Seasonal Labs
11 surviving pairs apply to 2026 now: in years when it was a midterm election year, Oct 1 → Mar 31 (the midterm-low window) was 17-2 against 52-24 in all years (+21 pts); -0.6% so far.
▶ View the live, interactive version with full charts →
Nobody picks the table. A few hundred conditions (the year's YTD bucket at each month-end, each month's sign, a 12-month high in a given month, distance from the 52-week high, the prior year's return, three-month streaks, the election cycle, the year digit) are crossed with every half-month, month, two-month span, quarter and classic named window on the S&P 500 since 1950, scored only where the window starts after the condition is known, and kept only if they survive four controls: 12+ years, the dates moved ±3 sessions, an early-half / late-half replication, and a 2,000-draw placebo of random years. The survivors are one table; the popular stats that fail are in the graveyard with the control that killed them, and the false-positive arithmetic is printed next to both.
- 5,607 condition × window pairs were tested on the S&P since 1950; 65 survive the nudge, split-half and placebo controls, against roughly 140 that chance alone would push through the placebo gate.
- 11 surviving pairs apply to 2026 now: in years when it was a midterm election year, Oct 1 → Mar 31 (the midterm-low window) was 17-2 against 52-24 in all years (+21 pts); -0.6% so far.
- The popular tables live in the graveyard: the calendar supplies most of any seasonal win rate, and the exact dates supply the rest.
When to run this study
Read it whenever a seasonal stat is doing the rounds and you want to know whether it is one of the few that hold up or one of the many that do not. The first table is the short list of condition-and-window pairs that survive every control; the graveyard is where the high win rates go to die, each with the control that killed it. Check "Live now / coming up" for the survivors that apply to this year.
Historical results
Edges that survive: 65 of 5,607 pairs tested
| In years when… | the S&P from… | Years | Record | Win % | All years | Base % | Edge (pts) | Avg edge (pts) | Median % | Nudged low % | Early / late half | Placebo pct | Chance odds | Clears 10% FDR | 2026 | Circulating |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| the year was up 5-10% at the end of January | Jun 1 → Jul 31 (Jun–Jul) | 12 | 12-0 | 100 | 47-29 | 61.80 | 38.20 | 4.42 | 5.40 | 83.30 | 100% / 100% | 99.9 | 1 in 594 | no | 2026 does not qualify | |
| the first 100 sessions were up 8% or more | Jun 1 → Jun 30 (June) | 23 | 18-5 | 78.30 | 36-40 | 47.40 | 30.90 | 1.65 | 2.11 | 69.60 | 82% / 75% | 99.9 | 1 in 2,497 | no | 2026 qualifies (+9.9% d100) · window done -1.3% | |
| the year was up 5-10% at the end of January | Jul 1 → Jul 31 (July) | 12 | 11-1 | 91.70 | 47-29 | 61.80 | 29.80 | 3.18 | 4.30 | 83.30 | 83% / 100% | 98.2 | 1 in 56 | no | 2026 does not qualify | |
| the year was up 5-10% at the end of February | Jun 1 → Jun 30 (June) | 17 | 13-4 | 76.50 | 36-40 | 47.40 | 29.10 | 1.49 | 2.11 | 64.70 | 75% / 78% | 99.0 | 1 in 151 | no | 2026 does not qualify | |
| it was a pre-election year | Apr 1 → Apr 30 (April) | 19 | 18-1 | 94.70 | 50-26 | 65.80 | 28.90 | 1.95 | 3.20 | 78.90 | 100% / 90% | 99.9 | 1 in 765 | no | 2026 does not qualify | |
| the S&P had three straight up months through April | Nov 1 → Nov 15 (first half of November) | 19 | 16-3 | 84.20 | 42-34 | 55.30 | 28.90 | 0.79 | 1.50 | 63.20 | 78% / 90% | 99.5 | 1 in 332 | no | 2026 does not qualify | |
| the year was up 5-10% at the end of February | Jun 1 → Jul 31 (Jun–Jul) | 17 | 15-2 | 88.20 | 47-29 | 61.80 | 26.40 | 2.20 | 2.76 | 64.70 | 88% / 89% | 99.4 | 1 in 109 | no | 2026 does not qualify | |
| it was a midterm election year | Nov 1 → Apr 30 (the six good months) | 19 | 19-0 | 100 | 56-20 | 73.70 | 26.30 | 7.47 | 15.13 | 94.70 | 100% / 100% | 99.9 | 1 in 901 | no | 2026 qualifies · starts Nov 1 (6 days) | yes |
| the year was up 10% or more at the end of May | Jun 1 → Jun 30 (June) | 19 | 14-5 | 73.70 | 36-40 | 47.40 | 26.30 | 1.28 | 2.11 | 68.40 | 78% / 70% | 98.9 | 1 in 124 | no | 2026 qualifies (+10.7% YTD May) · window done -1.3% | |
| the S&P had three straight up months through July | Sep 16 → Sep 30 (second half of September) | 18 | 12-6 | 66.70 | 31-45 | 40.80 | 25.90 | 0.93 | 0.45 | 44.40 | 78% / 56% | 98.7 | 1 in 87 | no | 2026 does not qualify | |
| the year was up 5-10% at the end of February | Nov 16 → Nov 30 (second half of November) | 17 | 16-1 | 94.10 | 52-24 | 68.40 | 25.70 | 0.90 | 1.74 | 76.50 | 88% / 100% | 99.3 | 1 in 141 | no | 2026 does not qualify | |
| the S&P was 3-10% below its 52-week high at the end of May | Jun 1 → Jun 15 (first half of June) | 28 | 23-5 | 82.10 | 43-33 | 56.60 | 25.60 | 1.11 | 1.63 | 64.30 | 79% / 86% | 100.0 | 1 in 1,885 | no | 2026 does not qualify | |
| it was a pre-election year | Mar 1 → Apr 30 (Mar–Apr) | 19 | 18-1 | 94.70 | 53-23 | 69.70 | 25 | 2.90 | 5.14 | 84.20 | 100% / 90% | 99.5 | 1 in 238 | no | 2026 does not qualify | |
| the S&P had three straight up months through April | Sep 1 → Oct 31 (Sep–Oct) | 19 | 16-3 | 84.20 | 45-31 | 59.20 | 25 | 3.07 | 3.26 | 78.90 | 67% / 100% | 99.0 | 1 in 110 | no | 2026 does not qualify | |
| the S&P had three straight up months through April | Nov 1 → Dec 31 (Nov–Dec) | 19 | 18-1 | 94.70 | 53-23 | 69.70 | 25 | 2.14 | 4.78 | 89.50 | 89% / 100% | 99.7 | 1 in 238 | no | 2026 does not qualify |
Showing 15 of 65 rows — see the live version for all.