Historical backtest of 44 technical signals at the per-stock level: each (signal, stock) pair has its own win rate; only pairs above 70% are kept. Larger samples mean more trustworthy win rates.
Rebuilt 2026-10-21: the three backtest books (members only) were rebuilt on corrected inputs, and the table below with them. Four fixes: only past outcomes that entered on the bar after their trigger count (about 58,000 misplaced old A-share outcomes dropped); ties in the ranking go to more known outcomes, then turnover (they used to fall to the signal name, alphabetically); trades use bars adjusted for splits and dividends with the store's own factors (unadjusted bars booked bonus issues as losses); HK uses the raw price series the signals were computed on. The books still follow the variants chosen on 2026-10-15 — US contrarian worst-MAE + MFE target, CN contrarian lowest win rate + MFE target, HK contrarian turnover + trail 15%. That choice was made after the fact over this same period, and the 2026-10-20 control studies found it indistinguishable from a random pick among the same day's judged signals; variant ranks also do not persist from year to year. It is not a "best" strategy; it is kept so each book has one continuous history. On the corrected inputs: US +277.3% (the 10-15 version: +195.8%), CN −24.3% (was +27.2%), HK −11.7% (was +49.4%). Corrected 2026-10-11: the returns published earlier (e.g. US +419.9%) used future data and are withdrawn. Two problems: (signal, stock) pairs were pre-selected on their win rate over the whole history, which keeps the ones that went on to win; and earlier triggers whose outcome was not yet known were counted in the win rate and the target. The table below has both removed.
Rule: the 10-day win rate of a (signal, stock) as known on the signal date is above 70% (at least 3 earlier triggers that had already run their 10 days; earlier triggers still running do not count), and signal-day turnover is above 100M local. Buy the next open with 25% of equity, at most 4 holdings. Exit: historical best MFE target (10 days), or a trailing stop (10% / 15% / 20%); the stop moves to entry after +10%. Each trade risks 2% of equity. Excludes signal-day limit-ups and IPO first day (CN); BSE uses the 30% limit.
Contrarian: the same rules and period with the gate inverted — the 10-day win rate known on the signal date is below 30% (at least 3 finished triggers), ranked by lowest win rate, worst MAE, or turnover. Trend and contrarian variants (24 per market) share one table; sort by any column.
Period:US 2024-10-01 → 2026-10-21;CN 2024-10-08 → 2026-10-21;HK 2024-10-02 → 2026-10-21
| Market | Strategy | Period | Rank | Exit | Trades | Win rate | Avg / trade | Return | Max drawdown |
|---|---|---|---|---|---|---|---|---|---|
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Worst MAE | MFE target | 156 | 19.6% | +4.14% | +277.3% | -36.9% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | MFE potential | Trail 20% | 75 | 22.5% | +4.34% | +101.8% | -29.4% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Win rate | Trail 15% | 57 | 30.2% | +5.80% | +98.9% | -19.4% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Win rate | MFE target | 98 | 38.9% | +2.46% | +81.9% | -17.1% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Turnover | Trail 20% | 62 | 15.5% | +1.48% | +79.8% | -36.3% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Worst MAE | Trail 20% | 139 | 16.9% | +2.54% | +49.8% | -43.7% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Win rate | Trail 10% | 179 | 34.3% | +1.26% | +48.5% | -39.6% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Turnover | Trail 15% | 26 | 31.8% | +5.19% | +36.2% | -13.5% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | MFE potential | Trail 10% | 191 | 32.1% | +0.62% | +31.5% | -29.9% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Turnover | Trail 15% | 64 | 23.3% | -0.11% | +30.8% | -32.2% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Worst MAE | MFE target | 110 | 17.9% | +1.50% | +30.5% | -31.0% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Turnover | MFE target | 67 | 31.7% | +1.74% | +30.3% | -25.8% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Turnover | Trail 10% | 152 | 34.9% | +0.53% | +27.2% | -30.8% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Lowest win rate | Trail 10% | 151 | 36.1% | +0.65% | +19.3% | -23.1% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Lowest win rate | Trail 20% | 97 | 15.1% | +0.65% | +17.8% | -39.7% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Win rate | Trail 10% | 91 | 32.2% | +0.46% | +16.4% | -23.6% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | MFE potential | Trail 15% | 126 | 27.0% | +0.51% | +14.5% | -38.1% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Turnover | Trail 20% | 16 | 8.3% | -1.19% | +8.2% | -14.1% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Turnover | Trail 20% | 93 | 16.7% | +0.70% | +6.9% | -48.8% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Turnover | Trail 10% | 208 | 34.1% | +0.29% | +4.0% | -33.9% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Win rate | Trail 20% | 50 | 10.9% | +0.43% | +3.3% | -36.1% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | MFE potential | Trail 15% | 123 | 24.4% | +0.67% | +3.0% | -35.1% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | MFE potential | Trail 20% | 154 | 18.5% | +0.06% | +1.1% | -51.4% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Turnover | Trail 10% | 159 | 36.9% | +0.20% | -0.9% | -38.3% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Turnover | Trail 15% | 186 | 23.6% | -0.18% | -7.4% | -36.5% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Worst MAE | Trail 15% | 100 | 29.2% | -0.11% | -7.6% | -35.7% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Worst