Up 0.14% for the week. On December 24th, the strategy rebalanced to:
52% SPY , 32% TLT , 16% GLD
*All investing strategies come with the risk of loss, including this one. This portfolio may not be appropriate for your investment goals and requirements, and it is not investment advice.
It should not be assumed that recommendations made in the future will be profitable or will equal the performance of the securities in this list.
Calculations are my own. Returns shown do not include trading costs. They do not include any fees. Past performance is not indicative of future performance. Dividends are re-invested.
Hi BTM,
I stilll love your ideas, but am also still sceptical that the method actually works – my theory is that it’s easy to trick oneself into thinking it works without a suitable benchmark, two of which, as I’ve already suggested, should be the Permanent Portfolio and, more suitably, an equal split of risk assets (e.g. 1/3 Bonds, 1/3 Stocks, 1/3 Gold).
Running a backtest with these and realistic assumptions, namely:
1) Rebalance only once per year.
2) S&P Yield included in S&P Returns
3) High Trade costs of 0.3% for Bonds, 0.8% for Stocks and 1% for Gold
We get the following:
https://i.imgur.com/XmRgMTH.png
Annualised Return:
S&P = 20.56%
PP = 15.75%
1/3 Split = 20.85%
Annualised Vol:
S&P = 28.57%
PP = 7.6%
1/3 Split = 10.12%
The main thing is that a simple 1/3 split seems to beat both the S&P AND Geometric over this time period.
Along long time frames, the 1/3 begins to lose out, but my suspicion remains that almost all the benefit of the Geometric method comes from simply having a split of uncorrelated risk assets and the additional complexity of changing the % allocations gives the illusion of control, but simply adds to the trading costs.
PLEASE put a post comparing Geometric to a fixed (occasionally rebalanced) 1/3 risk asset split if you truly want to prove that Geometric has an edge.
In all the testing I have done, recent correlation and volatility do nothing to predict asset allocation %.
Yes, this period it has preformed similarly to the permanent portfolio. There are also periods where the permanent portfolio will do better. But more often then not, the PP doesn’t keep up.
Here are the stats of a 1/3 split rebalanced annual SP500, 30Yr Treasury, Gold portfolio from 1978 to 2019.
Equal Weight Gold, SP500, 30yr. Bonds,
1978-2019, Rebalanced Annually
Average Ret. 9.9% Neg. Years 7
Geo Return 9.4% Best Year 54.4%
Max DrawDown -25.9% Worst Year -12.6%
Ann StDev 9.% Sharpe 0.57
I’ll do a post expanding on this comparison, but you can compare to the table in this post https://breakingthemarket.com/my-current-portfolio/
In summary, the difference is nearly 2% more return /year, 10% lower volatility, and 60% of the drawdown.
Thanks. Your returns don’t account for trading fees though, and weekly rebalancing is going to eat into that return.
Yes, but in there are low cost options for trading etfs (in the USA at least). I actually tried to estimate trading costs for a very frequent rebalancing strategy using interactive brokers costs before Robinhood and all the others opened up a no commissions option and included it an early post in a table.
https://breakingthemarket.com/geometric-balancing-unlevered/
Another interesting portfolio is a 75/25 Stocks / Gold, rebalanced annually but ONLY partially rebalanced.
I came across the idea of “asymptotic” rebalancing from belangp –
Rebalanced only 20 – 70% towards the target weight provides returns that equal Geometric over the same period (around 11.2 %), albeit with slightly higher vol (around 14.5%).
Fully rebalanced hurts returns, but only slightly – 10.9%, but also lowers vol (very slightly) to 14%.
That is interesting. Thanks.
I’ll leave it here because the next post strangely doesn’t have comment field.
It’s interesting that in 2020 geometric rebalancing underperformed every asset it consists of – TLT, SPY and GLD (at least in my version of it which follows yours pretty closely on average). This haven’t happened in any previous year in backtest (I tested from 2005). I attributed it to unusually uniform yearly return for all 3 assets and to unusual volatile gains that happened quite often during the year.
Do you have thoughts on that or do you have a data for longer backtest? I’m curious if that’s the first time ever. IMO it would be a warning sign.
That’s an interesting observation. Geo Balancing also includes cash, which it held at an average of about 11% weighting this year which is part of what you see. Sometimes it holds cash to increase the return (there was some of this in March), and sometimes it just holds it to control potentially drawdowns (November). The second version could easily hurt returns and it probably did this year.
So I think your understanding of why is correct. The returns are very similar, and the volatility threw the fund into a larger cash position than usual.
Historically, it’s pretty rare for all three to move so in sync, although recently it’s been more normal. If you use the weekly rebalancing, GB lost to all three just barely in 2005. The opposite, Geo Balancing beating all three, very very nearly happened in 2018 an 1995.