Works but not worth it
Turn-of-the-month: invested 12% of the time, earned 4% a year
The idea
Stocks have historically earned more around the turn of the month. Lakonishok and Smidt, “Are Seasonal Anomalies Real? A Ninety-Year Perspective” (Review of Financial Studies, 1988), used the last trading day of a month through the first three of the next. The rule: take the 200-day trend overlay, but only be invested in those four days.
What happened
| Sharpe | Yearly return | Worst drop | Invested | |
|---|---|---|---|---|
| 2017–2021, turn-of-month rule | 0.75 | 3.8% | 6.7% | 12% |
| 2017–2021, plain 200-day rule | 0.97 | 9.0% | 13.4% | 65% |
| 2022–2023, turn-of-month rule | −0.32 | 2.9% | 4.6% | 7% |
| 2022–2023, plain 200-day rule | −0.05 | 3.3% | 6.1% | 39% |
The close-to-close premium was real but modest: 0.098% on turn-of-month days against 0.068% on other days in 2017–2021, and 0.046% against 0.021% in 2022–2023.
The official runs
These are the official default-cost results for the pre-registered turn-of-month rule, one of the 15 trials in the literature batch. 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, which slightly favours it. The 2005–2016 window was used to form the 200-day average; this rule was not an official pass/fail run there, so it is not in the table.
| 2017–2021 | Sharpe | Yearly return | Worst drop | Time invested |
|---|---|---|---|---|
| Turn-of-month rule Pre-registered · Failed its own rule | 0.75 | 3.82% | 6.7% | 12% |
| Plain 200-day overlay | 0.97 | 8.97% | 13.4% | 65% |
| SPY, same invested days Reference | 1.29 | 7.17% | 4.3% | 12% |
| SPY, fully invested Reference | 0.92 | 18.28% | 33.7% | 100% |
| 2022–2023 | Sharpe | Yearly return | Worst drop | Time invested |
|---|---|---|---|---|
| Turn-of-month rule Pre-registered · Failed its own rule | −0.32 | 2.87% | 4.6% | 7.0% |
| Plain 200-day overlay | −0.05 | 3.28% | 6.1% | 39% |
| SPY, same invested days Reference | −0.33 | 2.74% | 2.7% | 7.0% |
| SPY, fully invested Reference | 0.12 | 4.26% | 22.1% | 100% |
The rule traded about 15 times its capital a year in 2017–2021 and 9 times in 2022–2023, which cost it about 0.23% and 0.12% a year. Each window is one official run at default costs.
The premium itself is in the daily numbers. This is SPY close to close, on the four turn-of-month days against every other day:
| Average daily return, SPY close to close | Turn-of-month days | All other days |
|---|---|---|
| 2005–2016 (build window, context only) | 0.040% (n=575) | 0.035% (n=2,445) |
| 2017–2021 | 0.098% (n=236) | 0.068% (n=1,003) |
| 2022–2023 | 0.046% (n=92) | 0.021% (n=389) |
The gap is positive in all three windows and small in each. The executable book also earns it only on four days a month, which is why it sat in cash 88–93% of the time.
Why it failed
A small premium cannot carry a book that sits in cash 88–93% of the time against a strategy that is invested through most of the market's drift. In 2017–2021 it earned 3.8% a year, with a Sharpe of 0.75, against 9.0% and 0.97 for the plain 200-day rule. The rule also trades about 15 times its capital a year.
Receipts: the rule was pre-registered before the run, as one of 15 official trials in that batch.