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 +1.5% ± 1.1% · promising, not proven
Edge hit rate 63% direction beyond the latest price, 62 leaned days
Range coverage 70% target 68%
Days learned 133 first 15 are warmup and not counted
Call
Tue, Oct 20, 2026 Graded All inputs inClose target $766.24, range $762.71 to $769.79, 48% chance of closing above the prior close ($766.48). Closed at $769.05.
| Checkpoint | P(up) | Target | Range | Actual | In range | vs naive |
|---|---|---|---|---|---|---|
| 10:00 | 37% | $765.89 | $764.73 – $767.04 | $766.66 | Yes | Beat (0.225 vs 0.273) |
| Noon | 40% | $765.74 | $763.43 – $768.06 | $769.56 | No | Lost (0.747 vs 0.689) |
| 2:00 | 46% | $766.14 | $763.42 – $768.87 | $769.62 | No | Beat (0.537 vs 0.571) |
| Close | 48% | $766.24 | $762.71 – $769.79 | $769.05 | Yes | Beat (0.380 vs 0.426) |
pushes downpushes up
- Long bonds (TLT) yesterday
- IWM minus QQQ yesterday
- Fed leaning hawkish (Jev)
- Distance to gamma flip
- Gamma above vs below spot
- VIX futures (VIXY) yesterday
- Net dealer gamma
- Baseline lean
Track record
-10.0%-5.0%+0.0%+5.0%+10.0%
naive+1.5%
May 22Jul 20Sep 10Nov 3
40%60%80%100%
68% target75%
May 22Jul 20Sep 10Nov 3
- Kalman: everything, slow memory 15%
- Gaps keep going 12%
- Analog days (10 nearest) 11%
- Kalman: options positioning 10%
- Kalman: price action 9%
- Everyone else 43%
0%50%100%
Apr 27Jun 30Sep 1Nov 3
0%50%100%
24
16
16
19
43
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 | 0.4714 | +0.0081 ± 0.0099 | 15.5% | ~579 more days to confirm |
| Gaps keep going | 0.4735 | +0.0060 ± 0.0078 | 12.0% | ~666 more days to confirm |
| Kalman: options positioning | 0.4763 | +0.0033 ± 0.0077 | 9.8% | ~2402 more days to confirm |
| Kalman: price action | 0.4785 | +0.0011 ± 0.0096 | 9.1% | ~37304 more days to confirm |
| Analog days (10 nearest) | 0.4790 | +0.0006 ± 0.0078 | 10.7% | ~93179 more days to confirm |
| Latest price holds (random walk) | 0.4796 | — | 8.6% | the benchmark |
| Yesterday's move reverses | 0.4810 | -0.0014 ± 0.0047 | 6.6% | no edge so far |
| Kalman: Jev text reads | 0.4831 | -0.0035 ± 0.0048 | 6.8% | no edge so far |
| Yesterday's direction continues | 0.4850 | -0.0055 ± 0.0049 | 7.7% | no edge so far |
| Dealer hedging drift | 0.4854 | -0.0058 ± 0.0045 | 5.9% | no edge so far |
| Kalman: everything, fast memory | 0.4893 | -0.0098 ± 0.0149 | 4.2% | no edge so far |
| Gaps fade | 0.5017 | -0.0221 ± 0.0083 | 3.0% | no edge so far |
By checkpoint
| Checkpoint | Skill | From range | From edge | Coverage | Edge hit rate |
|---|---|---|---|---|---|
| 10:00 | +10.6% ± 1.8% | +2.1% | +8.5% | 73% | 76% (72) |
| 12:00 | +1.6% ± 1.0% | +0.7% | +0.9% | 69% | 68% (40) |
| 14:00 | +2.3% ± 0.9% | +2.1% | +0.2% | 71% | 53% (34) |
| Close | +1.5% ± 1.1% | +0.7% | +0.8% | 70% | 63% (62) |
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.