DIARY
A diary of testing well-known trading ideas on historical data and keeping the results as they came out. This time: 26 plans, A to Z.
Tested: October 31, 2026
16 plans were positive before costs, and 4 stayed positive after the spread. Two passed the criteria set in advance, and none survived the recheck. Simply adding up all 26 plans gives +73.9% before costs and −645.3% after. The spread had a far bigger effect than the differences between methods.
Full-period result of each plan. The length of each line is the spread paid.
Before costsAfter costs
| Plan | Method | Trades | Win rate | Before costs | After costs | 1st half | 2nd half | Random avg | Verdict |
|---|---|---|---|---|---|---|---|---|---|
| A | Golden cross | 11,537 | 34.7% | −11.0% | −29.4% | −10.1% | −19.3% | −14.3% | No win |
| B | RSI mean reversion | 8,209 | 62.6% | −1.3% | −13.6% | −5.8% | −7.8% | −14.4% | No win |
| C | Bollinger mean reversion | 19,101 | 62.3% | −7.1% | −36.3% | −23.2% | −13.1% | −31.8% | No win |
| D | Bollinger breakout | 19,020 | 34.7% | +5.1% | −23.1% | −15.1% | −8.0% | −28.2% | No win |
| E | MACD cross | 39,932 | 34.0% | +19.7% | −41.7% | −34.2% | −7.5% | −63.1% | No win |
| F | Turtle (daily) | 855 | 34.2% | −6.2% | −7.4% | −1.7% | −5.7% | +8.1% | No win |
| G | Ichimoku three signals | 14,582 | 38.0% | +2.7% | −19.8% | −15.7% | −4.1% | −18.5% | No win |
| H | Stochastic | 25,982 | 49.2% | −6.0% | −46.8% | −30.8% | −16.0% | −44.4% | No win |
| I | SMA200 deviation | 5,340 | 69.3% | −5.1% | −13.2% | −4.0% | −9.2% | −11.6% | No win |
| J | Engulfing candle | 35,442 | 34.0% | −0.3% | −54.9% | −36.6% | −18.3% | −55.1% | No win |
| K | Pin bar | 4,513 | 33.2% | +1.1% | −5.9% | −5.9% | −0.0% | −6.7% | No win |
| L | Three soldiers (daily) | 2,422 | 49.9% | −5.6% | −9.1% | −4.1% | −4.9% | +0.2% | No win |
| M | Fade 5 in a row | 14,008 | 48.0% | +9.2% | −12.8% | −7.0% | −5.8% | −23.1% | No win |
| N | London breakout | 17,736 | 43.3% | +8.9% | −17.4% | −11.5% | −5.8% | −27.6% | No win |
| O | NY opening range breakout | 3,667 | 44.3% | +30.9% | +7.9% | +3.1% | +4.8% | −22.1% | Failed recheck (on hold) |
| P | Previous-day breakout | 22,567 | 40.0% | +1.3% | −35.6% | −31.4% | −4.2% | −34.0% | No win |
| Q | False-break fade | 26,673 | 50.4% | +9.5% | −32.1% | −19.3% | −12.8% | −42.9% | No win |
| R | Round-number bounce | 41,352 | 47.7% | −26.8% | −208.8% | −119.2% | −89.6% | −183.8% | No win |
| S | Weekend gap fill | 1,719 | 59.3% | +12.5% | +4.3% | +0.3% | +4.0% | −3.2% | Failed recheck |
| T | Tokyo fix (every day) | 1,253 | 49.9% | +18.5% | +1.3% | +3.2% | −1.9% | −20.0% | No win |
| U | Tokyo fix (gotobi only) | 281 | 46.3% | +2.1% | −1.6% | −0.6% | −1.1% | −4.4% | No win |
| V | Buy the day after a crash | 638 | 50.8% | −0.7% | −1.7% | −0.6% | −1.1% | +4.2% | No win |
| W | Turn of the month | 342 | 52.6% | +9.5% | +7.8% | +6.5% | +1.3% | +6.8% | No win |
| X | Higher timeframe + pullback | 5,310 | 34.2% | +2.8% | −5.0% | −2.8% | −2.2% | −7.1% | No win |
| Y | Random entry + trailing stop | 14,639 | 36.2% | +6.1% | −15.9% | −9.0% | −6.9% | −20.2% | No win |
| Z | Everything vote | 25,117 | 25.4% | +4.0% | −34.4% | −18.2% | −16.2% | −36.1% | No win |
Before costsAfter costs
Lines: cumulative weekly result (%). Each chart has its own vertical scale. Left of the dotted line is the first half; right is the second.
Rules: Enter on the SMA20/SMA50 cross and reverse on the opposite cross. H1.
Result: Already negative before costs. Few symbols ended positive.
Rules: Buy below RSI(14) 30, sell above 70, exit back at 50 (max 48 bars).
Result: A high win rate and a negative total. The result was almost the same as random entries.
Rules: Fade a close outside ±2σ; exit at the middle band (max 48 bars).
Result: A high win rate, yet worse than random.
Rules: Follow a close outside ±2σ; exit at the middle band (max 120 bars).
Result: Doing the opposite of plan C doesn't work either. The small pre-cost gain went to the spread.
Rules: Reverse on the MACD(12,26,9) cross. H1.
Result: Positive before costs and better than random, but it traded so often that it could not cover the spread.
Rules: Enter on a 20-day breakout, exit on a 10-day opposite extreme, 2 ATR stop.
Result: Negative even without swap despite long holds, and behind random entries of the same direction and length.
Rules: Buy on the bullish three-signal setup, sell on the bearish one, exit on the Tenkan/Kijun cross.
Result: No difference from random.
