Our test results
No method has reached +0.1R a trade yet. Some are close to break-even. Many failed. We keep the failures in this table, because they teach the most.
| Method | 2025 (picking) | 2026 (unseen) | Decision |
|---|---|---|---|
| FVG + 4-hour trend | −0.01R | +0.06R | Live paper bot |
| London breakout | +0.01R | +0.04R | Live paper bot |
| FVG + daily trend | +0.07R | +0.02R | Live paper bot |
| AI team (172 trades) | −0.11R over the whole test, Feb 2025 to Oct 2026 | Not split by year | Live paper bot, watching |
| FVG on 4-hour candles | +0.09R | −0.12R | Dropped: it only fit the past |
| FVG on 15-minute candles | Worse | Worse | Dropped |
| Liquidity sweep + CHoCH/BOS + FVG | Looked OK | Negative | Dropped |
| Session filter, "best of 288 settings" | Looked good | Failed | Dropped |
| Hammer, best of 192 versions | +0.03R | −0.09R | Dropped |
| Hammer, average of 192 versions | −0.10R | −0.17R | Dropped |
| Indices (S&P, NASDAQ, US30, GER40) | Mixed | Mixed, weak | Not added |
Lesson
Famous patterns, like the hammer and the liquidity sweep, mostly fail when tested honestly. A pattern can be popular and still not make money.
Gold
Gold is the most consistent symbol in our tests. In the FVG + 4-hour trend test, gold alone made +0.18R in 2026. This is one part of the test, not a whole method, so it does not prove an edge. We are watching it.
Quick check