Bollinger Bands: Reading Price Position and Volatility with Standard Deviation
How Bollinger Bands are calculated, what the default 20-period/2-standard-deviation setting means, how to read band walks and squeezes, common misreadings of the ±2 SD range, and what backtests actually found.
·Trading-Journl
On this page
Bollinger Bands use a 20-period simple moving average as a center line, with an upper and lower band set two standard deviations of price above and below it. Because the distance between the bands and the center line is tied directly to that standard deviation, the bands widen when volatility rises and narrow when it falls.
Touching a band, however, is not a buy or sell signal on its own. John Bollinger's own rules page states that a close outside the bands should first be read as a sign of trend continuation, not reversal. And the ±2 standard deviation range should not be treated as a 95% statistical confidence interval. This post walks through the calculation and default settings, how to read squeezes and band walks, two common misreadings, and what the backtesting research actually found.
What Are Bollinger Bands?
Bollinger Bands consist of a center line (SMA 20), an upper band (SMA 20 + 2σ), and a lower band (SMA 20 − 2σ). All three lines are calculated from the same price window, where σ is the standard deviation of closing prices over that period (BollingerBands.com; Fidelity).
The distance between the bands and the center line is directly tied to that standard deviation. When volatility rises, the bands widen; when it falls, they narrow (Fidelity; Charles Schwab). Band width shows how much volatility is present, but on its own it doesn't tell you which direction price will move next.
The 20-Period, 2-SD Calculation and How Settings Differ
The default setting is 20 periods and 2 standard deviations. Non-default settings, though, vary by source. Bollinger's own rules reduce the standard deviation multiplier for shorter periods and increase it for longer ones to keep the proportion of price action inside the bands roughly consistent, while Fidelity's education material gives different example values for short-, medium-, and long-term use (BollingerBands.com; Fidelity).
Period / Use
Bollinger's official rules (BollingerBands.com)
Fidelity's example setting
Short-term
10 periods, 1.9 SD
10-day, 1.5 SD
Default
20 periods, 2 SD
20-day, 2 SD
Long-term
50 periods, 2.1 SD
50-day, 2.5 SD
Both sources agree on the 20-period/2-SD default, but the non-default settings serve different purposes and shouldn't be swapped between the two frameworks. Don't generalize any single combination as the optimal setting for every asset or market.
Reading the Bands: Relative Highs/Lows and the Band Walk
Touching the upper or lower band is not, by itself, a sell or buy signal. In a strong trend, price can ride along one band for an extended stretch — what's called a band walk (BollingerBands.com; Charles Schwab). The bands aren't an absolute overbought/oversold boundary; they mark a relative high or low within the current window.
Consistent with that, Bollinger's own rules read a close outside the bands as a signal that the trend is more likely to continue than reverse (BollingerBands.com). False breakouts do happen, though, so a close outside the bands should be confirmed with trend direction, volume, momentum, or a price pattern.
For Korean (KRX) stocks, the daily price-limit system and lower liquidity in smaller names deserve extra attention. Thinly traded KOSDAQ small caps can see standard deviation swing sharply on just a handful of quotes, narrowing or widening the bands in ways that don't reflect real supply-and-demand change.
Using the Squeeze, BandWidth, and %b
BandWidth normalizes band width as (upper band − lower band) ÷ center line, and %b shows where the current price sits within the bands (BollingerBands.com; the AUT working paper). A squeeze is a stretch where BandWidth drops to a notably low level, and it's used to watch for volatility that may be about to expand.
A low BandWidth reading alone, though, doesn't tell you which direction a breakout will go or whether it will succeed. A squeeze is a signal about the volatility regime, not about direction. That means a squeeze calls for having volume, price pattern, and a stop-loss plan ready, not for guessing a direction in advance.
Two Common Misreadings
The first misreading treats a band touch as a mechanical trade signal. As covered above, band walks can repeat through a trend, so entering against the touch alone can mean taking several losses before the trend actually ends.
The second misreading treats ±2 standard deviations as a 95% statistical confidence interval. Bollinger himself warns against this, since price distributions aren't normal and the calculation window is short — he reports that the actual share of price action contained within the default bands has typically run closer to about 90% (BollingerBands.com). That ~90% figure, though, is his own practical observation rather than a reproducible formal estimate — the market sample, period, and calculation method behind it aren't published.
What the Research Found: Performance and Limits
The backtesting literature suggests Bollinger Bands don't hold up well as a standalone trading system. Lento, Gradojevic, and Wright (2007) tested BB(20,2), BB(20,1), and BB(30,2) against buy-and-hold on the TSX, DJIA, and NASDAQ indices and the CAD/USD exchange rate from May 9, 1995 through December 31, 2004. After accounting for transaction costs, only 1 of 12 return comparisons was statistically significant at the 10% level.
