Prior-Move Bucket Seasonal
Bucketing ^GSPC years by Aug as a whole; 2026 sits in 0 to 3%. Sep baseline 45% positive.
▶ View the live, interactive version with full charts →
What the month just gone says about the month ahead, bucketed by the SIZE of that prior move rather than its sign. The general form of the "August was strong so September will be weak" family: pick a month, measure it either in full or by its first N trading days, and every year in the sample is bucketed and shown its next-month, 3-month and rest-of-year returns — each against the all-years baseline, so a 60% bucket can be read next to the ~60% that any month clears unconditionally. Also reports the typical drawdown INSIDE the next month, because most of these claims are really about a dip, not a close.
- 76 signals since 1950-01-01.
- Bucketing ^GSPC years by Aug as a whole; 2026 sits in 0 to 3%. Sep baseline 45% positive.
When to run this study
Run at a month-end, or a few days into a month, when someone quotes a stat of the form "month X did Y, so month X+1 will do Z". Buckets every historical year by the size of the same prior move and shows the next month, next quarter and rest-of-year returns for each bucket against the all-years baseline. Set lookback_mode=first_n to test the "first N trading days" variants.
Historical results
What followed, bucketed by the all of Aug
| All of aug | N | Sep | Sep win | Worst point in Sep | 3M | 3M win | Rest of yr | RoY win |
|---|---|---|---|---|---|---|---|---|
| < -3% | 18 | -1.67 | 44.40 | -3.58 | 3.52 | 72.20 | 3.41 | 66.70 |
| -3 to 0% | 16 | 0.28 | 50 | -1.56 | 2.56 | 81.20 | 4.83 | 93.80 |
| 0 to 3% | 23 | 1.02 | 56.50 | -1.19 | 4.97 | 78.30 | 5.38 | 82.60 |
| >= 3% | 19 | -0.70 | 26.30 | -2.29 | -0.90 | 42.10 | 2.10 | 63.20 |
| ALL YEARS (baseline) | 76 | -0.30 | 44.70 | -1.69 | 2.93 | 68.40 | 3.97 | 76.30 |