No Calendar Month Is Best for All 8 EAs: Best Month Made 13.6%-31.3% of Profit; a Net-Loss May Won in 6 of 8 Years

We took the published MT5 backtests of 8 of our EAs, added up the net profit of all Januaries, all Februaries and so on, and asked whether the best and the worst calendar month are the same ones from EA to EA. This article uses backtests only. It contains no forward-account figures. Figures are as of 2026-10-19.

- No calendar month is the best for all 8. The best month is March for 3 (GOLD KING, DONCHIAN ENGINE, GOLD VIPER), November for 2 (BITCOIN COMET, GOLD NEURON), and May, October and February for one each. The worst month is spread over 6 different months, August being the weakest for 3.

- The best calendar month produced 13.6% (GOLD KING) to 31.3% (GOLD VIPER) of the whole-test net profit. An even split over 12 months would be 8.3% each.

- Across the 8 EAs and 12 calendar months (96 cells), 15 cells have a net loss in total. June, July and November are positive for all 8; December is positive for only 5.

- A calendar month with a total loss is not a month that lost every year. GOLD VIPER's May total is -141,138.46, yet May was profitable in 6 of 8 years. One month, 2026-05 (-192,160.63), outweighs the rest: the other seven Mays together made +51,022.17.

- The best calendar month was not reliable either: it was profitable in 3 of 6 years for ETHEREUM TREND and 13 of 22 for GOLD NEURON.

- Each closed trade is assigned to the calendar month of its close time in the Strategy Tester. Commission and swap are included. Open positions at month end are not valued. All years of a test are then added up by calendar month.

- Money amounts are in the test account currency. Every test starts from a 10,000 deposit. Several tests compound, so late years carry much larger amounts than early ones, and a calendar month's total mostly reflects the years in which the balance was large. This is why we also show in how many years the month was profitable, which does not depend on the amount.

- "Years profitable" counts only the years in which that calendar month had at least one trade; a test that starts or ends mid-year has fewer observations for some months.

- The drawdown in the last column is the balance drawdown from the MT5 report for the whole test, measured on closed-trade balance only. An equity drawdown, which includes open trades, is larger: for the same 8 tests it is between 10.33% and 42.8%. We do not put the two kinds in one column.

- TRINITY and AIFW are not in this article: the original report files behind their published backtests are not in our breakdown, so we could not split them by month.

EA | Test period and data | Best calendar month (total net profit) | Share of whole-test net profit | Years that month was profitable | Worst calendar month (total net profit) | Years that month was profitable | Calendar months with a net loss in total (of 12) | Balance drawdown (MT5 report, whole test)

GOLD VIPER | M30, 2019.01.01-2026.09.01, 8% real ticks | March (612,953.45) | 31.3% | 7 / 8 | May (-141,138.46) | 6 / 8 | 3 | 19.6%

GOLD NEURON | M30, 2004.01.01-2026.09.01, 100% real ticks | November (115,851.13) | 27.8% | 13 / 22 | April (-53,295.25) | 10 / 23 | 3 | 22.84%

ETHEREUM TREND | H4, 2019.01.01-2026.06.28, 100% real ticks | February (7,789.69) | 24.1% | 3 / 6 | December (-2,913.81) | 1 / 6 | 3 | 6.7%

BITCOIN COMET | H4, 2018.01.01-2025.12.31, 98% | November (269,019.09) | 22.1% | 4 / 8 | September (-78,431.68) | 2 / 8 | 3 | 14.06%

MEGAMAX DONCHIAN (USDJPY) | H1, 2017.01.01-2026.06.01, 100% | May (39,557.91) | 18.4% | 7 / 10 | January (-2,844.90) | 2 / 10 | 2 | 15.19%

DONCHIAN ENGINE (free) | H1, 2017.01.01-2026.06.01, 100% real ticks | March (3,002.71) | 18.2% | 6 / 10 | August (6.55) | 2 / 9 | 0 | 7.31%

ATLAS PORTFOLIO | H4, 2019.01.01-2026.06.28, 100% | October (70,335.43) | 14.2% | 4 / 7 | August (-5,899.83) | 2 / 7 | 1 | 16.29%

GOLD KING | M1, 2019.01.01-2026.06.01, 5% real ticks | March (83,898.83) | 13.6% | 8 / 8 | August (24,351.47) | 7 / 7 | 0 | 7.45%

The tests cover different periods and different data quality, so compare the numbers inside one row, not the rows with each other. The profit totals, the years and the whole-test balance drawdown come from the same test. MEGAMAX DONCHIAN and DONCHIAN ENGINE have almost the same trade count (629 and 631) over the same period, so we do not treat their months as independent evidence. For DONCHIAN ENGINE the "worst" month is August at +6.55, not a loss. For GOLD KING no calendar month lost in total or in any year.

GOLD VIPER's May is the clearest case. The sum of its 8 Mays is -141,138.46, the lowest of any calendar month, but 6 of the 8 Mays ended above zero. The worst single month of the whole test is 2026-05 at -192,160.63 on 143 trades; remove it and the other Mays add up to +51,022.17. A total can be negative because of one month. So a "bad May" is a statement about one test, not a rule for the next May.

The reverse also holds. GOLD NEURON's best calendar month, November, made 115,851.13, which is 27.8% of the whole-test profit, but it was profitable in 13 of 22 years (59.1%). A large total can sit on a thin majority of years.

Only weakly. August is the lowest month by total for GOLD KING, ATLAS PORTFOLIO and DONCHIAN ENGINE, but it is the second-best month for GOLD VIPER (412,385.82), so it is not a common weak month. March is the best month for 3 EAs, and the worst for none. With 6 to 22 years per EA and totals dominated by a few large months, we would not trade on any of these patterns. We show them so that you can see how little of the result a calendar rule would have explained.

- Do not read a month-of-year table as a schedule. In these tests the best and the worst calendar month differ from EA to EA, and a total says little about how often that month was positive.

- Check the number of years behind each month. 3 of 6 and 13 of 22 are not the same strength of evidence as 8 of 8.

- A single month can decide a calendar-month total. Ask what the total is without its largest month.

- When you compare drawdowns, compare the same kind over the same period.

- These are historical simulations. They describe how a past test spread across calendar months, not how future months will go, and no figure here is a promise of future results.

The two EAs discussed most in this article: GOLD VIPER https://www.mql5.com/en/market/product/182297 · GOLD NEURON https://www.mql5.com/en/market/product/187329