A working paper covering 14 major stock indices across 13 countries through March 2014 ("Popularity versus Profitability: Evidence from Bollinger Bands," Auckland University of Technology) shows a sharper decline over time.
Measure
Pre-1983
1983-2001
Post-2002
Markets with a significant positive result, breakout rule (out of 14)
14
(not reported)
2
Markets with a significant positive result, squeeze rule (out of 14)
9
(not reported)
1
Markets where risk-adjusted Sharpe ratio significantly beat the market, after a 1% switching cost
10
2
1
In the same study, the average daily buy-versus-sell conditional return spread for the BB(20,2) breakout rule fell from 0.454% pre-1983 to 0.002% post-2002. That drop coincides with the period when the indicator became widely known, but the study only shows that historical association — it does not establish that popularity caused the decline. The results are also limited to 4-14 markets and a specific historical sample, and outcomes can shift with different signal definitions and cost assumptions.
Fidelity's and Charles Schwab's education material both recommend using Bollinger Bands alongside other indicators, price patterns, volume, and risk management rather than as a standalone system. Combining indicators, though, doesn't guarantee a profit or cap a loss.
A Checklist Before Acting
Did you first identify whether the market touching the band is ranging or trending?
Did you consider a close outside the bands as a possible continuation signal, not just a reversal?
Did you check volume, momentum, or a price pattern to guard against a false breakout?
Are you treating a squeeze as a volatility signal rather than a directional one?
Are you reading ±2 SD as a relative position, not a 95% probability range?
Is the stock's trading value and liquidity high enough to trust the band width reading?
Do you have another indicator or risk-management rule ready alongside Bollinger Bands?
Frequently Asked Questions
Should I sell when price touches the upper band?
No. A touch of the upper band isn't a sell signal on its own — in a strong uptrend, price can ride the upper band for an extended stretch (a band walk).
Check trend direction, volume, and price pattern alongside the touch rather than reading it in isolation.
Can I assume 95% of price stays within ±2 standard deviations?
No. John Bollinger himself warns against that assumption, since price isn't normally distributed and the calculation window is short — the actual observed figure has typically run closer to about 90%.
That ~90% is his own practical observation, not a reproducible formal probability, so it's safer to treat the bands as a relative position marker rather than an absolute probability range.
Can I trade using Bollinger Bands alone?
The backtesting evidence doesn't support that. A 1995-2004 study across four markets found the strategy didn't consistently beat buy-and-hold after transaction costs, and a 14-country index study found the share of markets with significant results dropped sharply after 2002.
Fidelity and Charles Schwab both recommend pairing Bollinger Bands with other indicators, price patterns, volume, and risk management.
Do I have to stick with the default 20-period, 2-SD setting?
The default is the one point of agreement across sources, but non-default settings vary. Bollinger's own rules suggest 10-period/1.9 SD and 50-period/2.1 SD, while Fidelity's example uses 10-day/1.5 SD and 50-day/2.5 SD.
Changing the setting changes your comparison baseline, so avoid generalizing one combination as optimal for every asset — validate it consistently within the same stock and timeframe instead.
Key Takeaways
Bollinger Bands combine a 20-period moving average with standard deviation to show both relative price position and volatility — and the most important thing to avoid is reading a band touch as a trade signal on its own.
The bands consist of a center line (SMA 20) and upper/lower bands (±2σ), and band width widens or narrows with volatility.
A band touch is not a trade signal; band walks can repeat through a trend.
±2 SD is not a 95% confidence range — Bollinger's own reported figure runs closer to about 90%.
A squeeze and BandWidth show a volatility regime, not a direction.
Backtesting research found that performance of simple Bollinger Band strategies varied sharply by market, period, and cost assumptions, with fewer significant results in more recent samples.
If you act on a band touch or squeeze, recording the band width, volume, and market regime at that moment makes it easier to review later whether the read was correct.
References
John Bollinger, BollingerBands.com, "Bollinger Bands Explained. Rules to use Bollinger Bands successfully"
Fidelity, "Bollinger Bands®"
Charles Schwab, "Bollinger Bands: What They Are and How to Use Them"
Camillo Lento, Nikola Gradojevic, Chris S. Wright, "Investment information content in Bollinger Bands?", Applied Financial Economics Letters (2007)
"Popularity versus Profitability: Evidence from Bollinger Bands," Auckland University of Technology working paper (2014)
For reference only: The information in this post is provided for general reference and does not constitute investment advice or a solicitation to trade. Your responsibility: Final investment decisions and their outcomes are entirely your own. No warranty: This information is not guaranteed to be accurate and cannot be used as a basis for legal liability for any investment result.