Rules: Enter on Stochastic(14,3,3) crosses below 20 or above 80. TP and SL both 1.5 ATR.
Result: Close to a coin flip, so the spread simply added up with each trade.
Rules: Fade a 3 ATR deviation from SMA200; exit back at SMA200 (max 120 bars).
Result: The highest win rate of all the plans, but the total matched random: a loss.
Rules: Enter on an engulfing candle. Stop beyond the candle, take profit at 2R.
Result: About zero before costs. Not a single symbol ended positive.
Rules: Fade a 20-bar extreme on a wick at least 3× the body. Take profit at 2R.
Result: About zero before costs, and no difference from random.
Rules: Buy three white soldiers, sell three black crows on daily bars. Hold for 5 days.
Result: Negative before costs and worse than random.
Rules: After 5 bars in one direction, fade it. Hold for 6 bars.
Result: Better than random, but still negative after costs.
Rules: Follow a break of the pre-8:00 London range between 8:00 and 16:00. Stop = range width; exit at 16:00.
Result: Better than random, but still negative after costs.
Rules: Three US indices. Follow a break of the 9:30–10:00 NY range. Stop = range width; exit at the close. M5.
Result: One of two plans that passed the criteria. It then failed the recheck, which required both halves to be positive at double spread. Break-even is about 1.3× spread, which leaves little room, and 2022 stands out in the yearly results by symbol. The pre-cost gain itself was not disproved, so this one stays “on hold”.
| Condition | After costs | 1st half | 2nd half |
|---|---|---|---|
| Original | +7.9% | +3.1% | +4.8% |
| Double spread | −15.1% | −14.8% | −0.3% |
| Execution one bar later | +9.2% | +2.3% | +6.9% |
| One bar later + double spread | −13.7% | −15.5% | +1.8% |
| 15-minute range | +9.0% | −1.0% | +10.1% |
| 60-minute range | −0.4% | −6.8% | +6.4% |
Rules: Follow a break of the previous day's high or low. 2 ATR stop; exit at the day's end.
Result: No difference from random.
Rules: Fade a bar that breaks the previous day's extreme and closes back inside. TP and SL both 1.5 ATR.
Result: Better than random, but still negative after costs.
Rules: Six FX pairs. Trade the bounce off round numbers (0.5 for JPY pairs, 0.005 for others). TP and SL both 2 ATR. M5.
Result: Last of all 26. Trading often on a short timeframe made the spread count in full.
Rules: Trade toward filling a weekend gap of at least 0.5 ATR. Max 24 bars.
Result: The other plan that passed, and it was an illusion. H1 bars understate the spread just after the weekend open. With the actual M5 spread at that moment, the result turned negative before any stress test.
| Condition | After costs | 1st half | 2nd half |
|---|---|---|---|
| Actual M5 spread | −1.1% | −1.6% | +0.6% |
| 5 minutes later | −1.6% | −2.0% | +0.4% |
| 15 minutes later | −0.7% | −2.3% | +1.6% |
| Double spread | −14.6% | −7.9% | −6.7% |
| 5 minutes later + double spread | −14.2% | −7.9% | −6.3% |
Rules: USDJPY. Buy at 9:20 Tokyo time, exit at 9:55. Every weekday.
Result: The pre-cost gain clearly differs from random, so a time-of-day bias seems to exist. After costs it is near zero and the second half is negative: the spread takes almost all of it.
Rules: Plan T restricted to gotobi days (multiples of 5 and month-end).
Result: Gotobi days are said to work best, but restricting to them made it worse.
Rules: Buy after a daily close at −2σ or below; hold for 5 days.
Result: Worse than buying at random.
Rules: Six stock indices. Buy the day before month-end and exit on the third trading day of the next month.
Result: Positive in every symbol and both halves, so it looks like the best plan. But buying at random with the same count and length made about the same. It simply held longs while indices rose; no turn-of-month effect was confirmed. Swap is also excluded.
Rules: Trade with the daily SMA50 direction, entering on H1 RSI 30/70. SL 1.5 ATR, TP 3 ATR.
Result: A smaller loss, but no clear difference from random.
Rules: Enter by coin flip once a day and exit only with a 3 ATR trailing stop.
Result: “Exits alone can win with random entries” found no support under these conditions.
Rules: Enter when all five votes agree (SMA, MACD, RSI, close vs middle band, Kijun). Exit when the votes split. 2 ATR stop.
Result: Stacking indicators only lowered the win rate.
From 371 plan-and-symbol combinations with at least 30 trades, we took the top 10 by first-half result and looked at their second half. The first-half average of +36.7% became −9.3%, and 4 of 10 stayed positive. Across all combinations, 52 were positive in both halves, but with this many tries chance alone produces some. “This method works on this symbol” is easy to say in hindsight. Figures are % with all capital on that one symbol.
| Plan | Symbol | 1st half | 2nd half |
|---|---|---|---|
| F | USDJPY | +59.4% | −3.4% |
| G | GBPJPY | +43.9% | −35.3% |
| D | USDJPY | +40.0% | +16.1% |
| I | EURUSD | +34.0% | −12.5% |
| L | BTCUSD | +32.7% | +2.1% |
| A | EURJPY | +32.3% | −41.2% |
| I | AUDUSD | +32.2% | +9.8% |
| B | GBPUSD | +32.2% | −9.3% |
| Y | BTCUSD | +30.2% | +19.5% |
| A | GBPJPY | +29.9% | −39.0% |
Check your own ideas, yourself.
MT5 Analyzer lets you test your ideas on historical data.
See MT5 Analyzer →This diary records tests on historical data. It does not recommend any method or symbol and does not indicate future profit or loss. Please make investment decisions at your own risk.
