Backtest Trade Counts Run From 154 to 10,340: Win Rate Margin Is ±7.7 Points on 154 Trades, ±1.0 on 10,340

We took the trade counts behind the backtest figures of our listed EAs and asked one question: how many trades stand behind each win rate? This article uses backtests only. It contains no forward-account figures. Figures are as of 2026-10-22.

- Across 11 backtests the number of trades runs from 154 (ETHEREUM EA) to 10,340 (GOLD VIPER). Per year of verified history that is 20.5 to 1,342.9 trades.

- 6 of the 11 have fewer than 650 trades. For 7 of the 11 the textbook 95% margin on the win rate is 3.5 points or more.

- The 38.3% win rate of ETHEREUM EA rests on 154 trades and carries a margin of ±7.7 points, so 30.6% to 46.0%; that span contains the win rates of 6 other EAs in the table. The 56.4% of GOLD VIPER rests on 10,340 trades and carries ±1.0.

- MEGAMAX DONCHIAN (34.98%) and DONCHIAN ENGINE (34.5%) differ by 0.48 points, far less than the ±3.7 margin of either.

- More trades does not mean safer: GOLD KING v6 has the highest win rate in the table (83.1%, margin ±1.4) and also the highest maximum drawdown (91.3% on the listed settings).

- Trades, years verified, win rate, profit factor and maximum drawdown are the figures listed on the site for each EA on its listed settings. Trades per year (trades divided by years verified) and the margin are our own arithmetic.

- The margin is the textbook 95% interval for a proportion, 1.96 × √(p × (1 − p) / n), with p the win rate and n the number of trades. It assumes every trade is an independent draw. Trades that belong to one sequence or one basket are not independent, so the real uncertainty is wider than shown, by an amount we cannot compute from this data.

- The maximum drawdown here is the one measured on the listed settings. It is a different kind of figure from the balance and equity drawdowns taken from MT5 reports, so do not put it beside those. Years verified differ between EAs (3.7 to 22.5). TRINITY is included with its listed-settings figures only, because its report details are not in our data.

EA | Trades | Years verified | Trades per year (our division) | Win rate | 95% margin on win rate, points (our arithmetic) | Profit factor | Maximum drawdown (listed settings)

ETHEREUM EA | 154 | 7.5 | 20.5 | 38.3% | ±7.7 | 2.83 | 17.1%

AIFW | 228 | 3.7 | 61.6 | 30.7% | ±6.0 | 1.77 | 4.6%

TRINITY | 420 | 7.5 | 56.0 | 43.1% | ±4.7 | 1.91 | 23.3%

GOLD TRIPWIRE | 616 | 6.8 | 90.6 | 58.6% | ±3.9 | 2.72 | 12.7%

MEGAMAX DONCHIAN | 629 | 9.4 | 66.9 | 34.98% | ±3.7 | 1.55 | 58.6%

DONCHIAN ENGINE | 631 | 9.4 | 67.1 | 34.5% | ±3.7 | 1.5 | 10.8%

BITCOIN COMET | 775 | 8 | 96.9 | 40.1% | ±3.5 | 1.72 | 31.2%

ATLAS | 1,193 | 7.5 | 159.1 | 33.2% | ±2.7 | 1.48 | 26.3%

GOLD NEURON | 2,766 | 22.5 | 122.9 | 53.29% | ±1.9 | 1.35 | 55.5%

GOLD KING v6 | 2,827 | 7.5 | 376.9 | 83.1% | ±1.4 | 4.66 | 91.3%

GOLD VIPER | 10,340 | 7.7 | 1,342.9 | 56.4% | ±1.0 | 1.64 | 30.3%

Sorted by number of trades, smallest first.

With 154 trades a win rate of 38.3% could easily have come out anywhere from 30.6% to 46.0% on a slightly different history. With 10,340 trades the same width of doubt shrinks to about a point. The margin covers the win rate only; the profit factor and the drawdown rest on the same trades but we show no margin for them.

- Before comparing two win rates, look at the trade counts. A gap smaller than the margin (0.48 points against ±3.7) is not a difference you can rely on.

- A small sample is not evidence of a bad EA. It means every figure in its row, profit factor and drawdown included, rests on few trades.

- A big sample says nothing about drawdown. GOLD KING v6 has 2,827 trades and still a maximum drawdown of 91.3% on the listed settings.

- A backtest value is a measurement of the past on one history, not a limit. Nothing here is a promise of future results.