Bollinger Bands

Petr enables Bollinger Bands in TradingView and sees three lines around the bitcoin price. The traditional center is an SMA, an ordinary average of prices in a chosen window that gives each observation equal weight. The upper boundary adds a multiple of standard deviation to the center; the lower boundary subtracts it. The deviation is calculated from prices themselves, not from the already smoothed average line. New observations move and reshape the corridor. It is not a predetermined track that bitcoin must follow. [2]

John Bollinger gives a length of 20 periods and a deviation multiplier of 2 as defaults. A period here is one chart interval: twenty hourly candles give a different view from twenty daily candles. When comparing charts, Petr also checks the market, price source and deviation calculation. Changing settings changes the question he asks of the chart. Defaults offer a starting point for learning the tool, not proof that they are best for every market. [1]

Fidelity explains band width through volatility, meaning variability in price. If prices in the chosen window spread out more while the multiplier stays unchanged, the boundaries move farther apart. As prices cluster closer to the average, the bands narrow. This tells Petr something about observed fluctuations, not the direction of the next move. A narrow corridor is not itself a promise of a rise, and a wide one does not automatically mean a fall is near. [3]

Petr sees price at the lower line and thinks it looks cheap. John Bollinger explicitly distinguishes a band touch from a trading signal. Price is low relative to the selected window, not necessarily relative to bitcoin’s underlying value. During a continuing decline, it may keep following the lower boundary or move beyond it. Likewise, touching the upper band need not be a reason to sell. The lines describe a relationship to the past; they do not create a price barrier. [1]

Petr remembers from statistics that standard deviation relates to probability. John Bollinger warns against precisely this shortcut. Particular statistical conclusions require assumptions about how data are distributed; naming a calculation does not establish those assumptions. Prices need not follow the familiar bell-shaped normal distribution, and a short window cannot establish certainty about the future. The bands are not a promised interval for future prices or insurance against loss. Using a multiplier of two adds no such guarantee. [1]

TradingView also offers %b, which expresses price’s position relative to the bands. Subtract the lower boundary from current price, then divide by the corridor’s width, the difference between its upper and lower boundaries. A value of 0 marks the lower edge and 1 the upper. Price can be outside the bands, and the result can go beyond these values too. Petr is measuring position, not width or the probability of profit. If the boundaries coincide, the ratio is undefined: division by zero is not possible. [4]

Petr watches price leave a band during the day, yet the chart looks different by evening. TradingView explains that Close on an open candle represents the latest price as it changes. A candle summarizes one time interval; its final close is not known yet. If the calculation uses Close, both price and the boundaries can change before the candle ends. Touching a band during a candle and closing outside it are therefore different events. A historical picture does not retain every state observed along the way. [5]

Petr might try a rule based on returning to the center, or one based on continuing after a break beyond a boundary. These are different assumptions; the name Bollinger Bands validates neither. TradingView’s strategy documentation separates chart calculations from simulated execution. A test must include fees and slippage, the gap between expected and achieved price. It must also address overfitting: settings refined on past data may fail on other data. Testing data that were not used for tuning is useful, but still creates no guarantee of future profit. [6]