The Truth About Technical Analysis
Is TA Just a Self-Fulfilling Prophecy? The Data Says Yes and No
There is a certain type of trader who discovers the relative strength index and behaves as if he has been handed a private instrument of prophecy. He opens a chart, sees the number rise above seventy, and announces that the asset is overbought. He sees it fall below thirty and announces that the asset is oversold. The words comfort him. They have the pleasing sound of measurement. They appear scientific enough to protect him from the shame of guessing. Then the asset keeps rising or keeps falling.
An indicator used in the technical analysis world, at the most simple definition, is a compression of what has already happened. Most popular technical indicators take what millions of people have done for the past x number of days and uses this data to predict what will happen for the next y number of days. This makes technical analysis like a map. Every map only shows a certain part of the world, and even complete maps do not show the world properly. They are distorted. The fallacy that most people encounter in technical analysis is that they forget what has been distorted, or that the map has been distorted at all; the market is the real world, the chart is only the map.
The real world contains forced sellers, leveraged buyers, pension funds rebalancing, options dealers hedging, institutions executing orders quietly, retail traders chasing green candles, panic, boredom, career risk, margin calls, algorithms, narratives, liquidity holes, and the occasional lunatic with size. This is the central problem of technical analysis. Its tools are useful enough to attract intelligent people and incomplete enough to ruin the people who demand more from them than they can give. A useless tool is easy to discard; a partially useful tool is far more dangerous. It wins your confidence in one regime, then invoices you in another.
Technical analysis is a two-sided blade. It can work. It can also cut the hand holding it. The better question is always narrower and less flattering: what does this tool measure, how late is the measurement, under what conditions does the measurement matter, and what does the tool fail to see? Signals give the illusion of obedience to something external. The trader can say the RSI told him to sell, the moving average told him to buy, the MACD told him momentum had shifted. This is how people outsource judgment while preserving the feeling of sophistication.
Jesse Livermore understood price better than almost anyone of his generation. In 1929, he made a fortune by reading what the market was doing before the crowd understood what it had already done. He studied price, volume, speed, hesitation, exhaustion, and the behavior of stocks at critical levels. He did this before modern terminals, before modern disclosure, before the thousand little luxuries that make contemporary traders feel informed while leaving them no wiser.
He saw distribution in the late stages of the 1920s bull market. He saw rallies reaching new highs with declining force. He saw violent reversals around important levels. He saw strong hands passing inventory to late buyers. He positioned himself for the crash and held through the violence that came before the reward. Livermore later lost his fortune. More than once. He died broke. The man who read price with extraordinary skill could not preserve the results of that skill. Most people fail to look at this side of the story and get enticed into joining day trading by hearing about his profits.
I start with the simple fact that markets are full of people who are right and bankrupt. This sounds paradoxical only to those who confuse analysis with trading. Analysis is the intellectual part. Trading is the behavioral part. The market pays for the second only when it is joined to the first. A trader can correctly identify the direction of the market and still lose money through poor sizing, bad timing, impatience, leverage, or the inability to sit through normal adverse movement.
Technical analysis was born from necessity. Charles Dow observed trends because price was one of the few public records available. Munehisa Homma studied rice prices and created candlestick methods because the open, high, low, and close revealed something about the psychology of the session. They were reading behavior through the evidence available to them. It was a way of extracting meaning from price when balance sheets were unreliable, economic data was delayed, and privileged information moved through private channels. Price had one advantage over almost every other source: it recorded what people had actually done with money. Technical analysis was created so that everything could be visible on the screen; so that everything could be countable and predicted.
The academic attack on technical analysis became powerful because much of retail technical analysis deserves to be attacked. Eugene Fama and the efficient market tradition challenged the idea that historical price data could be easily converted into excess returns. Many simple trading rules failed once they were tested properly. Moving average crossovers that looked elegant on finished charts became unimpressive after transaction costs, slippage, taxes, and bad fills. Patterns that looked obvious in textbooks became ambiguous in real time. Backtests produced miracles that vanished when exposed to future data.
The mathematics is unforgiving. Search long enough through randomness and randomness will reward you with patterns. It will even reward you with convincing ones. The trader then takes this accidental discovery into the market and learns the difference between a backtest and a wound. The efficient market people were right about many things. Simple rules are often arbitraged away. Historical price patterns are frequently overfit. Cost and execution matter; hindsight is a liar with excellent graphic design.
