John Bollinger’s contribution in the early 1980s was to make a price envelope breathe. Fixed percentage bands sit too wide in a quiet market and too narrow in a violent one. Bollinger tied the width to the rolling standard deviation of price, so the envelope expands when the stock gets volatile. The call is @BBANDS(close, 20).
How it is computed
middle = SMA(close, 20)
upper = middle + k * stdev(close, 20)
lower = middle - k * stdev(close, 20)
The conventional multiplier k is 2. Three series come out of one call, and a rule names which of them it wants. All three are rupee prices, so they compare directly to price:
close > @BBANDS(close, 20)
using the upper band as the output. Because the grammar allows arithmetic around an indicator, you can also express the band width as a fraction of price to compare volatility across securities.
Typical parameters
| Setting | Convention | Effect |
|---|---|---|
| 20-period window | Bollinger’s original | roughly one trading month on NSE |
k = 2 | Bollinger’s original | under a normal distribution, about 95% of observations |
| 10-period window | shorter | tighter, reacts faster, more touches |
| 50-period window | longer | wider, fewer touches |
The 20 and 2 pairing is Bollinger’s own choice, chosen for readability on a chart. It is not derived from Indian equity returns, and the 95% figure that gets quoted alongside it assumes normally distributed returns, which equity returns are not. Real return distributions have fat tails, so price touches the bands more often than the normal approximation implies.
How to read the output
Price at the upper band means price is two standard deviations above its own 20-day mean. Whether that is a reason to sell or a reason to buy depends entirely on the regime, and the bands themselves do not tell you which regime you are in. Bollinger has said as much repeatedly.
Band width is the more durable read. A narrow envelope means realised volatility has been low. That is a measurement, not a prediction that a large move follows.
Risks and caveats
The failure mode is the mirror of the RSI trap. In a strong trend a stock can ride the upper band for weeks, and a rule that fades every touch will short strength repeatedly. RELIANCE and index heavyweights do this during sustained runs.
The standard deviation is computed on the same window as the mean, so in a trending market the bands widen because of the trend itself, not because of any change in noise. That inflates the estimate exactly when you want it to be clean.
Two free parameters, the window and the multiplier, tuned on one history will produce a flattering result. Test the survivor on a period you did not search.