Up 0.9% for the week. The strategy is back within a quarter of a percent from the daily high set on the 21st of February. It set a new monthly high today, up nearly 5% for the year.
Why Is This Working?
Since the strategy is now above it’s prior level on the day the S&P 500 peaked (Feb 19th), I’m going to illustrate why it has worked so well over this time.
The market has taken a gigantic curve this year. Curves do not help returns. The market fell 34% in February and March (x 0.66). It has rebounded 36% (x 1.36) off the bottom. This equates to 0.66 x 1.36 = 0.899 or a 10.1% loss over that time.
Of course this is exactly where the S&P 500 rests since Feb 19th.
But what I want you to notice is the arithmetic return is implied to be positive over this time.
(36% + -34%) / 2 = 1%.
Of course the market didn’t experience the arithmetic return because it never does. It experiences the geometric return. It drove through the curves.
Over this time, the daily returns are about as flat as they should be (0.999 average). However they have a standard deviation of 3.7%, which is quite high. This leads to:
Arithmetic Return – 1/2 Volatility2 = Geometric Return
0.999 – 1/2 * 0.0372 = .99832
Geometric Return^ #days = Realize Return
0.9983270 = 0.889
So after these 70 days the approximation of the volatility drag shows the market should be down 11.1%. Pretty close to reality (the realized data is slightly skewed causing the difference).
How Does Geometric Balancing Approach This?
There is a good reason my first investment post was about cutting risky games with cash to convert them from a losing game to positive games. Understanding the first post is foundational to understanding Geometric Balancing.
If you held a 50% cash, 50% S&P 500 portfolio through this time (rebalanced daily), you would be down 4% from Feb 19th.
Not half of 10% = 5% as most would assume.
The benefits of cash are not linear, they are convex (the same is true for leverage).
Geometric Balancing aims to eliminate the losses that come from high volatility as much as possible. It does this first by mixing the stock market with two assets — Treasury Bonds and Gold — that are usually uncorrelated to stock returns. This reduces volatility without greatly impacting the arithmetic returns.
Secondly it scales the portfolio with cash (or leverage) to profit off the convexity.
Through these two methods, the strategy keeps the portfolio aligned to grow over the long term. Over a full cycle, in this case, 35% up 35% down, it harvests out the volatility to preserve actual returns. The 35% down, 35% up, will behave more like arithmetic returns, aka, it’s essentially flat since the top.
It’s not perfectly flat since you can’t remove volatility entirely. The daily volatility over this time for the strategy has been 0.77%, or nearly 20% of the market’s volatility.1
Twenty percent of market volatility means experiencing only 4% of the market’s volatility drag (20%2 = 4%). In times of high volatility, this leads to significant outperformance.
This is the benefit of taking a straighter road through the market.
Weekly Portfolio
To me the market this week felt similar to last week. The strategy agrees and isn’t changing anything, although if you’re paying attention to the scale (leverage), it’s a bit more comfortable now. On May 29th, 2020, the strategy remained at:
40% SPY , 40% TLT , 20% GLD
1-The strategy is not this excellent all the time. So don’t expect the same ratio in future drawdowns.
Aren’t you just time speculating? Your thesis tries to time the market so you reduce volatility but holding cash in a bull market will dramatically reduce returns. This only benefits you during down markets. Markets are up about 75% of the time. You will underperform the market (s&p 500) over a decade.
Everyone is time speculating, most don’t realize it though.
https://twitter.com/breakingthemark/status/1260770275794333697?s=20
I believe I time speculate less.
The benefit on the way up comes from the philosophy’s tendency to be more aggressive in calm markets. Bull markets usually come with lower volatility than bear markets. Cash is not common in the portfolio during bull markets.
I noticed the strategy keeps high allocation to TLT even in a raising market (even in 2019/beginning of 2020).
How do you think lower expected returns of bonds are going to affect strategy returns long term? TLT can go up even more short term but over the course of the next several years it is unlikely to return more than the yield which is very low. Wouldn’t it return pretty much the same as cash but with higher volatility?
There is also a risk of raising rates (at some point in the future) which will affect TLT negatively.
Not terribly concerned about the lower rates as long as the negative correlation to stocks continues. Most of the benefit has come from the correlation lately not the yield. Rising rates are problematic and will impact returns, but in that case I’d hope (expect) either gold or stocks or both to counteract those losses.
What kind of tax drag and turnover does the strategy have?
Turnover’s usually well over 100% per year. Taxes under FIFO aren’t good. I feel taxes under LIFO are reasonable as you just run the top part of the portfolio through short term gains and the bottom part only get touched every few years during big changes (like in March). Of course in tax advantaged accounts this doesn’t matter.
How are you shrinking / constraining the model? I run my own version of this, and it appears you are generally running more diversified than my calculations would suggest. Are you solving under some max volatility or historic drawdown constraint? I’ve calculated “Shadow price” values of expected returns that would match your allocations and they often have one component at nearly exactly zero – suggests you have a separate constraint of some sort that limits the extreme max geometric mean answer. E.g. I would be at something like 70/30/0 today where you are 40/40/20. Happy to compare notes directly if you want to email back to address on this
CHI_CAPITALIST.