Bollinger on Bollinger Bands — Bollinger: COMPLETE

The birthplace was a TV studio. In the early 1980s Bollinger worked as chief market analyst at Financial News Network, FNN, and every day had to have an answer ready on EVERYTHING: stocks, indexes, commodities, currencies. The standard tool of the era was the percentage bands — a moving average with a fixed percentage width above and below — and Bollinger used them like everyone else, with David Bostian's Intraday Intensity as an extra vote on supply and demand.

He has told it himself, quoted openly: we used percentage bands and compared the price's position in the band with tools like Bostian's Intraday Intensity. The problem was that the bands did not hold up: the same width that fit a sleepy large-cap stock was a joke for a volatile small-cap stock — and the same stock switched volatility regime between earnings week and holiday week, between bubble and crisis. The tool needed to BREATHE.

This was where the microcomputer stepped in — and with it the ability to compute standard deviations per stock, per day, at scale. Bollinger systematically tested combinations of periods and multipliers across broad selections of stocks, and the result was Bollinger Bands: a 20-day simple average surrounded by bands at the average plus and minus two standard deviations. The point — and it is the book's entire thesis, worth memorizing before chapter two — is not the numbers but the definition: the bands give a RELATIVE definition of high and low.

The price is high at the upper band and low at the lower one, whether the price is 15 or 15 000 kronor, whether the market is tense or sleepy. The book documenting all of this did not arrive until 2001 — Bollinger on Bollinger Bands, McGraw-Hill — after nearly two decades in which the tool had already been used all over the world. It is the only book about the bands written by the bands' own creator, and this course follows it chapter by chapter.

The book's architecture in brief: first the history and the construction (the predecessors, the mathematics, the parameters), then the tools (%b, BandWidth, volume indicators, patterns in the bands' language) and finally the system building with two documented example systems. The differentiation from the neighbouring courses, openly edited: the Edwards & Magee and Murphy courses own the grammar of chart reading, Bulkowski owns the pattern statistics — but only this book explains how a tool is CONSTRUCTED and tested, by the person who made it.

That makes it the catalog's methods course in indicator building: whoever understands why 20/2 was chosen understands how to judge any other indicator's parameters — and their temptations.

AKM1 BRIDGE · V12 revenue stability — volatility lives in the fundamentals too: companies with stable, predictable revenue generally have calmer price volatility than companies with swinging revenue; the bands' breathing is partly V12's shadow in the price.

AK1TS ROW: the Breakpoint dimension — the course's rails: the bands' core idea, compression begets breaks, lives on in the engine as zones and breakouts, and that connection is built out properly in chapter 7 and 13.

The bands consist of three lines computed on the same N closes, standard N = 20. The middle band: the SIMPLE moving average — the sum of the last twenty closes divided by twenty (Bollinger chooses simple over exponential deliberately: for simplicity and transparency). The upper band: the middle band plus the multiplier times the standard deviation. The lower band: the middle band minus the same product. Standard N = 20, multiplier = 2. One technical detail the book takes seriously: there are two versions of the standard deviation — the population (divide by n) and the sample (divide by n − 1).

Bollinger himself uses the population version; the difference at twenty periods is a fraction of a percent but it exists, and it explains why different chart services can show microscopically different bands. Know which one your service uses — but do not fight over the number.

The standard deviation step by step, in plain text. Step one: compute the mean of the N most recent closes. Step two: compute each day's deviation from the mean (close minus mean). Step three: square each deviation — it makes all deviations positive and punishes the large ones harder. Step four: sum the squares and divide by N — that is the variance. Step five: draw the root — that is the standard deviation σ, the dispersion around the mean in the same unit the price is quoted in — the unit of account in this course.

The bands are thus the mean plus and minus the multiplier times the dispersion: when prices sit tightly clustered σ becomes small and the bands narrow; when they soar σ becomes large and the bands widen. σ is the bands' lungs — that is the whole secret of why they breathe.

The worked example, complete and verifiable — with five days instead of twenty, since the arithmetic is identical and only longer. Closes: 100, 102, 101, 103, 104. Mean: 510 / 5 = 102. Deviations: −2, 0, −1, +1, +2. Squares: 4, 0, 1, 1, 4 — sum 10. Variance (population): 10 / 5 = 2. Standard deviation: the root of 2 ≈ 1,41. Upper band: 102 + 2 × 1,41 = 104,83. Lower band: 102 − 2 × 1,41 = 99,17. With the sample version the variance would have been 10 / 4 = 2,5 and σ ≈ 1,58 — the bands 105,17 and 98,83: a visible but small difference. For real, with twenty days, exactly the same five steps are done — the table below is the whole construction in miniature.

The parameters: 20 and 2 are DEFAULTS, not magic, and the book gives its only recommended alternatives: a shorter average requires a lower multiplier — 10 days with 1,9 σ; longer requires higher — 50 days with 2,1 σ. And Bollinger's own warning, quoted verbatim in the end: do NOT make statistical assumptions from the computation — price is not normally distributed, and the bands are not confidence intervals (the controversy returns in chapter 12).

AKM1 BRIDGE · V04 P/S, V05 P/B, V06 EV/EBITDA — the multiples are also lagging averages in disguise: a P/B below its own five-year average is %b thinking applied to valuation — where in its band is the multiple trading? The wave foundation asks that question per variable. AK1TS ROW: the Trend dimension — MA50 and MA200 in the engine are the same tool family as the middle band; the engine classifies waves (impulse wave, correction, base building) while the bands measure the distance to the mean.

The construction is kin; the use is separate.