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 +1.4% ± 1.1% · promising, not proven
Edge hit rate 63% direction beyond the latest price, 59 leaned days
Range coverage 70% target 68%
Days learned 126 first 15 are warmup and not counted
Call
Wed, Oct 14, 2026 Graded All inputs inClose target $774.76, range $770.28 to $779.27, 29% chance of closing above the prior close ($778.01). Closed at $769.60.
| Checkpoint | P(up) | Target | Range | Actual | In range | vs naive |
|---|---|---|---|---|---|---|
| 10:00 | 32% | $776.60 | $774.49 – $778.70 | $775.83 | Yes | Beat (0.205 vs 0.241) |
| Noon | 28% | $775.35 | $771.67 – $779.04 | $771.73 | Yes | Beat (0.426 vs 0.427) |
| 2:00 | 28% | $775.20 | $771.43 – $778.98 | $769.43 | No | Lost (0.595 vs 0.587) |
| Close | 29% | $774.76 | $770.28 – $779.27 | $769.60 | No | Beat (0.447 vs 0.495) |
pushes downpushes up
- Long bonds (TLT) yesterday
- Implied vs realized vol
- Gamma above vs below spot
- Baseline lean
- Net dealer gamma
- Distance to gamma flip
- Put skew
- High yield (HYG) yesterday
Track record
-10.0%-5.0%+0.0%+5.0%+10.0%
naive+1.4%
May 22Jul 15Sep 3Oct 23
40%60%80%100%
68% target70%
May 22Jul 15Sep 3Oct 23
- Kalman: options positioning 14%
- Kalman: everything, slow memory 13%
- Gaps keep going 11%
- Analog days (10 nearest) 10%
- Yesterday's move reverses 10%
- Everyone else 43%
0%50%100%
Apr 27Jun 26Aug 25Oct 23
0%50%100%
22
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 | 0.4756 | +0.0063 ± 0.0099 | 13.4% | ~995 more days to confirm |
| Kalman: options positioning | 0.4763 | +0.0056 ± 0.0078 | 13.7% | ~746 more days to confirm |
| Gaps keep going | 0.4773 | +0.0046 ± 0.0077 | 10.9% | ~1151 more days to confirm |
| Yesterday's move reverses | 0.4808 | +0.0011 ± 0.0044 | 9.6% | ~7155 more days to confirm |
| Latest price holds (random walk) | 0.4819 | — | 9.4% | the benchmark |
| Analog days (10 nearest) | 0.4829 | -0.0011 ± 0.0082 | 9.8% | no edge so far |
| Kalman: price action | 0.4845 | -0.0027 ± 0.0096 | 6.4% | no edge so far |
| Kalman: Jev text reads | 0.4860 | -0.0041 ± 0.0035 | 7.0% | no edge so far |
| Dealer hedging drift | 0.4876 | -0.0058 ± 0.0048 | 6.6% | no edge so far |
| Yesterday's direction continues | 0.4900 | -0.0081 ± 0.0046 | 6.4% | no edge so far |
| Kalman: everything, fast memory | 0.4959 | -0.0140 ± 0.0151 | 2.8% | no edge so far |
| Gaps fade | 0.5025 | -0.0207 ± 0.0081 | 4.0% | no edge so far |
By checkpoint
| Checkpoint | Skill | From range | From edge | Coverage | Edge hit rate |
|---|---|---|---|---|---|
| 10:00 | +10.0% ± 1.9% | +2.0% | +8.0% | 74% | 76% (68) |
| 12:00 | +1.6% ± 1.0% | +0.7% | +0.9% | 69% | 65% (37) |
| 14:00 | +1.9% ± 0.9% | +1.7% | +0.2% | 71% | 52% (31) |
| Close | +1.4% ± 1.1% | +0.7% | +0.7% | 70% | 63% (59) |
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.