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 +1.8% ± 1.1% · promising, not proven
Edge hit rate 64% direction beyond the latest price, 61 leaned days
Range coverage 71% target 68%
Days learned 131 first 15 are warmup and not counted
Call
Wed, May 6, 2026 Graded 1 input missingClose target $727.74, range $725.23 to $730.26, 93% chance of closing above the prior close ($723.77). Closed at $733.83.
This call was made without everything it normally sees. Each one was read as an average day (no signal), so the call leaned on the rest:
- WSB euphoria (Jev) missing
| Checkpoint | P(up) | Target | Range | Actual | In range | vs naive |
|---|---|---|---|---|---|---|
| 10:00 | 100% | $727.73 | $726.45 – $729.00 | $729.07 | No | Lost (0.507 vs 0.504) |
| Noon | 98% | $727.74 | $725.86 – $729.62 | $731.53 | No | Lost (1.291 vs 1.276) |
| 2:00 | 95% | $727.74 | $725.41 – $730.08 | $731.77 | No | Lost (1.109 vs 1.093) |
| Close | 93% | $727.74 | $725.23 – $730.26 | $733.83 | No | Lost (1.604 vs 1.580) |
pushes downpushes up
- Baseline lean
Track record
-10.0%-5.0%+0.0%+5.0%+10.0%
naive+1.8%
May 22Jul 17Sep 9Oct 30
40%60%80%100%
68% target75%
May 22Jul 17Sep 9Oct 30
- Kalman: everything, slow memory 16%
- Gaps keep going 13%
- Kalman: options positioning 11%
- Analog days (10 nearest) 10%
- Kalman: price action 8%
- Everyone else 42%
0%50%100%
Apr 27Jun 29Aug 31Oct 30
0%50%100%
24
16
15
19
42
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 | 0.4674 | +0.0094 ± 0.0100 | 16.1% | ~409 more days to confirm |
| Gaps keep going | 0.4692 | +0.0076 ± 0.0078 | 13.0% | ~366 more days to confirm |
| Kalman: options positioning | 0.4717 | +0.0051 ± 0.0077 | 10.9% | ~936 more days to confirm |
| Kalman: price action | 0.4754 | +0.0014 ± 0.0097 | 8.5% | ~23378 more days to confirm |
| Analog days (10 nearest) | 0.4761 | +0.0007 ± 0.0079 | 9.9% | ~64411 more days to confirm |
| Yesterday's move reverses | 0.4763 | +0.0005 ± 0.0043 | 7.5% | ~31394 more days to confirm |
| Latest price holds (random walk) | 0.4768 | — | 7.8% | the benchmark |
| Kalman: Jev text reads | 0.4777 | -0.0009 ± 0.0041 | 8.3% | no edge so far |
| Dealer hedging drift | 0.4823 | -0.0055 ± 0.0046 | 5.6% | no edge so far |
| Yesterday's direction continues | 0.4842 | -0.0074 ± 0.0045 | 5.6% | no edge so far |
| Kalman: everything, fast memory | 0.4849 | -0.0081 ± 0.0150 | 4.6% | no edge so far |
| Gaps fade | 0.5007 | -0.0239 ± 0.0083 | 2.2% | no edge so far |
By checkpoint
| Checkpoint | Skill | From range | From edge | Coverage | Edge hit rate |
|---|---|---|---|---|---|
| 10:00 | +10.6% ± 1.8% | +1.9% | +8.6% | 73% | 76% (72) |
| 12:00 | +1.4% ± 1.0% | +0.7% | +0.8% | 69% | 66% (38) |
| 14:00 | +2.2% ± 0.9% | +2.0% | +0.2% | 72% | 53% (32) |
| Close | +1.8% ± 1.1% | +0.8% | +1.0% | 71% | 64% (61) |
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.