Bollinger Bands Strategy Tested on the 5-Minute Chart: The Fade Won 60% of Trades and Still Lost

Bollinger Bands are one of the most searched indicators in forex, and “Bollinger Bands strategy” is one of the phrases people add to it. Most articles answering that phrase describe the same three rules: sell the upper band, buy the lower one; wait for price to come back inside first; or, in a strong move, ride the band.

I already tested one Bollinger idea here — the squeeze breakout, in my Bollinger squeeze test. This is the other half: the everyday band strategies, on the chart most beginners actually use, the 5-minute one.

How were the Bollinger strategies tested?

The data is 16,926 EURUSD 5-minute bars from June 3 to August 26, 2026 — the same bars as every EURUSD test on this site. Bollinger Bands are calculated the standard way: a simple moving average of the close, with bands a set number of standard deviations above and below.

Three entry rules, all on the close of the bar:

• A. Fade: the bar closes above the upper band — sell. Closes below the lower band — buy.
• B. Re-entry fade: one bar closes outside the band, the next closes back inside — trade against the move. This is the “wait for confirmation” version.
• C. Walk the band: the bar closes outside the band — trade with it, expecting the move to continue.

One exit for all three, so only the entry is compared. The trade closes when a bar closes through the middle band, when it loses as much as the distance from entry to the middle band (a 1:1 stop, checked on closes), or after two hours. Every trade pays 1 pip, the EURUSD cost measured in my spread and swap test. One position at a time.

The grid is period 10, 20 and 50 against deviation 1.5, 2, 2.5 and 3. The default in MetaTrader is 20/2.

Which Bollinger Bands strategy did best on the 5-minute chart?

The plain fade was the best of the three and the only one with a clear edge before costs: +0.66 pips per trade gross. Then the spread took a full pip and left -0.34.

The interesting number is the win rate. Six trades in ten closed in profit. Most people who see that on a demo account would call the strategy working. It is not, because the losers were 1.7 times the size of the winners: 60.3% × 4.09 is 2.47 pips earned, 39.7% × 7.07 is 2.81 pips lost.

That shape comes straight from the rule. The target — the middle band — keeps moving toward the entry as the band contracts, so winners get cut short. The stop does not move. The Bollinger fade did better here than the plain overbought trade in my Stochastic overbought test, which won only 44.7% of the time — the band picks more extreme moments — but the payoff problem is the same: on a fast chart, “too far” snaps back often and not far.

Does any Bollinger setting make money after costs?

Sorted by net pips per trade; the halves split the data on July 15, 2026. Only the fade on 10/2.5 stayed positive after costs, and it did so in both halves. That is the good news. The rest of the news: +0.05 pips per trade is 10.5 pips across three months. One slightly wider spread and it is gone.

Two patterns are more useful than the winner:

• The 50-period bands were the worst fades. The middle band sits so far away that 29.8% of 50/2 trades ran out the two-hour clock instead of reaching it.
• Deviation 3 barely trades. On 10/3, price closed outside the band 10 times in three months. Too few trades to judge — I am leaving 10/3 out of every conclusion here.

Tuning the grid until one setting turns positive is the trap my optimised vs default settings test warned about. 10/2.5 holding in both halves makes it less likely to be pure luck, but the margin is too thin to trade.

Is it better to wait for price to close back inside the band?

Net pips per trade. On every setting with enough trades, waiting for confirmation made the fade worse. The logic of waiting is that you avoid selling into a runaway move. The cost is that by the time the bar closes back inside, half of the snap-back has already happened, and the middle band is closer. You still take the same full stop risk for a smaller target.

It is the same lesson as my RSI divergence test, where 75% of the move was gone before the confirmation arrived. Confirmation feels safer; on a 5-minute chart it mostly arrives late.

Does “walking the band” work on EURUSD M5?

No. Rule C lost in every setting that traded more than 10 times — before the spread as well as after. The best of them, 50/3, lost 0.20 pips per trade gross over 125 trades.

The reason shows up in a simpler count. On 20/2, 11.34% of all 5-minute closes landed outside a band. I followed 1,913 of those closes forward for up to four hours:

On a 5-minute EURUSD chart, a close outside the band was more often the end of a move than the start of one. The squeeze version — a breakout after the bands had been unusually narrow — is a different setup, and it did better in my squeeze test. The plain “price broke the band, follow it” rule did not.

That changes on slower charts. On 30-minute bars, the same walk-the-band rule made money in both halves of the data — the timeframe question is the next test in this series, the same one my Supertrend MT5 vs MT4 comparison asked of another indicator.

When in the day did the band fade work best?

The London open was the worst time to fade the bands: the morning moves were the ones most likely to keep going. The overlap was the only positive window, at +0.29 pips on 156 trades. That is one slice of one test, and I have not checked it in both halves, so treat it as a lead rather than a rule. It does match what my best hours to trade EURUSD measurement found about where the range is.

Frequently asked questions

What is the best Bollinger Bands strategy for the 5-minute chart?

Of the three tested on 16,926 EURUSD 5-minute bars, fading a close outside the band was the best: 60.3% winners on 20/2, but -0.34 pips per trade after a 1-pip cost. Only 10/2.5 was positive after costs, at +0.05 pips per trade over 204 trades.

What is the win rate of the Bollinger Bands strategy?

About 60% for the fade on most settings, 54.6% for the re-entry version on 20/2 and 35.1% for walking the band. The fade’s high win rate came with winners of 4.09 pips against losers of 7.07, so it still lost money.

What are the best Bollinger Bands settings for scalping?

10/2.5 was the only setting of 36 that made money after costs, and it was positive in both halves of the data — but by only 0.05 pips per trade. The default 20/2 lost 0.34. Settings with period 50 were the worst for fading on M5.

Should you trade Bollinger Band breakouts on the 5-minute chart?

Not with the simple rule. Following a close outside the band lost in every setting with more than 10 trades, before costs too. After a close outside 20/2, price touched the middle band first 68.3% of the time.

Is it better to wait for a close back inside the band?

No, in this test. Waiting for the re-entry made the fade worse on every setting with enough trades, cutting 20/2 from -0.34 to -0.91 pips per trade.