The Lab
Every trading day at 9:25 ET the Lab forecasts where SPY will be at 10:00, noon, 2:00 and the close, then learns from what actually happened. The bar it has to clear is the simplest honest forecast: the latest pre-market price holds.
Skill at the close +1.3% ± 1.1% · promising, not proven
Edge hit rate 62% direction beyond the latest price, 58 leaned days
Range coverage 70% target 68%
Days learned 124 first 15 are warmup and not counted
Call
Wed, Oct 7, 2026 Graded All inputs inClose target $775.43, range $769.79 to $781.11, 29% chance of closing above the prior close ($779.09). Closed at $777.22.
| Checkpoint | P(up) | Target | Range | Actual | In range | vs naive |
|---|---|---|---|---|---|---|
| 10:00 | 26% | $776.88 | $774.27 – $779.50 | $776.39 | Yes | Beat (0.201 vs 0.237) |
| Noon | 25% | $775.59 | $770.95 – $780.27 | $775.67 | Yes | Beat (0.223 vs 0.236) |
| 2:00 | 27% | $775.65 | $770.88 – $780.44 | $777.26 | Yes | Beat (0.232 vs 0.248) |
| Close | 29% | $775.43 | $769.79 – $781.11 | $777.22 | Yes | Beat (0.237 vs 0.244) |
pushes downpushes up
- IWM minus QQQ yesterday
- Implied vs realized vol
- High yield (HYG) yesterday
- Long bonds (TLT) yesterday
- Gamma above vs below spot
- Net dealer gamma
- Baseline lean
- VIX futures (VIXY) yesterday
Track record
-10.0%-5.0%+0.0%+5.0%+10.0%
naive+1.3%
May 22Jul 15Sep 1Oct 21
40%60%80%100%
68% target70%
May 22Jul 15Sep 1Oct 21
- Kalman: options positioning 15%
- Kalman: everything, slow memory 14%
- Analog days (10 nearest) 10%
- Gaps keep going 10%
- Yesterday's move reverses 9%
- Everyone else 42%
0%50%100%
Apr 27Jun 25Aug 24Oct 21
0%50%100%
20
16
15
19
39
0%25%50%75%100%
Leaderboard
Average CRPS at the close in sigma units (lower is better), and each expert's edge over "latest price holds". Weight is how much the ensemble listens to it today.
| Expert | CRPS | vs naive | Weight | Verdict |
|---|---|---|---|---|
| Kalman: everything, slow memory | 0.4789 | +0.0071 ± 0.0100 | 13.9% | ~765 more days to confirm |
| Kalman: options positioning | 0.4792 | +0.0069 ± 0.0078 | 15.0% | ~460 more days to confirm |
| Gaps keep going | 0.4822 | +0.0038 ± 0.0078 | 9.6% | ~1726 more days to confirm |
| Yesterday's move reverses | 0.4848 | +0.0012 ± 0.0045 | 9.4% | ~5929 more days to confirm |
| Latest price holds (random walk) | 0.4860 | — | 9.0% | the benchmark |
| Analog days (10 nearest) | 0.4870 | -0.0009 ± 0.0083 | 9.6% | no edge so far |
| Kalman: price action | 0.4877 | -0.0017 ± 0.0097 | 6.9% | no edge so far |
| Kalman: Jev text reads | 0.4911 | -0.0050 ± 0.0035 | 6.1% | no edge so far |
| Dealer hedging drift | 0.4918 | -0.0058 ± 0.0048 | 6.4% | no edge so far |
| Yesterday's direction continues | 0.4941 | -0.0081 ± 0.0047 | 6.2% | no edge so far |
| Kalman: everything, fast memory | 0.4986 | -0.0126 ± 0.0153 | 3.1% | no edge so far |
| Gaps fade | 0.5051 | -0.0190 ± 0.0081 | 4.7% | no edge so far |
By checkpoint
| Checkpoint | Skill | From range | From edge | Coverage | Edge hit rate |
|---|---|---|---|---|---|
| 10:00 | +8.9% ± 1.7% | +2.4% | +6.6% | 74% | 76% (66) |
| 12:00 | +1.6% ± 1.0% | +0.7% | +0.9% | 70% | 67% (36) |
| 14:00 | +1.9% ± 0.9% | +1.6% | +0.3% | 71% | 53% (30) |
| Close | +1.3% ± 1.1% | +0.6% | +0.7% | 70% | 62% (58) |
How it learns
- Senses. Options positioning from the 9:25 snapshot, yesterday's moves in stocks, bonds, credit, the dollar and VIX, and Jev's reads of the morning headlines and WSB.
- Experts. Simple fixed hypotheses ("gaps fade"), Kalman filters whose weights drift slowly with the market, and an analog expert that asks what happened on the most similar past days.
- Ensemble. Experts earn weight by forecasting better (Hedge), and every expert keeps a small share so one that fell out of favor can come back when the market changes.
- Range. The range widens after a close lands outside it and narrows after one lands inside, so over time about 68% of closes fall inside it.