Works but not worth it

The plain 200-day rule: smaller drops, not enough to pass

· 4 min read

The idea

After the 32-variant search was ruled out, we tried the simplest thing nobody had tuned: give each of 23 ETFs an equal slot, hold the slot while the fund's close is above its 200-day average, check monthly, otherwise sit in T-bills.

What we did

The rule, both windows (2017–2021 and 2022–2023), and the pass criteria were committed before the run, and it was run once. Pass rule per window: worst drop at most 60% of an equal-weight basket of the same 23 ETFs, and Sharpe not more than 0.10 below that basket. Both windows had to pass.

What happened

Sharpe Yearly return Worst drop
2017–2021
200-day rule 0.97 8.97% 13.4%
Equal-weight basket 0.85 9.04% 17.7%
SPY buy and hold 0.92 18.28% 33.7%
2022–2023
200-day rule −0.05 3.28% 6.1%
Equal-weight basket −0.07 2.74% 12.4%
SPY buy and hold 0.12 4.26% 22.1%
  • 2017–2021: fail. Worst drop was 0.76 of the basket's; the bar was 0.60.
  • 2022–2023: pass. Worst drop was 0.49 of the basket's. The year stocks and bonds fell together.

The official runs

These are the official default-cost results for the pre-registered 200-day overlay and its equal-weight control, both run once per window. Sharpe and yearly return credit idle cash at the 3-month T-bill rate, like the figures already on this page. Worst drop is from the simulated book as traded. The SPY rows are reference points, not trials; the same-days row is frictionless.

2017–2021SharpeYearly returnWorst dropTime invested
200-day rule Pre-registered0.978.97%13.4%65%
Equal-weight basket of the same 23 funds Control0.859.04%17.7%68%
SPY, same invested days Reference0.9711.18%20.3%65%
SPY, fully invested Reference0.9218.28%33.7%100%
2022–2023SharpeYearly returnWorst dropTime invested
200-day rule Pre-registered−0.053.28%6.1%39%
Equal-weight basket of the same 23 funds Control−0.072.74%12.4%60%
SPY, same invested days Reference0.003.52%6.9%39%
SPY, fully invested Reference0.124.26%22.1%100%
Growth of one dollar for the 200-day rule, the equal-weight basket of the same 23 funds, SPY on the same days and fully invested SPY, in 2017–2021 and 2022–2023.
Growth of $1. The rule credits idle cash at the 3-month T-bill rate, the same way as the yearly returns above. The SPY mixes are frictionless. Official windows only; the January 2022 embargo between them is not shown.
Drawdown from peak for the 200-day rule, the equal-weight basket of the same 23 funds, SPY on the same days and fully invested SPY, in 2017–2021 and 2022–2023.
Drawdown from peak. The rule uses the simulated book as traded. Official windows only.

The rule's worst drop was 0.76 of the basket's in 2017–2021 (bar 0.60: fail) and 0.49 in 2022–2023 (pass). Each of the two was also run at zero costs and at double the usual spread, which makes up the 6 overlay and 6 control trials. Costs hardly move either:

Sharpe (cash at T-bill rate)Zero costsDefault costsDouble spread
200-day rule, 2017–20210.970.970.96
Equal-weight basket, 2017–20210.850.850.85
200-day rule, 2022–2023−0.05−0.05−0.06
Equal-weight basket, 2022–2023−0.06−0.07−0.07

Why it isn't worth it

It does cut the damage, especially in 2022-style markets. The registered miss is the 2017–2021 drop target: the worst drop was 0.76 of the equal-weight basket's, and our bar was 0.60. We did not adjust the pass rule after seeing this.

The rule did not give up a large return relative to the same 23 funds. In 2017–2021 it earned 8.97% a year against the basket's 9.04% (−0.07 percentage points), and in 2022–2023 it earned 3.28% against the basket's 2.74% (+0.54). Both the rule and the unfiltered basket lagged SPY in 2017–2021 (18.28% a year; about 18.3% in the prose on this site). That gap is the cost of holding this basket, not something the 200-day rule itself caused. In 2022–2023 its Sharpe was still below SPY's (−0.05 against 0.12).

It is the base for the drawdown brake, which passed its own test against this rule but did not beat holding SPY on the same days.

Receipts: the rules were written down and committed before any run. 6 overlay and 6 control trials.