&P 500 PEs and CAPEs predict SWRs and future returns. There is no guarantee that this relationship also holds in the cross-section. In other words, you can’t compare the drawdown risk and Sequence Risk of an S&P 500 portfolio at CAPE=18 with a SCV CAPE at 18, when the SPX CAPE is above 30.
Being the suspicious type, I wonder if there is some ulterior motive for the proponents of SCV “pushing” Avantis and Dimensional small cap funds. Its so contradictory to the Bogle philosophy of sticking to low cost index funds.
Here is information I pulled up comparing the DFA small cap value fund DFSVX, with Vanguard’s Small Cap value index fund:
1 year: DFSVX 7.15 VSIAX 10.44
3 Year: DFSVX 7.35 VSIAX 6.65
5 Year: DFSVX 16.10 VSIAX 13.01
10 year DFSVX 8.27 VSIAX 8.40
Case closed! If you must invest in SCV stay with Vanguard!
Yes, I agree. The SCV, DFA, and Avantis stuff is the continuation of an old trend: sell folks some new shiny objects. True, there were some decent returns in the past, but everything looks crummy in the more recent data. Do VSIAX is you like. I would skip the VSIAX and go straight to VTSAX, though. 🙂
This is just attributable to the different factor loadings of the index that VSIAX follows compared to DFSVX. It’s well-established that SCV has underperformed the total US stock market (although I attribute that more to a LCG bubble than anything). VSIAX/VBR tracks the CRSP US Small Cap Value Index and simply has less exposure to SCV and more mid caps. Compare with VSMVX/VIOV, which tracks the S&P 600 and has somewhat better factor loadings. I would be careful about conflating the realized returns of two funds following two different indices with the performance of an asset class.
Hi, and thanks for the great analysis! I agree that the recent under performance of SCV is concerning. However I was wondering what you would think about adding a very aggressive LCG ETF like QQQ (which has done significantly better than standard growth ETFs like SCHG or VUG) to balance it out? For example, using testfol.io to simulate QQQ back to 1995, 35% QQQ, 35% DFLVX and 30% DFSVX has significantly outperformed the S&P 500 since 1995, and even starting with the massive bull run of LCG in 2010 it still very slightly outperforms. Starting in 2015 you get a 0.5% underperformance, and starting in 2020 you get a slight outperformance again. It seems that this way you can get a large part of the benefits of SCV (if it still exists) while barely under performing if does not, or does not for extended periods.
I’d love to hear your thoughts.
Adding the QQQ and SCV you’re likely back to just the total market again. Then just invest in the ITOT or VTI and be done with it. But I understand that tinkering with the portfolio like that is more fun. 😉
I (and I suspect any others) would appreciate your comments on Episode 223 [corrected the episode number] of Risk Parity Radio, where Frank summarizes what he says was a Zoom call and some subsequent emails exchanged with you, but since he didn’t actually have you on as a podcast guest, I am wondering if any of his characterization of your positions are different than you’d describe them.
Most interesting I think is that he described a portfolio including Fama French factors and gold which showed 1926 backtesting results in your SWR toolbox of nearly 5% SWR for 30 years. He then said that if he adjusted the CAPE range to 1926 and later (claiming that the earlier CAPE data you had didn’t go back further than 1926 so wouldn’t work correctly if factors are non-zero), even CAPE>20 using your modified CAPE showed a nearly 5% SWR with 0% failure rate.
I tried duplicating the toolbox results using your latest toolbox, and it seemed to closely match the results he claims.
So I am wondering your thoughts on using that portfolio and approach (reduced volatility closer to efficient frontier) to reduce SORR and thus increase SWR, since it appears to achieve significantly higher SWR than a simple S&P500 / bond portfolio.
Is there a mischaracterization here? If not, what if any objections do you have to that portfolio, even if it significantly boosts SWR?
Sorry I meant episode 223 of Risk Parity Radio above (replying as there is no way to edit to correct my typo).
I corrected the episode number in the other comment. Thanks.
And (one more typo correction) adding words between asterisks: “(claiming that the earlier CAPE data you had didn’t go back further than 1926 *for factors* so wouldn’t work correctly if factors are non-zero),”
Noted
Ah, yes, Frank. Bless his heart.
He uses my toolkit without knowing how to read the numbers.
1: Small-Cap value had outsized alphas during the 1926-2006 era that will likely not repeat.
2: His new portfolio generates a sub-4% SWR in the 1930s, so conveniently ignores that.
I have written about the weaknesses of the Risk Parity and Small-Cap value method in Part 34 of my series. I have nothing else to add.
