All posts
PublishedUpdatedBybollinger bandsstandard deviationsigmabreakoutmean reversionbacktestUSDJPY

Bollinger Bands: 1, 2 or 3 Standard Deviations? Backtest Results

We tested Bollinger Bands from 0.5 to 4 standard deviations on USD/JPY. The hourly breakout favored 1σ; wider fade settings failed to repeat across two years.

The Bollinger Band dialog has a box called “Deviation” or “Standard Deviations.” Two is the familiar default, but whether 1σ, 2σ or 3σ is best depends first on what you do at the band.

We tested USD/JPY from 0.5σ through 4σ in 0.25 steps, crossed with seven periods from 10 to 50. A lower-band fade, an upper-band fade and a close-based breakout were run on 15-minute, hourly and four-hour bars across 2024, 2025, and the two halves of 2025: 3,780 backtests in all.

The short answer is that 1σ is a defensible starting point for an hourly breakout. From 0.5σ through 1.25σ, all seven periods were profitable in both 2024 and 2025. At 1σ, the seven-period mean was +1,568.8 pips in 2024 and +1,081.9 in 2025. At 1.75σ it was +1,850.2 in 2025 but only +590.2 in 2024.

That does not make 1σ universally optimal. It means that, among these candidates, 1σ made the best compromise between profit, survival across periods and transfer to another year. The fade did not produce a repeatable “best sigma.”

0246770.570.757171.2551.551.756232.2522.512.753323.2543.543.7544deviation (σ)periods positive both years
0.5 to 1.25σ1.5 to 4σseven periods; USD/JPY hourly breakout
All seven periods stayed positive in both years from 0.5σ through 1.25σ. The apparent recovery beyond 3.5σ rests on only 6.3, 4.1 and 2.3 trades a year on average.

What Bollinger Bands sigma changes

Let SMA(n) be the simple moving average of the last n closes, SD(n) their standard deviation and k the deviation setting. The bands are:

Upper = SMA(n) + k × SD(n)

Lower = SMA(n) − k × SD(n)

The 2 in the settings dialog does not mean 2%. It multiplies the rolling standard deviation by two. Raising sigma pushes both bands away from the mean, producing fewer touches and fewer close-based breaks. Raising period also tends to reduce frequency because the centre line reacts more slowly.

SettingWhere it sitsRole in this hourly test
Close to the meanBreakout candidate with many signals
The conventional widthMany breakout settings survived both years
Far from the meanFew trades, highly dependent on rare moves
Reached very rarelyOnly a handful of trades per year

The often-quoted 68%, 95% and 99.7% figures for 1σ, 2σ and 3σ describe a normal distribution of independent observations. A market has serial dependence, trends and jumps, so those percentages are not market signal frequencies.

0%25%50%75%100%0.50.7511.251.51.7522.252.52.7533.253.53.75445%87.1%99.1%deviation (σ)
closes finishing insideperiod 20; 6,226 USD/JPY H1 bars in 2025
Measured on price rather than a theoretical distribution, 45.0% of closes sat inside 1σ, 87.1% inside 2σ and 99.1% inside 3σ.

On 6,226 hourly USD/JPY bars in 2025, using period 20, 45.0% of closes finished inside 1σ, 87.1% inside 2σ and 99.1% inside 3σ. That leaves 12.9% outside 2σ. Merely reaching 2σ does not establish that price is abnormal or due to reverse.

The settings dialog

For the conventional 20-period, 2σ display in MT4 or MT5, the inputs are below. TradingView and Formiq expose the corresponding period and deviation fields.

FieldExampleWhat it controls
Period20Lookback for the moving average and standard deviation
Shift0Horizontal displacement; normally zero
Deviations2.000Multiplier applied to standard deviation
Apply toClosePrice used in the calculation; this test uses closes
Style / ColorAnyDisplay only; it does not change the calculation

Formiq's backtester adds a trigger choice.

