{"schemaVersion":"1.0","type":"BacktestEvidence","id":"https://formiq.jp/evidence/bollinger-band-sigma-settings/data.json","evidencePage":"https://formiq.jp/evidence/bollinger-band-sigma-settings","language":"en","article":{"title":"Bollinger Bands: 1, 2 or 3 Standard Deviations? Backtest Results","url":"https://formiq.jp/blog/bollinger-band-sigma-settings","published":"2026-08-27","updated":"2026-08-30"},"author":{"name":"Yuichi Saiki","alternateName":"斉木勇一","url":"https://formiq.jp/about/yuichi-saiki"},"verified":"2026-08-30","question":"Which Bollinger deviation should be used, and do wider bands perform better?","finding":"The breakout favoured 1σ, and wider was not stronger. On hourly USDJPY every period from 10 to 50 was profitable in both 2024 and 2025 at 0.5σ through 1.25σ; at 1σ the seven-period mean was +1,568.8 pips in 2024 and +1,081.9 in 2025. At 1.75σ the 2025 mean reached +1,850.2 pips but fell to +590.2 in 2024. Beyond 3σ the samples are too small to read: the hourly breakout averaged 21.7 trades a year at 3σ, 6.3 at 3.5σ and 2.3 at 4σ in 2025. The fade had no stable deviation — of 105 hourly settings only 23 fade-long and one fade-short were profitable in both years. Containment on 6,226 hourly bars at period 20 was 45.0% inside 1σ, 87.1% inside 2σ and 99.1% inside 3σ, against the normal-distribution figures of 68%, 95% and 99.7%.","methodology":"Measured with Formiq's production backtester. The fade enters at the close of a bar whose low or high touched the band and exits at the close of a bar touching the opposite 1σ band; the exit is fixed at 1σ so that only the entry deviation varies. The breakout enters on a close beyond the band and reverses on a close beyond the opposite one.","scope":["USDJPY. 1 Jan to 31 Dec 2025, with 2024 and the two halves of 2025 as comparisons","15-minute, hourly and four-hour bars; 24,903 M15 bars in 2025, 6,226 H1 and 1,610 H4","7 periods (10 to 50) x 15 deviations (0.5σ to 4σ in 0.25 steps) x 3 trading rules = 315 settings, on 3 timeframes and 4 windows: 3,780 runs","0.3 pip spread, 0.1 lots, zero slippage"],"keyResults":[{"label":"Deviations profitable in both years, hourly breakout","value":"every period at 0.5σ to 1.25σ; the 1σ seven-period mean was +1,568.8 / +1,081.9 pips"},{"label":"Hourly breakout trades a year in 2025","value":"21.7 at 3σ, 6.3 at 3.5σ, 2.3 at 4σ"},{"label":"Closes inside the bands, period 20 hourly","value":"45.0% inside 1σ, 87.1% inside 2σ, 99.1% inside 3σ"}],"limitations":["One pair, USDJPY, and two independent calendar years.","Selecting the maximum from a grid of periods and deviations carries multiple-comparison uplift. The confirming year is not untouched future data.","Beyond 3σ the trade counts are small and win rates and profit factors are wide. An average of 2.3 trades a year at 4σ cannot be read with the confidence of 1σ.","Touches use highs and lows; breakouts use closes. Counting a wick piercing as a breakout changes the result.","Fills use bar closes and the spread is fixed throughout.","The fade grid fixes the exit at the opposite 1σ band so that only the entry deviation varies; a test that also optimises the exit answers a different question."]}