Indian F&O data, and the evidence about what not to trade · educational
🛡 Gate Scorecard: we test our own risk flags in public
Every pick on this site carries a Darwin Gate verdict. This page holds the Gate accountable, two ways: a full reconstruction over every historical pick since 2020 (point-in-time correct, split-adjusted), and a live forward test that accrues nightly from verdicts recorded BEFORE outcomes are knowable.
Reconstruction: all 4,463 historical picks, vetted as-of their pick date
| Group | Picks | Avg 5d | Volatility (sd) | Up rate | Worst | Avg 21d |
|---|
| PASS | 3643 | +0.27% | 4.0 | 51% | -26.3% | +1.20% |
| WARN | 733 | +0.73% | 5.3 | 55% | -45.6% | +2.72% |
| VETO | 87 | +1.03% | 5.3 | 60% | -20.1% | +1.87% |
By individual flag
| Flag | Picks | Avg 5d | Volatility (sd) | Up rate | Worst | Avg 21d |
|---|
| earnings window | 374 | +0.12% | 5.2 | 49% | -45.6% | +0.63% |
| up-gap on pick day | 96 | +1.64% | 7.3 | 66% | -45.6% | +4.38% |
| in F&O ban | 66 | +1.15% | 5.6 | 64% | -20.1% | +1.98% |
| high-VIX regime | 311 | +1.37% | 5.5 | 60% | -20.1% | +5.27% |
| compressed market dip | 311 | +1.37% | 5.5 | 60% | -20.1% | +5.27% |
The honest reading (this is not the result we expected, and we publish it anyway):
1. Flags measure risk, and the data confirms it: flagged picks carry 1.3-1.8x the volatility of Clear picks, and the deepest single-pick losses.
2. Within this dip-buying population, that risk was often PAID: high-VIX and gap-day picks averaged better bounces than Clear picks, with wider swings both ways. Caution flags here mean "bigger distribution, size smaller", NOT "lower average".
3. The earnings flag is the exception: lower average than Clear picks (+0.27% vs +0.12%) AND fatter tails: unknowable direction with big stakes.
4. The ban flag stands on tradability, not returns: banned names are reduce-only in F&O (you often cannot open the position at all) and carry the widest single-name outcomes.
This page updates nightly and the interpretation will change if the data does.
Live forward test
Started 2026-07-28. live table below
Method: historical picks reconstructed with the exact public screen; each vetted using only information available on its pick date (that day's actual NSE ban file, the earnings calendar as of that date, trailing price/volume/delivery windows, expanding-window market-state percentiles). Returns are close-to-close, split/bonus adjusted. No costs deducted (this measures information content, not a trading system). Backtested reconstruction and live accrual are reported separately and labelled.
Educational data reference: NOT investment advice. Historical statistics do not predict future results. We are not SEBI-registered advisers. All data from NSE public sources; figures may contain errors. Trading, especially leveraged F&O, can cause significant losses.