TriggerRule
Band touchBuy when the low reaches the lower band, or sell when the high reaches the upper band
Band breakoutBuy on a close above the upper band, or sell on a close below the lower band

The same 20 and 2σ can give opposite results when a touch is faded and a close outside is followed. The trading direction has to be fixed before sigma can be compared.

How this was measured

ItemValue
PairUSD/JPY
Windows1 Jan to 31 Dec 2025; 2024 and the two halves of 2025 as comparisons
Timeframes15-minute / one-hour / four-hour
2025 bars24,903 M15, 6,226 H1, 1,610 H4
Periods10, 14, 20, 25, 30, 40, 50
Deviations0.5σ to 4σ in 0.25 steps, 15 values
Fade longBuy the close of a bar whose low touched the lower band; exit at the close of a bar touching the opposite 1σ band
Fade shortSell the close of a bar whose high touched the upper band; exit at the close of a bar touching the opposite 1σ band
BreakoutBuy a close above the upper band, sell a close below the lower band, reverse on the opposite close
Grid7 periods × 15 deviations × 3 trading rules = 315 settings, on 3 timeframes and 4 windows: 3,780 runs
Costs0.3-pip spread, zero slippage, 0.1 lots
FillsSignal-bar close; opposite-side reversals may occur on the exit bar, while same-side re-entry waits for a later bar
Stops and targetsNone in the main grid; added separately below
MeasurementFormiq's production backtester, with pips re-derived from every fill

The fade's opposite band stays fixed at 1σ so the experiment changes the entry sigma only. Moving the entry and exit together would not reveal which one caused the result.

Which sigma works best?

On the hourly chart the 0.5σ to 1.25σ band was the steadiest. Each row below averages its seven periods. “Positive both” counts the periods that made money in both 2024 and 2025.

SigmaMean pips 2025Mean pips 2024Mean 2025 tradesPositive both
0.5σ+1,069.1+1,430.3363.47 / 7
0.75σ+1,113.0+1,217.8301.17 / 7
+1,081.9+1,568.8260.47 / 7
1.25σ+999.7+1,088.6222.47 / 7
1.5σ+1,368.1+693.0188.75 / 7
1.75σ+1,850.2+590.2151.75 / 7
+979.6+742.2129.06 / 7
2.25σ−18.9+1,530.997.73 / 7
2.5σ−812.6+895.665.02 / 7
2.75σ−1,133.3+990.740.01 / 7
−124.9+462.221.73 / 7
3.25σ+772.2+114.012.72 / 7
3.5σ+509.1+210.86.34 / 7
3.75σ+329.7+697.94.14 / 7
+394.5+989.92.34 / 7

All seven periods survived both years from 0.5σ through 1.25σ. At 1σ, the means were +1,568.8 pips in 2024 and +1,081.9 in 2025. Two sigma was also viable, with six of seven periods positive in both years, but its result depended more on period.

The apparent return to four survivors from 3.5σ upward needs the trade column beside it. Those settings averaged 6.3, 4.1 and 2.3 trades in 2025. A few successful outliers are not the same as a stable setting.

The whole 105-setting breakout grid changed with timeframe.

TimeframeYearProfitable settingsMedian netPositive in both years
M15202555 / 105+138.5 pips44 / 105
M15202465 / 105+404.3 pips44 / 105
H1202575 / 105+672.9 pips67 / 105
H1202486 / 105+902.5 pips67 / 105
H4202532 / 105−270.8 pips27 / 105
H4202484 / 105+926.6 pips27 / 105

Hourly bars retained 67 settings in both years. Four-hour bars went from 84 winners in 2024 to 32 in 2025, which is why the answer here is explicitly an hourly one.

Are 2 standard deviations any good?

At period 20 alone, 1.75σ beat both 1σ and 2σ in both years. Fixing the period makes the frequency change easier to see.

