Research, updated November 2026
Out-of-sample results / 2026
Eighteen fortnights, scored against what happened.
We trained the simulator on data to December 2025, then ran it forward on every two-week window of 2026 for books of 8, 16, 24 and 50 US stocks. These are the results, including where it still falls short.
2 January to 25 September 2026
Four book sizes, equal-weight
4 of 5 at 50 stocks: the raw range runs a little narrow
The test
What we ran, and on what.
For each window, the simulator saw only the recent prices of every stock in the book, nothing from after the window’s start. It drew 32 joint 10-trading-day futures for the whole book. We then compared those futures with what actually happened over the next 10 trading days. Windows do not overlap, and no 2026 data was used in training.
How the books were chosen. The 8-, 16- and 24-stock books are hand-picked large US companies across sectors. The 50-stock book is drawn from S&P 500 names. All books are equal-weight. The 24-stock book, our worked example: AAPL, MSFT, NVDA, AMZN, JPM, XOM, PG, JNJ, GOOGL, NFLX, TSLA, META, V, UNH, HD, AVGO, ABBV, COST, MRK, CVX, PEP, KO, CRM, ADBE.
The five checks
All five pass up to 24 stocks. One just misses at 50.
| Check | 8 stocks | 16 stocks | 24 stocks | 50 stocks |
|---|---|---|---|---|
| 1. Do the stocks move together the right amount? (slope, 1.00 = perfect) | 0.90 Pass | 0.92 Pass | 0.92 Pass | 0.92 Pass |
| 2. When one crashes, do others crash too? (share of real joint crashes) | 94% Pass | 87% Pass | 88% Pass | 88% Pass |
| 3. Is the size of each stock’s moves right? (simulated / real) | 0.99 Pass | 0.99 Pass | 0.99 Pass | 0.98 Pass |
| 4. Is the raw range centred, and the right width? (centre shown, 0 = centred) | +0.014 Pass | −0.012 Pass | −0.012 Pass | −0.019 Just misses |
| 5. Is the loss estimate reliable? (breaches of 18; 3 or fewer is green) | 1 Pass | 3 Pass | 2 Pass | 1 Pass |
Check 2 passes at every size since the heavy-tailed coupling (November 2026); we still print the share of real joint crashes it reproduces, because it is still below 100%. Check 4 at 50 stocks: the centre passes, but the raw range runs narrower than its threshold allows, by four thousandths. We report it as a miss. Check 5: a correct 5% estimate is breached about once in 18 windows by chance, and three or fewer breaches is consistent with that.
The range
A range that corrects its own width.
The raw 90% range was too narrow for larger books. So the range corrects its own width from the simulation’s scored track record, using only earlier windows; the earliest windows of the year have no record yet and stay raw. In 2026 the correction settled near x1.4 for the 24-stock book.
| Stocks | Raw range | Self-corrected, all 18 windows | Self-corrected, windows 6–18 |
|---|---|---|---|
| 8 | 86.1% | 89.6% | 89.4% |
| 16 | 83.3% | 87.8% | 90.4% |
| 24 | 82.9% | 88.0% | 91.3% |
| 50 | 81.9% | 88.7% | 94.0% |
How sure can we be? Eighteen windows is a small sample. A hit rate measured on 18 windows carries about ±15 points of uncertainty, so these results support “roughly a 90% range”, not “exactly 90%”. Testing over earlier years is the next step.
The bad fortnight
The loss estimate, against real losses.
| Stocks | Bad-fortnight estimate, average | Estimate, worst window | Worst real fortnight | Breaches of 18 |
|---|---|---|---|---|
| 8 | -3.9% | -6.1% | -3.3% | 1 |
| 16 | -4.1% | -5.8% | -5.9% | 3 |
| 24 | -3.0% | -4.4% | -4.3% | 2 |
| 50 | -3.1% | -5.4% | -4.0% | 1 |
The bad-fortnight estimate is the 10-day loss the book should exceed in one fortnight out of twenty (a 10-day 95% value at risk). Read the raw estimate as a floor: because joint crashes are still slightly under-reproduced, the true tail is probably somewhat deeper.
The remaining gap
Joint crashes are still slightly under-shown.
We count the half-hours in which two stocks both have one of their worst 5% moves. In the 24-stock book that happened at a rate of 0.225 in reality and 0.198 in the simulation: 88% of real. A textbook bell-curve model at the same correlations gives 0.089. Before the November 2026 engine update the simulation reproduced 57% of the real rate; it now passes the check, and still falls a little short of reality. Until that changes, we say so next to every loss number.
The market is alive
Correlations moved more than chance allows.
We split the 24-stock book’s half-hour returns from December 2025 to September 2026 into 7 back-to-back blocks of equal length and measured how much each pair’s correlation changed from one block to the next. On average it moved by 0.099. Shuffling the same returns in time 500 times gives the change that chance alone produces: 0.062. Real correlations moved 1.6 times more, and no shuffle came close.
The clearest case was the March 2026 sell-off. The average correlation across the book went from 0.10 to 0.21 within weeks, then back to 0.08 by May. This is why the simulator re-measures co-movement from recent prices on every run.
What the test does not show
Limits.
No direction. The middle of the range carried no information about which way prices went (its correlation with outcomes ranged from −0.02 to +0.13 across book sizes). The product is the shape of the range, not a forecast.
One year. All results come from 2026. Testing over earlier years has not been done yet.
Large US stocks only. Small caps, ADRs, ETFs and other markets are untested in book form.
Prices only. The model does not know earnings dates or news. Gap-prone names are flagged instead: in the 50-stock book the screen flagged NFLX, ORCL, META, UNH, MCD, CSCO, IBM.
Questions about these results?
Read how the simulation works, or talk to us directly.