Three claims sit under everything here. Charles Dow stated them in roughly this form in his Wall Street Journal editorials at the turn of the twentieth century. William Peter Hamilton and Robert Rhea organised them after his death, and Robert Edwards and John Magee gave them their standard textbook shape in Technical Analysis of Stock Trends (1948).
What are the three claims behind technical analysis?
The first claim is that the price already discounts everything. Every fact, every opinion and every forced seller is in the last traded price, so studying price is studying the aggregate of all of it. The second is that prices move in trends, and a trend in force is more likely to continue than to reverse. The third is that history repeats, because the patterns in a chart are patterns in human behaviour and human behaviour is stable.
Notice what happens if the first claim is true and strong. If price already contains all information, then studying past prices is studying information the market has already absorbed, and there is nothing left to extract. The strong version of claim one destroys claims two and three. Practitioners resolve this by holding claim one loosely, as "price is the best available summary" rather than "price is complete", which is a reasonable position and also an admission that the framework is not a theory in the sense a physicist would recognise.
Advertisement
What does the evidence say?
The empirical record splits cleanly along a line that most textbooks do not draw. Cross-sectional and time-series momentum, meaning the tendency of recent winners to keep winning over horizons of three to twelve months, is one of the best-documented anomalies in finance, established by Narasimhan Jegadeesh and Sheridan Titman in 1993 and replicated across markets and decades since. That is a technical finding by any definition, since it uses only past prices. It is studied and traded as a factor, and the quantitative galaxy builds it properly. Breadth statistics, which aggregate across hundreds of securities rather than reading one price, have a respectable if noisy record as risk indicators.
Almost everything else, meaning the named patterns, the oscillator thresholds and the wave counts, has either failed careful testing or has never been specified precisely enough to be tested at all.
That is the scorecard. A value-first programme still teaches the whole of it, for three reasons that have nothing to do with whether the patterns work.
Why learn technical analysis anyway?
Your counterparty uses it. A large part of futures and currency trading is run by systematic trend followers whose entire input is past price, and a large part of retail order flow in India is placed by people reading exactly the charts described in the technical analysis module. When a widely-watched line breaks, orders fire. Those orders move the price, and the move is real money whatever you think of the reasoning that caused it.
It is examinable. Chart types, trend, patterns, oscillators and the elementary intermarket ideas appear in professional curricula, including the CFA Program's first level (check against the current curriculum outline, which is revised annually). An analyst who cannot define a head and shoulders is going to look unserious in a room where several people can.
And the arithmetic is genuinely useful even when the conclusions are not. Computing a delivery percentage, an advance-decline line and a rolling standard deviation from raw data is a skill that transfers to every other part of the craft. The habit of demanding that a claimed signal survive a trial count is the same habit that protects a discounted cash flow model from its own assumptions.