I understand (and agree with) your point #1 – skepticism that the SCV premium still exists, and that it does not provide diversification, despite the claim that it does. Though I’m having trouble understanding how SCV not providing diversification is consistent with the toolbox result I reference below. I am not advocating for the below portfolio – just trying to understand the result from your toolkit.
However, to clarify, I would not have submitted the above question if I had seen your #2 above to be the case with your current SWR toolkit. Before asking my previous question reply, I duplicated what he said on the referenced podcast episode (timecode ~10:30 on Spotify), and the latest version of your SWR toolkit currently shows 4.91% in the 0% failure rate, since 1926 cell (F20), not sub-4%. It even shows 4.91% in the 0% failure rate “All” cell (E20) if the ‘SCR time series’!$E8 formula is changed to ‘SCR time series’!$E667 in E20 (which Frank claimed was necessary to match the timeframe for “Fama-French Style Factors. Only post Jul-1926” in what is now A17.
Specific numbers to fill in to get above result are:
Stocks: S&P 500, US large-cap 55%
Bonds: 10y U.S. Treasury 30%
Gold 15%
Fama-French Style Factors. Only post Jul-1926
Small Stocks (SMB) 30.00%
Value Stocks (HML) 27.50%
Retirement Horizon (Months) 360
— all other portfolio inputs left with default values (most 0%)
and the above change to E20 formula for aligning SCV CAPE to 1926+.
That result does not seem consistent with your statement #2 above.
So, does it devolve into just your point #1, or have I missed something regarding #2?
If just #1, I’m at a loss as to how useful the Fama-French factor inputs are in the Toolkit if you’re saying you don’t believe that the result the SWR boost they’re showing remains relevant.
Thanks!
With those parameters, the pre-1926 SWR was 3.54%. Lower than with a 75/25 portfolio. If SCV again produces only 0% outperformance, you’re cooked. Your SWR isn’t 5% but only 3.5%. But hey, if you like Frank so much, please follow his advice and not mine and withdraw 5% out of a whacky portfolio. Also, keep in mind that this is for a 30y with 0% final value. Not a good idea for early retirees.
It is also a totally false claim from Frank that the pre-1926 cohorts can’t be used. The SCV was simply 0 before 1926. Just like the realized SCV was 0 during the last 30 years.
About the 1930s: That was the case with another simulation I’ve seen previously. But it’s true that his current one avoids that issue. Data-snooping and hindsight bias to the rescue!
To be clear, I am not following Frank’s advice – in fact, it is due to my skepticism of it, that I am investigating further using your SWR Toolkit (since your toolkit is so thoroughly researched and back-tested, I consider it a great way to back-test his claims). And I am definitely not in agreement with using 30 year planning horizon for a longer FIRE RE period!
I am simply trying to resolve the “he said / he said” that he referenced in episode 233 regarding his claim when using your SWR toolkit. Since I was able to reproduce the result he claimed with your toolkit, I just want to understand the disconnect, since I know you don’t agree with his conclusions.
I interpret your conclusion to be that it is invalid to change the CAPE lookback formulas to only go back to 1926 even when using the Fama-French factors. Without that change, the CAPE>20 (with or without SPX high) and thus All SWR values drop back down to 3.54%. Did I interpret your conclusion correctly?
In that case, how should we interpret “Only post July-1926” in A17? That SCV was actually 0 before that, or that there is no SCV data before then, or something else?
Also, re your hindsight bias comment, I’m interpreting you to mean that you believe the high since 1926 result using the combination Fama-French factors is overfitting the backtest data, and therefore not something that can be relied on for forward looking prediction? Is that what you meant?
Thanks again.
And, of course, I meant episode 223 above again (I’m beginning to think autocorrect has it in for me 😉)
No worries. 223 or 556, it’s all the same.
The disconnect is not the toolkit. It’s the interpretation of the toolkit results.
The Fama-French factors were not replicable in 1926, so I don’t even believe the 1926-2006 performance is to be used unfiltered due to transaction costs, data delays, etc.
Moreover, there is no way of telling what the premium was before. So, a 0% premium is not a bad assumption. Fun fact: right out of the gate, there was a bad drawdown in the Great Depression.
In that case, why did you include Fama French factors in the SWR Toolkit at all?
Sounds like you wouldn’t trust any result which included them due to the reasons you listed above and backwards looking bias (overfitting).
Is there any use case you would trust for filling anything non-zero into the Fama French factors?
I can include whatever I want. If people misuse/misinterpret the results afterward that’s their problem, not mine.
I have made some changes in the Google sheet to allow model the FF factors without the outsized outperformance. Take out the SMB and HML trend with a HP-filter.