The simple truth is that every useful tool in markets carries its own poison. The moving average (a technical analysis tool that calculates the weighted or unweighted mean of n number of days) is the easiest example. It’s price smoothening is another example of the double edged blade we looked at a few paragraphs ago. Short moving averages with lower timeframes give faster signals, but there are more false buy/sell opurtunities. Longer moving averages bands provide us with more meaningful buy/sell orders, but they arrive later and infrequently.
Do not get me wrong here, I am not saying that moving averages don’t work, I’m saying they only work at times. In an upwards trend, a moving average can be magnificent. It provides meaningful buying opportunities, allows traders not to get faked out by small counter trends, etc. In a sideways market, the same moving average becomes a machine for humiliation. Price may cross above for the trader to buy. Then it crosses back below, prompting the trader to sell, then it crosses back above. The moving average did not fail in some mystical sense. It did what it was built to do. It summarized recent price. The trader failed by asking it to identify the regime in which its own signal deserved trust.
Take RSI (relative strength index) as another example. It suffers from a different abuse. The indicator measures the strength of recent gains relative to recent losses. A reading above seventy means recent buying pressure has been strong; it does not mean the asset must fall. A reading below thirty means recent selling pressure has been strong; it does not mean the asset must rise. Strong markets can remain overbought longer than a trader can remain solvent shorting them. Weak markets can remain oversold while bargain hunters are carried out one by one. A high RSI can indicate exhaustion; it can also indicate powerful demand. A low RSI can indicate panic near a bottom; it can also indicate the early stage of a collapse.
Finished charts are where these patterns look best. The completed chart is a courtroom in which the verdict has already been delivered. Live charts are different. The right side is blank. Relative strength index over 70 may stay over 70 for a longer period of time, and vice versa. This does not make indicator useless. It makes them conditional. A pattern must have invalidation. It must say where it is wrong. If the trader can redraw it endlessly, it is not analysis (it’s something else).
I understand that many of you may say,” But isn’t confluence of several indicators the way to go? If many indicators line up, then surely the chart pattern must resolve higher (or lower).” Sadly, the answer to this question is no (not always). Confluence is the next trap. RSI agrees with MACD. Price is near the moving average. Bollinger Bands are tightening. A Fibonacci level sits nearby. The trader feels protected by the crowd of confirmations.
Most of the times, the trader who has many indicators on his/her chart is reading the same thing many times. Many indicators are derived from price. RSI, MACD, moving averages, Bollinger Bands are different languages from the same ancestor. This creates a fallacy in which a trader feels that many indicators lining up prompts a buy or a sell. However, true confluence requires different kinds of evidence: price structure, volume, volatility, liquidity, catalyst, market regime, positioning. A crowded chart can create the emotional sensation of certainty without adding much information.
The serious trader uses technical analysis differently. He does not ask the indicator to tell him the future using the past. He asks what condition the indicator is measuring. He asks whether that condition has mattered in similar environments and how late the information being provided to him is. This last question separates trading from chart appreciation. A seventy percent chance of rain does not guarantee rain. If the sun appears, the forecast has not necessarily failed. The thirty percent outcome occurred. A good forecast must be judged over many instances, not by one emotionally convenient example.
A high-volume reversal at support can be a good trade and still fail. A bearish divergence can be valid and still arrive early. One trade proves almost nothing. The edge appears across repetition, under consistent rules, with survivable losses.
Technical analysis works sometimes because markets are made of behavior, and behavior leaves traces. It fails because traders confuse traces with destiny. Price can reveal pressure. Indicators can reveal trend, momentum, volatility, and exhaustion. They can help. They can even produce an edge. But every indicator is late, partial, and regime-dependent.
It’s ironic as I use technical analysis as my primary way of analyzing a chart. After some reflection, I realized that technical analysis uses the past to predict the future. Even after writing this, though, I will continue to use technical analysis as my way of analyzing a chart. As I continue to use technical analysis, and for everyone else who uses technical analysis, you have to understand that technical analysis is just a tool that measures a conditional. Every technical indicator is late, partial, and poisoned by its own limitations—yet it remains an indispensable weapon, because in the chaos of the markets, a flawed edge is the only difference between a trader and a gambler.