# A Bollinger Bands strategy

> Build the Bollinger Bands bounce for Bitcoin without code and backtest it on Binance candles since 2020: the rule, the settings and what the numbers actually show.

Guide · Janorium · Published 2026-11-06  
Web page: https://janorium.com/guides/bollinger-bands-strategy-backtest

A Bollinger Bands strategy buys when the price comes back inside the lower band after leaving it, and sells at the middle band. This recipe builds one for Bitcoin without code and tests it. On BTC since 2020 it lost money at every setting tried, which is worth seeing before trusting the textbook.

## The rule

Bollinger Bands draw a 20-candle average with a band two standard deviations above and below it. The bet: a price that fell through the lower band stretched too far and returns to the average.

- **Enter** on a 4-hour candle that closes back above the lower band, with half of the free capital.
- **Exit** when the close is above the middle band — the 20-candle average.
- **Or exit earlier** at +5% profit or at a −4% loss.

> **Idea: Bounce off Bollinger Bands**
>
> Trades BTCUSDT on Binance spot, on 4-hour candles.
>
> Buys when the price crosses above the lower Bollinger Band (20, 2), using 50% of the free capital.
>
> Closes the trade at +5% profit or at a loss of −4%, whichever comes first. It also closes when the price is above the middle Bollinger Band (20, 2).
>
> Every entry and every exit pays 0.1% in fees plus 0.01% slippage.
>
> BTCUSDT 4h, 2020-01-01 → 2026-11-01 · a dry run on past candles: Return -46.98%, max drawdown 52.32%, 357 trades, Buy & Hold +976.84%.
>
> [Open in the wizard](https://janorium.com/strategies/new?idea=bollinger&setting=0&src=guide:bollinger-bands-strategy-backtest)

*The recipe as the wizard builds it. The button opens the wizard with these answers.*

## Three steps

1. **Open it.** Press **Open in the wizard** on the card and check the answers: coin, direction and market, candles, how far outside the bands, how much risk.
2. **Try it.** Press **Fine-tune in the editor**, then **Test on history**. Nothing is saved. Open the line under that button and tick **Compare against random entries** to see whether the bands beat a coin flip with the same exits.
3. **Keep it.** Save the strategy and press **Run backtest** on its page for the full report.

## What it did on Bitcoin

Dry runs on BTC 4-hour candles from 2020-01-01 to 2026-11-01. Holding BTC over the same period made about +980%.

| Setting | Trades | Return | Max drawdown |
| --- | --- | --- | --- |
| 2 standard deviations | 357 | about −47% | about 52% |
| 2.5 standard deviations | 215 | about −29% | about 33% |
| 3 standard deviations | 95 | about −14% | about 18% |
| 2, with the trend filter | 190 | about −32% | about 34% |
| 2, the short mirror on futures | 394 | about −33% | about 40% |

A wider band loses less mostly because it trades less. More than half of the trades won at every setting, and the strategy still lost: at 2 standard deviations the average win was a little over half of the average loss, so the average trade lost about 0.1% before costs and about 0.3% after them. One coin and one period, so run it on others. [The study on ten coins](https://janorium.com/articles/macd-bollinger-bands-crypto-backtest), which used no stop at all, points the same way: the textbook version won about two trades of three and still lost money in 16 of 20 runs.

## What to try next

- **A filter.** In the wizard, **with the bigger trend** keeps the strategy out while the price is below its 200-day average. [The trend-filter recipe](https://janorium.com/guides/trend-filter-trading-strategy) explains it.
- **Longer candles.** On daily candles there are far fewer trades, so costs take less. In the study four of ten coins ended in profit on daily candles and none on 4-hour ones.
- **The squeeze instead of the bounce.** The same bands, the opposite bet: [trade the break after they tighten](https://janorium.com/guides/volatility-squeeze-strategy-backtest).

## Frequently asked questions

### What are the best Bollinger Bands settings for crypto?

There is no best setting in general. (20, 2) is the textbook default and the place to start; a wider band trades less often, after sharper moves. On BTC in 2020–2026 none of the three settings tried here made money.

### Do Bollinger Bands work on crypto?

As a bounce rule, not on the platform’s own data: on ten large coins in 2020–2026 the textbook version won about two trades of three and lost money in 16 of 20 runs. [The study](https://janorium.com/articles/macd-bollinger-bands-crypto-backtest) has every run.

### How can a strategy win most trades and still lose?

The wins are small and the losses are larger, and every round trip pays costs. The win rate alone says little; look at the return and the max drawdown. [Every metric, explained](https://janorium.com/guides/backtest-metrics-explained).