Thanks! Of course you can include whatever you want! It’s your blog, and your SWR Toolkit. Just trying to figure out how to interpret the 1926 backtest 0% failure cell (F20) when using the FF factors above.
Using the new HP-filter (Hodrick-Prescott?) selector to smooth out the FF outsized performance which you believe may no longer be present (for numerous reasons you’ve written and spoken about here and on various podcasts), it lowered F20 in the above example from 4.91% (raw) down to 3.44% (HP-filter — assuming I entered everything correctly), which is now well within the expected SWR range. Even changing All/CAPE formula to only start in 1926 doesn’t increase the 3.44%.
So, based on everything you said, it sounds like leaving the now default HP-filter on aligns with your expert opinion about the most appropriate assumption about FF premium, which is that it wasn’t practical/possible to use it for much of the time in the past (due to factors you stated above) and may well be gone (if it wasn’t just hindsight bias/overfitting in the first place!), so shouldn’t be assumed for the future. And the SWR Toolkit default values now match your expert opinion.
Thanks, that was just what I was looking for.
Whew, I’m glad we’re on the same page. You were one of the harder to please customers here recently, haha!
And just for the record: we may assign a slight premium to SCV. Maybe 0.50% each for HML and SMB. But we will likely not see sustained outsized excess returns going forward.
Thanks again for sticking with it/me and adding the hp-filter to make your SWR Toolkit match your conclusions by default even when SCV is entered.
I believe the addition of hp-filter default will reduce future misinterpretation by others, especially those who already trust your expertise and conclusions and use your SWR toolkit, without having read or fully internalized your comprehensive SWR series. I personally have read several parts based on topic specific interest and curiosity, and continue to read more, but am nowhere near even halfway through. And yet, I believe the SWR toolkit spreadsheet is very useful to me, and even those who only read a couple parts or only hear you on a podcast and jump right to part 28, and hp-filter helps minimize misunderstanding in those cases.
I would recommend adding a brief explanation of hp-filter into the text of part 28, so those downloading the spreadsheet will know why hp-filter is there and what it vs raw means. Since you recommend users of the toolkit regularly download new copies to update their analysis, anyone using SCV and downloading a new copy may be in for a surprise, and part 28 text explaining hp-filter could proactively address that.
Thanks again.
Thanks!
I updated SWR28 and added the explanation of the HP Filter method and another dire warning not to take the FF Factor returns at face value.
Thanks. Your update to SWR28 looks good. It should well address the concerns I had raised above.
Whew! I’m glad I could calm the fiercest ERN commenter in a while! 🙂
I finally listened to your SCV debate and had a thought on why “2006” seems to be a turning point / head-scratcher as you said. Do you think this has to do with the “rise of passive” to a higher % of total investment, and specifically Total Market funds now included in more 401ks etc? It seems like that could increase the correlations between small and large stocks because wouldn’t a large-scale sale of a total market fund for example create a marginal sale on small cap stocks too? In other words in the past index funds were more 500-index oriented meaning someone selling large cap wouldn’t necessarily mean someone also selling small cap at the same time. With everything in the same total index you would expect the same type of drawdowns at all cap levels on down days, which is what it looks like actually happens now.
The problem with the “passive” theory is that the move to passive investing has been gradual and started way before 2006. Why would the SCV premium flip like someone flipped a light switch?
My theory: In 1993 the premium became popular. For another 13 years it became something of a bubble, because more people wanted to harvest it and this blew up the bubble even more; a self-fulfilling prophesy. Since 2006 all the ill-gotten gains have disappeared again.
If we look outside the US using your 2006 start date, DISVX has beat VEA by 1.36% per year. DEMSX has beat VWO by 2.41% per year.
I’m not saying that SCV sucks everywhere and all the time.
Agreed. That’s further evidence to me that the effects he’s plotting are more attributing to uncharacteristic LCG out-performance in US equities than SCV under-performance.
It’s both. LCG outperformance is implies SCV underperformance. And vice versa. It’s a distinction without a difference.
Looking at the chart with the x-value the starting point of each simulation, all of which end 10/2024.
Isn’t this just the effect of the relatively recent out-performance of US large-cap growth stocks? As you move along the x-axis, the last decade or so exert increasingly more influence on the CAGR. That doesn’t seem to me like evidence that a small cap value premium disappeared when it’s more directly explained by the fact that US large-cap stocks are overvalued and due for a correction.
Which do you think makes more sense? Emphasizing the more recent return data or the 1926 return data? The latter is what all of Merriman’s charts do, by starting each cohort in 1926.