Technical Analysis · Chapter 1 of 12
What technical analysis actually claims
Three premises, only two of which are defensible. Start here, because knowing which part of this discipline has evidence behind it is what separates a practitioner from a believer.
Technical analysis is the study of price and volume history to inform decisions about the future. It is enormously popular, widely dismissed by academics, and used daily by professional traders at firms that would not survive a year on superstition. Those three facts are all true at once, and reconciling them is the most useful thing this chapter can do for you.
The reconciliation is this: technical analysis is not one claim, it is a bundle of them, and they differ enormously in how much support they have. Treating the bundle as a single thing — either ‘it works’ or ‘it's astrology’ — is what makes the argument unproductive.
The three classical premises
- 1. Price discounts everything. All known information is already reflected in the price, so you do not need to analyse the news, only the price. This is close to the efficient markets hypothesis and is broadly defensible. It is also the premise that quietly undermines most of what people do with technical analysis, because if price already reflects everything, historical price cannot contain unexploited information.
- 2. Price moves in trends. A trend in motion is more likely to continue than to reverse. This one has genuine empirical support. Cross-sectional momentum — buying recent winners and selling recent losers — is one of the most replicated anomalies in the academic literature, documented across decades, across dozens of countries including India, and across asset classes. Time-series momentum, or trend following, has a similar record. This is the strongest leg the discipline has.
- 3. History repeats itself. Patterns recur because human psychology is constant. This is the weakest premise and where most of the folklore lives. It is not that psychology is not constant — it plainly is — but that ‘this shape appeared and then price rose’ is exactly the kind of claim that survives on selective memory and dies under a proper control.
The single most important sentence in this book
A pattern that precedes a rise 60% of the time is worthless if price rises 60% of the time anyway.
This is the control problem, and it invalidates the overwhelming majority of technical analysis you will read online. Almost nobody who publishes a pattern's ‘success rate’ also publishes the base rate it should be compared against.
Every chapter that follows will name the control you would need in order to believe the claim. That habit — asking ‘compared to what?’ before asking ‘how often?’ — is worth more than any indicator setting you will ever learn.
What the evidence actually supports
| Claim | Evidence position | Practical reading |
|---|---|---|
| Cross-sectional momentum | Strong. Replicated for 30+ years across markets, including India | The most defensible thing in the whole discipline. Buy relative winners, rebalance periodically |
| Time-series trend following | Good, particularly in futures and commodities | Works; has long painful drawdowns; the difficulty is behavioural, not analytical |
| Volatility clustering | Very strong. One of the most robust facts in finance | Volatile periods follow volatile periods. This is why ATR-based sizing works and why it is under-used |
| Volume confirming price | Mixed, and context-dependent | Weaker than commonly assumed for equities; more informative in Indian markets via delivery percentage and open interest |
| Support and resistance levels | Weak to moderate, and heavily dependent on how the level is defined | Round numbers and prior extremes do attract activity. Most drawn levels are not those |
| Candlestick patterns in isolation | Weak. Studies that use proper controls find little | Not useless as context. Close to useless as a standalone entry signal |
| Chart patterns (head and shoulders, flags) | Weak, and severely affected by the subjectivity of identification | If two analysts cannot agree it is present, it cannot be tested, and if it cannot be tested you are guessing |
| Fibonacci retracement levels | No credible support for the specific ratios | Price does retrace. That 61.8% is special rather than 60% or 65% has never survived a proper test |
Why professionals still use charts
Given that table, why does a systematic trading desk have charts on every screen? Three reasons, all of them honest, and none of them ‘the pattern predicts the future’:
- A chart is a compression of a great deal of data. Six months of tick data is unreadable; six months of daily candles is one glance. Using a chart to orient yourself is not a claim about prediction.
- Levels are self-fulfilling to a degree. If enough participants place orders at an obvious level, activity genuinely concentrates there. This is a real mechanism and it does not require the level to have any independent significance. It also means the obvious levels matter more than the clever ones, which is the opposite of what most people assume.
- Risk management needs reference points. The most valuable thing technical analysis provides is not entries; it is a defensible, non-arbitrary place to put a stop and a way to size a position relative to volatility. A trader with a mediocre entry rule and excellent risk placement will out-perform the reverse, reliably.
The trap to name out loud
Hindsight is the enemy. Every chart in every book has the pattern circled after the fact. On the right-hand edge of a live chart, the pattern is ambiguous, the swing points depend on the lookback you chose, and the same price action supports three incompatible readings.
The test: cover the right half of any chart and mark your decision on the left half only. Then reveal. Do this fifty times and record the results honestly. Almost nobody does this, and almost everybody who does is surprised.
The professional version of the same discipline is a walk-forward test: fit or choose parameters on one period, evaluate on the next, never look at the evaluation period while deciding. It is the same idea with the self-deception engineered out.
Check yourself