Do Chart Patterns Actually Work? What the Research Shows

Few questions in investing are as argued over as this one: do chart patterns such as the head and shoulders or the cup and handle actually tell you anything about future prices?

The honest summary is that the evidence is mixed, modest, and hard to interpret — and anyone quoting a precise "success rate" for a pattern deserves some skepticism about where it came from.

The Case Against

The traditional academic view comes from the efficient market hypothesis: if past prices contained reliable, easy-to-exploit information about future prices, traders would act on it until the opportunity vanished. In its weak form, the hypothesis says that patterns in past prices should not predict future returns after costs.

Researchers have also pointed to several practical problems with pattern claims:

• Subjectivity. Two analysts can look at the same chart and disagree about whether a pattern exists, where its lines go, and when it was confirmed. A rule that can't be defined precisely can't be tested precisely.
• Hindsight bias. Patterns are far easier to see after the fact. Charts shown in books and articles are usually the ones where the pattern worked, which is a form of survivorship bias.
• Data snooping. Test enough patterns, parameters, and time periods and some will look profitable by chance. A pattern that "worked" in a backtest chosen after the fact may not have been a real effect.
• Transaction costs. Even a small statistical edge can disappear after spreads, commissions, and taxes.

The Case For (Or At Least, Not Zero)

The evidence isn't entirely one-sided:

• In a widely cited 2000 paper in the Journal of Finance, Andrew Lo, Harry Mamaysky, and Jiang Wang used a computer algorithm to identify patterns — including head and shoulders and double tops and bottoms — in US stock data. They found that several patterns appeared to provide incremental information, though the effects were modest and varied by stock group and period.
• A 2007 survey by Cheol-Ho Park and Scott Irwin reviewed ninety-five studies of technical trading strategies and found that a majority reported some profitable results, but also noted that many studies were weakened by data-snooping and risk-adjustment problems. Results also tended to be stronger in earlier periods than later ones.
• Research by Carol Osler on foreign exchange markets found that published support and resistance levels were associated with clusters of orders, suggesting that some levels can matter because many participants act at them. That is a mechanism, not proof that traders can profit from it.

The pattern across this literature: some statistically detectable effects appear in some markets and eras, but they are modest, they weaken over time, and they are hard to turn into reliable after-cost profits.

The Self-Fulfilling Prophecy Argument

A popular defense of chart patterns is that they work because many people watch them. If thousands of traders buy when price breaks above a resistance level, the buying itself can push price higher.

There is something to this, particularly at widely watched levels. But the argument cuts both ways:

• It would only apply to patterns that most participants see and agree on. Most patterns are more ambiguous than that.
• If everyone acts on the same signal, early movers can front-run it, and the signal may be priced in before the "official" breakout. This is part of why false breakouts are common.
• It implies that patterns could stop working once they become too popular.

What "Reliability" Claims Usually Leave Out

If you see a statement like "pattern X works 75% of the time," ask:

1. How was the pattern defined? By an algorithm or by a person picking examples?
2. What counts as "working"? Reaching a target? Any positive return? Over what period?
3. What sample and time period? Does it hold out-of-sample, or only in the data that was used to find it?
4. After costs? And compared to simply holding the stock?
5. Who is publishing it? Is someone selling a course or a service?

Without answers, a headline success rate is a marketing claim, not evidence.

How Patterns Are Better Thought Of

A more defensible way to use chart patterns is as a framework for describing price behavior and defining risk, rather than as a prediction machine:

• They give a common vocabulary for what a chart has done.
• They give a specific price level (a neckline, a trendline, a resistance level) where the "story" of the pattern is proven wrong, which helps define how much to risk.
• They are probabilistic, and a pattern failing is a normal outcome, not an anomaly.

That is a very different claim from "this pattern tells you where price is going."

Honest Takeaways

• The research does not show that chart patterns are reliable money machines.
• It also does not show they carry zero information; some studies find small, inconsistent effects.
• The reliability numbers you see online are rarely backed by transparent methods.
• Patterns are subjective, so your results depend on how you define and apply them.

For how specific patterns are read, see our guides to triangles, flags and pennants, and support and resistance.

This article is for educational purposes only and is not investment advice. Past patterns do not guarantee future results.