Update (12/31/2022):
It seems that new users have trouble registering for the forum. The confirmation email never shows up. I’ve decided to phase out this forum and transition over to a new forum plugin. The new location is here:
https://earlyretirementnow.com/forum-3/
Policies – please read!
- Everyone can read posts, but to start a new topic or add a reply you’d need to register. To create an ID, please verify your email address. You will get a link at that email address and you can then pick a password.
- Be courteous to others. Treat others like you like to be treated. Discuss the issues. No ad-hominem attacks!
- It’s OK to include external links if they are relevant. But please avoid spamming! Please no affiliate links.
- Before starting a new topic, please check if that question/topic has been discussed before already and add to that discussion instead of starting a new topic.
- If you’ve written a cool blog post that you want to share with others please post this in the “Self-Promo” category only!
- Please read the usual Disclaimers and the Privacy Policy!
- The forum policies may be amended in the future!
Simple 10MMA for variable equity allocation
For those interested, attached is a version of the hacked spreadsheet using Meb Faber’s simple 10 Month Moving Average (10MMA) rule for market timing. Following I will attach a second sheet with some results for 4.0% and 4.5% IWR.
Using bonds (as opposed to gold) with this method and a 4.0% IWR again “saves” the 4% rule*, with all of the worst cohorts. Though here unsurprisingly it is the Nov 1965 cohort that is the worst, averaging just a hair under 4% at 3.966%.
I find it interesting that with this method, bonds make a consistently superior ballast for stocks than does gold. Not at all surprising for the Great Depression, nor for the 1980s where gold and nominal bond yields peaked at the beginning of the decade before descending, but even for the 1950s and early 1960s. I speculate that part of the reason is that given the allocations switches from 100% to 0% - the impact of gold going down during the early 1980s during those period when the S&P is below the 10MMA for stocks is profound.
[NOTE: the calculations for the first 8-9 rows for any cohort may be incorrect, because I have not calculated the actual 10MMA, but rather only the MMA starting from the single month preceding the retirement start, leveraging ERN’s Case Study Sheet as is.. That said, for those "worst case cohorts which usually start at at least a short term stock peak, it mostly eybealls as about right.]
I still think I prefer my “homespun” variable equity allocation alternative using gold as ballast - despite the risk that my solution might be somewhat “overfit” - since it seems much more likely to me given today’s nominal and real interest rates that the challenges we face going forward will be more like 1965 or 2000, rather than the deflationary / disinflationary environment of the Great Depression or the late 19070s / early 1980s (which would clearly favor bonds).
But even if you don’t like my homespun mechanism, or using gold as the ballast for equities, you still now have a method better than a static glidepath that “saves” the 4* rule for early retirees.
Second sheet with some results for 4.0% and 4.5% IWR using the 10MMA variable equity allocation.
APOLOGIES!
The calculations I posted last night using 10MMA were wrong because of how I was calculating the MMA for the first 9 months. I've corrected this, and am working to update the numbers.
I'll post the correction (along with the updated sheet that does the calculations correctly) hopefully in the next 24 hours.
[The original stuff with my home-brewed variable asset allocation scheme doesn't have this problem.]
@andyg42 Thanks for providing the files.
I have trouble putting together how any of this was calculated. Probably by hand through the "case Study" tab. But there is a much more straightforward way: Simply generate one "momentum-based strategy" return time series, feed that into the "custom return series column" and simply run with that. You'd be able to quickly generate an SWR simulation for all possible starting dates, all in one move.
I can't vouch for or against anything anybody creates by hand the way you did.
I also caution against confounding too many things all at once. If you do the guard rails it makes the comparison of different starting dates extremely difficult. I'd need to see charts like the ones I created in SWR Series Part 10, i.e., what's the distribution of possible withdrawals:
?resize=863%2C646&ssl=1
I believe that can only be done over a loop in Matlab or Python because the withdrawals are path-dependent. Not so easy to do this with only the SWR Google Sheet. Unless you want to run 1700 case studies by hand.
But in any case, I think this is all very important work. I definitely want to a SWR Series post utilizing the tactical asset allocation rules, including momentum.
I've already done a Stock/Bond/Cash model. But I should also do the Stock/Gold model. Hadn't thought of that one!
@earlyretirementnowcom Thanks for the reply.
Yes, I did all of this all "by hand" using a hacked clone of the Case Study tab. I don't know Matlab, and am in no position to do this with Python. I'd be thrilled if you did so!
Re generating "one 'momentum-based strategy' return time series, feed(ing) that into the 'custom return series column' and simply run with that", that too sounds great! But at the moment I don't quite understand how to edit / hack the sheet to do that. I'd be delighted if you did this, or perhaps could give me a bit more explanatory instructions how to go about this, and perhaps I can try it.
Re the confounding too many things at once point: I specifically show all of the a) static, b) guardrails only (with static allocation), c) variable allocation with static consumption, and d) combined guardrails and variable allocation, to address this very issue.
And I report these for the worst cohorts in the separate single page sheet. Per the point above, absent Matlab/Python, I'm not in a position to create the charts.
Attached is the corrected version of the hacked Google sheet using 10MMA (Meb Faber signal) for the variable equity timing signal.
Please discard and do NOT use the previous 10MMA - it incorrectly "knew" to avoid the 1929 crash, so the results were bogus...
Corrected Results page for the 10MMA variant attached.
With the corrected data, the results change quantitatively, but not qualitatively. Using 100% bonds as the ballast for stocks still seems best here overall. Using a 4.0% IWR, FVs all still exceed 99% of initial over 30 years. The November 1965 cohort shows a 3.96% average consumption, while the March 1937 cohort is “only” 3.88%.
If you use a 4.25% IWR, the 1965 cohort exceeds 4% average consumption, while the 1937 one gets up to 3.94%, and FVs go down but remain high enough to safely support a 60 year retirement. So using 10MMA you don’t quite “save” the 4% rule for early retirees, but it seems you can exceed 3.9%. And perhaps someone can do the math and achieve 4.0% or more if your 60 year FV target is below 50%.
And 4% appears quite doable for those of us not concerned about sustained deflation.