SigmaYearTradesWin ratePFNet
202527738.27%1.137+1,052.5 pips
202428034.29%1.281+2,071.7 pips
1.75σ202515844.94%1.348+1,846.0 pips
1.75σ202416940.24%1.407+2,331.5 pips
202513243.18%1.235+1,177.5 pips
202413741.61%1.270+1,434.6 pips
2.75σ20254841.67%0.703−955.6 pips
2.75σ20244850.00%1.219+791.9 pips
20252437.50%1.017+33.7 pips
20243444.12%1.255+780.1 pips

Across all seven periods in 2024, however, 1.75σ averaged +590.2 pips versus +1,568.8 at 1σ. A result after fixing one period is a different comparison from a result averaged across periods.

Which sigma for mean reversion?

No sigma was stable for the fade. The 105 hourly settings point in a different direction when the band is faded.

Trading rulePositive 2025MedianPositive 2024MedianPositive both
Fade long43 / 105−114.4 pips48 / 105−20.5 pips23 / 105
Fade short11 / 105−483.9 pips25 / 105−995.2 pips1 / 105
Breakout75 / 105+672.9 pips86 / 105+902.5 pips67 / 105

Widening the entry often made the fade-long loss smaller. Averaged over seven periods, 1σ made −811.6 pips in 2025 and −111.6 in 2024. At 2.75σ the figures were +230.2 and −45.3; at 3σ, +35.6 and −12.7.

The fade-long result at 2.75σ was +230.2 pips in 2025 and −45.3 in 2024; at 3σ it was +35.6 and −12.7. The short fade retained just one of 105 settings in both years. A wider band may simply trade less and therefore lose less; it did not establish a repeatable mean-reversion setting.

Does one year's best sigma hold?

The single-year maximum shrank the next year, while 1σ held in both. Choosing sigma by the largest 2025 mean selects 1.75σ at +1,850.2 pips. In 2024 it made +590.2. The 1σ setting made +1,568.8 in 2024 and +1,081.9 in 2025.

Selection yearSigmaNet in selection yearCheck yearNet in check year
20251.75σ+1,850.2 pips2024+590.2 pips
2024+1,568.8 pips2025+1,081.9 pips

Optimizing period and sigma together is less stable still. Period 10 and 1.75σ made +2,753.0 pips in 2025 but +686.0 in 2024. Period 25 and 2.25σ made +2,793.5 in 2024 but +231.3 in 2025. That is why the conclusion favors several nearby settings that stayed positive in both years over one maximum cell.

Can win rate choose sigma?

No. Averaged over all 105 hourly settings in 2025, the fade long won 61.9%, the fade short 61.2% and the breakout 41.0%.

Trading ruleMean win rateMean winMean loss
Fade long61.9%+39.2 pips−65.9 pips
Fade short61.2%+40.2 pips−84.8 pips
Breakout41.0%+188.8 pips−93.0 pips

The fade is right more often but loses more when wrong. The breakout's average winner is about twice its average loser. Looking only for the sigma with the highest hit rate discards that payoff shape.

Filters, exits and costs

We added common filters to the period 20 and 1σ breakout used as the robust baseline.

ConditionYearTradesAnnual net
Plain2025277+1,052.5 pips
Plain2024280+2,071.7 pips
ADX ≥ 202025188−1,033.2 pips
ADX ≥ 202024185−3.9 pips
ADX ≥ 252025134−1,572.2 pips
ADX ≥ 252024126+218.7 pips
ADX ≥ 30202594−972.7 pips
ADX ≥ 30202488−231.8 pips
Tokyo hours2025164+519.4 pips
Tokyo hours2024162+2,139.6 pips
London and New York hours2025207+1,073.5 pips
London and New York hours2024218+2,154.3 pips

None of the ADX thresholds beat the baseline in both years. London and New York hours improved it slightly in both, from +1,052.5 to +1,073.5 in 2025 and +2,071.7 to +2,154.3 in 2024, but that session result needs more pairs and years before it becomes a setting recommendation.

