Public · read-only · updates with every race
Distributed Validation Board
Every backtest race anyone runs on AIM Volatility Lab adds one independent out-of-sample window per ticker — every race anyone runs makes everyone's validation stricter. The spread across windows is the finding, so this board reports consistency, median and the worst window — never an average. Results only, never who ran them.
55 of 71 validated in their latest window · top 10 shown publiclyWhen the market regime stayed the same, past edges held up in 639 of 1044 community re-checks — after a regime change, 1534 of 1900 held.
| # | Ticker | Edge | Consistency | Range | OOS | Races |
|---|---|---|---|---|---|---|
| 1 | OSCR | validated | 1 window | — | 3.062 | 1 |
| 2 | MUU | validated | 1 window | — | 2.878 | 2 |
| 3 | UNH | validated | 30/30 consistent | 2.069 – 3.061 | 2.66 | 30 |
| 4 | HPE | validated | 1 window | — | 2.393 | 1 |
| 5 | AAVE-USD | validated | 3 windows | 2.274 – 2.699 | 2.362 | 5 |
| 6 | V | validated | 30/30 consistent | 1.519 – 2.838 | 2.346 | 36 |
| 7 | AAPL | validated | 31/31 consistent | 1.823 – 2.593 | 2.195 | 55 |
| 8 | NEAR-USD | validated | 3 windows | 1.976 – 2.667 | 2.015 | 4 |
| 9 | SPY | validated | 30/30 consistent | 1.518 – 2.511 | 1.987 | 61 |
| 10 | DLLL | validated | 1 window | — | 1.979 | 2 |
Aggregated across every account's races — results only, never who ran them. Each window is one day's shared end-of-day bars; each dot = one window's verdict (green = validated, amber = no edge); consistency needs n≥5 windows. Historical simulation for decision support only — hypothetical results, not financial advice, no prediction of future performance.
Beat-holding record board
The closest any engine has come to beating buy & hold in a 3-year duel — the slot fills itself only when the data earns it.
- #1 Zero-Cost Avg on ETH-USD +53.07 pts vs holding · 2024-06-29 → 2026-07-24 · run by AIM Labs
- #2 Zero-Cost Avg on SPY -10.54 pts vs holding · 2025-07-08 → 2026-07-08 · run by a member
- #3 Classic AIM on AAPL -20.52 pts vs holding · 2021-01-04 → 2023-12-29 · run by a member
Think your ticker beats it? Run the duel yourself →
How this stays honest
Every ticker races the same fixed configuration on shared end-of-day bars, so each day is one true out-of-sample fold — nine strangers' shifted windows are a robustness test no individual would bother running. Verdicts are gated at n≥5 windows, the same way our regime-probability engine is publicly scored before anyone sees a probability. See the calibration scoreboard →
Hypothetical simulations on historical data — not investment advice. aimvolatilitylab.com