Bollinger bands

A moving average tells you where a stock has been trading lately. It does not tell you whether today's price is unusual.

Bollinger bands add that second dimension. They use the normal distribution to calculate how far the market price has strayed from the moving average, measured in standard deviations of the price itself. Drawing a band at plus and minus two standard deviations creates an envelope that, statistically, should contain most of the price action.

When price pushes outside it, something out of the ordinary is happening — and the technical reading is that the move has gone too far.

Three lines. The middle one is a moving average of past prices; the workbook notes that 5, 10, 30, 50, 100 and 200-day moving averages are the ones generally calculated. The outer two sit a chosen number of standard deviations above and below it.

The interpretation is symmetric:

• price two standard deviations above the moving average — the stock might be regarded as overbought;
• price two standard deviations below — the stock might be regarded as oversold.

The bands are not fixed in width. Standard deviation is computed from recent prices, so the envelope widens automatically when the stock becomes volatile and narrows when it goes quiet. A band that is wide is not a stronger signal; it is a noisier stock.

The whole family of techniques rests on the technical analyst's premise: past prices and trading volume can be used to predict future prices. The workbook is explicit that these techniques apply to fixed income securities too, wherever price and volume data are available — the theory and rationale for technical analysis of bonds are the same as for stocks.

A stock is being tracked against its 30-day moving average, one of the periods the workbook lists.

The stock closes at Rs 508. It is above the upper band, so on this indicator it might be regarded as overbought.

Now the volatility adjusts. Results are announced, the stock swings hard for a fortnight, and the standard deviation over the window rises to Rs 21 while the moving average drifts up to Rs 492:

Upper band = 492 + 42 = Rs 534
Lower band = 492 - 42 = Rs 450

The envelope has widened from Rs 48 to Rs 84. The same Rs 508 close that was an overbought signal a fortnight ago is now comfortably inside the band and signals nothing at all.

In rupees. A trader holding 2,000 shares — a position of Rs 10,16,000 at Rs 508 — who sold on the first signal would have booked out of a stock that the indicator, two weeks later, would not have flagged. The indicator did not change its mind; the stock's volatility changed underneath it.

That is the practical lesson, and it is why the workbook frames the readings with "might be".

Chapter 8 (Investing in Stocks), section 8.7 on technical analysis, immediately after moving-average analysis, and just before section 8.7.5 on applying technical analysis to fixed income securities. Questions are definitional: what do the bands measure, what does a price two standard deviations above the moving average indicate, and which statistical distribution the tool assumes.

• Overbought is the upper band, oversold is the lower. Reversing the pair is the most common error, and both options are always offered.
• The bands measure deviation from a moving average, not from a fixed price or a trend line. The middle band moves every day.
• Standard deviation here is of the price, not of returns, and it is recomputed over the window — which is why the bands breathe.
• The workbook does not prescribe a lookback period. It gives the list of common moving averages and the two-standard-deviation reading; a question that demands "the standard 20-day period" is going beyond the syllabus text.
• A band touch is a signal, not an instruction. The workbook says a stock "might be regarded as" overbought, and technical analysis is explicitly a probabilistic method, not a valuation method.
• These tools work on bonds as well as shares. The workbook says the theory and rationale are the same, and many of the same trading rules are used in bond markets.
• Technical analysis ignores fundamentals by design. It is not an alternative computation of intrinsic value; it is a different question altogether.