Bollinger Band Strategy: Backtest Results on NSE Stocks

Bollinger Bands are a technical volatility indicator developed by John Bollinger in the 1980s. They consist of three distinct lines plotted directly on top of price action on a stock chart: a middle baseline, an upper band, and a lower band.

The middle line is a standard 20-period simple moving average (SMA). The upper band is positioned two standard deviations above that moving average, while the lower band sits two standard deviations below it. Because standard deviation measures statistical dispersion, the width of the bands widens automatically when market volatility surges and contracts when prices settle into tight consolidation ranges.

Most closes fall inside these outer boundaries, so a close outside them is unusual. In practical Indian stock trading across the National Stock Exchange (NSE), traders attempt to exploit this mathematical structure using two opposite philosophies: mean reversion (the bounce) and momentum expansion (the breakout).

Retail traders generally approach a bollinger band strategy through one of two distinct mechanics:

• The Mean Reversion Bounce: Assumes that prices touching or dipping below the lower band are statistically oversold and must snap back toward the mean. The system enters long when the candle crosses back above the lower band and exits when price reaches the 20-period middle SMA.
• The Volatility Breakout: Assumes that a sharp candle piercing above the upper band represents explosive institutional momentum. The trader buys on the upper band close and attempts to ride an extended trend until momentum exhausts.

Both approaches sound logical in theory, but theory ignores the heavy transaction costs of the Indian market. Below, we examine the quantitative reality across thousands of executed trades.

We backtested both mean reversion and breakout strategies on a portfolio of 15 highly liquid NIFTY constituents, including Reliance Industries, HDFC Bank, ICICI Bank, Infosys, and TCS. The testing period ran across one complete year with an account of ₹1,00,000, putting 20% of the account into each trade, with at most 5 trades open at once. Every test applied exact Indian market statutory levies including STT, exchange transaction fees, SEBI charges, stamp duty, GST, and execution slippage.

To strip away arbitrary stop losses and profit targets, we conducted a massive historical event study on all NIFTY 100 constituent stocks from July 2017 to October 2026. This study measured close-to-close forward returns exactly 1, 5, and 10 trading sessions following two specific signals: closing back above the lower band (bounce trigger) and closing above the upper band (breakout trigger).

The benchmark baseline shows the chance of any random NIFTY 100 stock being higher over 1, 5, and 10 days during the exact same historical period.

The data reveals a critical insight: following a close back above the lower band, stocks were positive 5 days later only 52.6% of the time, compared to 52.7% for all random trading days. In contrast, closing above the upper band saw stocks up 5 days later 53.2% of the time, slightly outperforming random days. However, as demonstrated in our backtest, this modest edge is completely overwhelmed once friction is deducted.

When trading mechanical setups on lower timeframes like 5-minute or 15-minute candles, the sheer volume of trades creates an insurmountable hurdle. For example, on the 15-minute mean reversion bounce strategy alone, the system executed 1,663 trades over the course of 12 months.

Before charges, the raw price moves generated -₹7,799. However, statutory friction—including Securities Transaction Tax (STT), brokerage, exchange transaction fees, SEBI turnover fees, stamp duty, GST, and execution slippage—consumed ₹26,414 in cash. This turned a modest gross loss into a severe net deficit of -₹39,185.

If you plan to utilize Bollinger Bands in your trading, you must avoid automated, high-frequency signal execution. Bands should be used as secondary context filters—such as identifying volatility compression before an earnings announcement or gauging extremes alongside volume—rather than standalone execution triggers.