MAE | Trail 20% | 89 | 15.3% | -0.59% | -9.7% | -28.0% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Turnover | MFE target | 173 | 29.6% | -0.41% | -10.9% | -30.8% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | MFE potential | MFE target | 152 | 10.8% | -0.65% | -11.7% | -52.3% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Turnover | Trail 15% | 179 | 24.0% | +0.11% | -11.7% | -42.2% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Lowest win rate | Trail 20% | 117 | 19.5% | -0.43% | -13.5% | -49.7% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Turnover | Trail 20% | 136 | 16.5% | -0.57% | -14.5% | -51.6% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | MFE potential | Trail 20% | 112 | 14.7% | -0.30% | -14.9% | -41.7% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | MFE potential | MFE target | 92 | 4.5% | -0.38% | -15.5% | -43.5% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Lowest win rate | Trail 20% | 98 | 16.0% | -0.56% | -15.8% | -46.4% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Lowest win rate | Trail 10% | 248 | 31.1% | -0.08% | -15.8% | -47.8% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Turnover | Trail 15% | 133 | 21.5% | -0.17% | -16.1% | -39.5% |
| US | Trend >70% | 2024-10-01 → 2026-10-21 | Turnover | Trail 10% | 67 | 22.2% | -1.59% | -18.7% | -28.7% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Lowest win rate | Trail 15% | 130 | 26.2% | -0.34% | -19.1% | -49.7% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Worst MAE | Trail 15% | 197 | 23.3% | -0.04% | -21.0% | -49.8% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Win rate | Trail 15% | 213 | 25.7% | -0.18% | -22.0% | -47.3% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Lowest win rate | Trail 15% | 167 | 27.0% | -0.44% | -22.6% | -43.6% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Win rate | Trail 20% | 159 | 17.9% | -0.59% | -22.7% | -56.5% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | MFE potential | MFE target | 156 | 9.2% | -0.82% | -24.2% | -55.0% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Lowest win rate | MFE target | 204 | 37.5% | -0.38% | -24.3% | -31.3% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Lowest win rate | Trail 10% | 134 | 32.3% | -0.60% | -25.7% | -39.1% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Turnover | Trail 10% | 248 | 33.5% | -0.40% | -27.8% | -41.1% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Worst MAE | Trail 20% | 177 | 16.2% | -0.41% | -29.3% | -45.2% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Turnover | Trail 20% | 107 | 21.9% | -1.55% | -30.3% | -41.7% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Turnover | Trail 20% | 114 | 16.2% | -1.20% | -32.6% | -45.4% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | MFE potential | Trail 15% | 211 | 24.0% | -0.63% | -32.7% | -49.1% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Win rate | Trail 10% | 269 | 31.6% | -0.38% | -32.8% | -40.4% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Worst MAE | Trail 15% | 182 | 26.8% | -0.84% | -33.6% | -50.0% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Win rate | MFE target | 178 | 10.9% | -0.56% | -33.8% | -54.1% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Turnover | MFE target | 123 | 12.4% | -1.38% | -34.6% | -50.7% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Turnover | MFE target | 168 | 10.3% | -0.90% | -35.4% | -50.9% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Win rate | MFE target | 128 | 14.5% | -1.16% | -35.9% | -56.7% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Worst MAE | Trail 10% | 154 | 34.7% | -1.33% | -37.0% | -44.3% |
| CN | Contrarian <30% | 2024-10-08 → 2026-10-21 | Turnover | MFE target | 130 | 20.5% | -1.84% | -37.0% | -40.6% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | MFE potential | Trail 10% | 218 | 34.6% | -0.78% | -37.1% | -47.7% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Turnover | MFE target | 167 | 20.1% | -0.63% | -37.5% | -47.5% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Lowest win rate | MFE target | 217 | 28.5% | -0.70% | -38.3% | -59.7% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Turnover | Trail 15% | 112 | 20.2% | -1.40% | -39.6% | -51.1% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Worst MAE | Trail 10% | 286 | 29.2% | -0.51% | -42.0% | -52.7% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Lowest win rate | MFE target | 217 | 33.3% | -1.12% | -43.2% | -54.9% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Worst MAE | Trail 10% | 314 | 30.2% | -0.58% | -44.4% | -53.1% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | MFE potential | Trail 10% | 318 | 32.1% | -0.63% | -46.0% | -63.8% |
| HK | Trend >70% | 2024-10-02 → 2026-10-21 | Turnover | Trail 10% | 276 | 30.1% | -0.88% | -48.4% | -54.0% |
| HK | Contrarian <30% | 2024-10-02 → 2026-10-21 | Worst MAE | MFE target | 176 | 16.2% | -1.25% | -52.3% | -66.6% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Win rate | Trail 20% | 103 | 14.1% | -3.40% | -54.2% | -62.1% |
| US | Contrarian <30% | 2024-10-01 → 2026-10-21 | Lowest win rate | Trail 15% | 117 | 19.5% | -2.67% | -55.2% | -59.3% |
| CN | Trend >70% | 2024-10-08 → 2026-10-21 | Win rate | Trail 15% | 120 | 24.1% | -2.81% | -56.9% | -58.9% |
No trading costs. Daily bars adjusted for splits and dividends (A-shares and HK multiplicatively; US is stored adjusted); the turnover gate still uses the money that really traded. Rebuilt on 2026-10-21 from the corrected signal data and inputs. The 24 variants have no consistent winner across the three markets, and their ranks move a lot with small input changes; the control studies (random picks, year-by-year walk-forward selection) show that no variant in this table is evidence that it works.