ExitYearAnnual net
Opposite 1σ only2025+1,052.5 pips
Opposite 1σ only2024+2,071.7 pips
SL 30 / TP 602025+1,687.6 pips
SL 30 / TP 602024+1,467.1 pips
SL 50 / TP 1002025+1,157.3 pips
SL 50 / TP 1002024+1,896.2 pips
SL 100 / TP 2002025+1,340.5 pips
SL 100 / TP 2002024+1,987.6 pips
24-bar time exit2025+989.1 pips
24-bar time exit2024+1,559.0 pips

Fixed stops and targets are executable levels measured from the filled entry. When reached, they fill at the configured distance without another exit-side adjustment being deducted from that level.

Fixed stops and targets improved 2025 at the cost of 2024. None beat the plain rule in both years.

With spread varied on the 2025 period 20 and 1σ breakout, net profit moved from +1,135.6 pips at zero to +1,052.5 at 0.3, +858.6 at 1, +581.6 at 2 and +304.6 at 3 pips. Its simple break-even spread was about 4.10 pips. One sigma trades often, so low cost is part of the setting.

Which Bollinger Bands period works best? compares lookbacks from 5 through 200 and tests whether the best bar count changes with timeframe. The broader comparison between fading and following the bands is in 315 Bollinger Band backtests. RSI periods and 30/70 levels also show why a higher fade win rate need not produce more profit. MA disparity settings put the band at a percentage of price rather than a standard deviation, while backtesting without code shows how to rebuild the condition on your own pair.

Notes

  • One pair, USD/JPY, and two independent calendar years. Nothing here guarantees the same order on other pairs or regimes
  • Picking a winner from a period-and-sigma grid creates a multiple-comparison uplift. The other year checks it, but is not an untouched future sample
  • Three sigma and wider settings have small samples. A 4σ mean of 2.3 trades cannot be compared with 260.4 trades at 1σ at equal confidence
  • Touch rules use highs and lows; breakouts use closes. Counting wick breaks as signals would produce a different test
  • Fills are at bar closes and spread is fixed. Live fills and changing spreads are not reproduced
  • The fade exit stays at the opposite 1σ band to isolate entry sigma. A study that optimizes the exit as well answers a different question

Questions people ask

What is the best standard deviation for Bollinger Bands?
For the hourly USD/JPY breakout in this test, 1σ is a defensible starting point. 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 it was +1,850.2 in 2025 but only +590.2 in 2024.
Are 2 standard deviations good for Bollinger Bands?
They were reasonable, but less stable than 1σ for the hourly breakout. Six of seven periods at 2σ were profitable in both years. Period 20 and 2σ made 1,177.5 pips in 2025 and 1,434.6 in 2024. Measured containment was 87.1%, below the familiar theoretical 95%.
Are 3σ and 4σ Bollinger Bands stronger signals?
They are rarer, not automatically stronger. The hourly breakout averaged 21.7 trades a year at 3σ, 6.3 at 3.5σ and 2.3 at 4σ in 2025. A few large moves can decide the whole result at those sample sizes, so the number of profitable settings is not enough evidence.
Which Bollinger Band sigma is best for mean reversion?
This USD/JPY test found no stable answer for the fade. Of 105 hourly settings, only 23 fade-long settings and one fade-short setting were profitable in both years. Wider bands reduced the number of trades and often reduced the loss, but few settings stayed profitable in both years.
How much price stays inside 1σ, 2σ and 3σ Bollinger Bands?
On 6,226 hourly USD/JPY bars in 2025 with period 20, 45.0% of closes finished inside 1σ, 87.1% inside 2σ and 99.1% inside 3σ. The familiar 68%, 95% and 99.7% are normal-distribution figures, not guaranteed market frequencies.

Formiq is a free browser-based FX terminal with replay practice and no-code backtesting. Open the chart or see what the free plan includes.