If past price movements don’t predict future ones, then every chart pattern, trend line, and “the stock is breaking out” alert is just noise dressed up as signal.
Imagine someone hands you a chart of a stock’s daily closing price over the past six months. The line zigzags across the page. There are a few peaks, a few valleys. The person points to a pattern and says, “See this shape? It’s called a head and shoulders. The stock is about to fall. I’m selling.”
That kind of analysis has a name: technical analysis — the practice of predicting where a stock’s price is going by studying where it has been. Charts, trend lines, pattern names, buy and sell signals. There are entire books written about it. There are software platforms built around it. Millions of investors use it every day.
And yet, one of the most careful, rigorous bodies of research in all of finance suggests that it doesn’t work.
Not “doesn’t work as well as other methods.” Doesn’t work at all — not better than chance, and not consistently, not for long.
That’s the argument we’re going to examine in this post.
A Quick Introduction to Our Guide
Before we go further, I want to introduce the book that this series is based on: A Random Walk Down Wall Street by Burton G. Malkiel.
Malkiel is a professor of economics at Princeton University and one of the world’s most influential advocates for the efficient market hypothesis and index investing. He first published A Random Walk Down Wall Street in 1973. It is now in its 13th edition — a sign of how durable the argument has proven. The book is not dense academic writing. It is clear, direct, and built around evidence.
We did not introduce this book in Part 1 because the opening post was focused on setting up the core ideas — the random walk, the efficient market hypothesis, and why they matter. But Malkiel is the architect of much of what we’ll cover in this series, and it’s time to name him directly. When this series references claims about market efficiency or technical analysis, those claims come from decades of research that Malkiel synthesized, tested, and explained better than anyone.
Now: back to those charts.
What Technical Analysis Actually Assumes
To understand why technical analysis fails, you first have to understand what it’s claiming to do.
A technical analyst — or chartist, as they’re sometimes called — believes that the history of a stock’s price movements contains useful information about its future movements. Not about the company itself. Not about its earnings or products or management. Just the price history.
The specific claim is that patterns repeat. That a “head and shoulders” formation, or a “double bottom,” or a “golden cross” (two moving averages crossing each other at a specific angle) has predictive power. That the stock will behave in a certain way after this pattern appears because it has behaved that way before.
This is a bold claim. And it runs directly into a problem.
The Weak Form — and Why It Kills the Chart
In Part 1, we introduced the Efficient Market Hypothesis. Economist Eugene Fama defined it in three forms. The first — and the one most directly relevant here — is called the Weak Form.
The Weak Form says this: all information contained in past prices is already reflected in the current price.
Read that again. Every piece of price history that anyone could possibly use — yesterday’s close, last month’s trend, the head-and-shoulders pattern that formed over the past six weeks — has already been seen by thousands of traders. They have already acted on it. Their buying and selling have already pushed the current price to reflect what that information implies.
So when you study a chart and find a pattern, you are not discovering hidden information. You are looking at information that has already been fully processed. The ship has already sailed.
Here’s another way to think about it. Suppose it were genuinely true that after a head-and-shoulders pattern, a stock reliably falls 10% over the next month. Every chartist would know this. They would all sell before the fall. Their selling would drive the price down immediately, not a month later. The very act of everyone using the pattern would destroy the pattern.
This is the fundamental problem with technical analysis: if a price pattern were real and widely known, it would immediately be traded away.
What the Research Actually Shows
This isn’t just theory. Researchers have tested specific technical patterns on real market data. Repeatedly. Over long time periods. Across different markets and different countries.
The results have been consistently unimpressive.
Moving averages are one of the most popular tools in technical analysis. A moving average is simply the average price of a stock over a recent period — say, the last 50 days. Chartists watch for the moment when a short-term moving average (say, 50 days) crosses above a long-term moving average (200 days) — a signal called a “golden cross.” They believe this signals a rising trend.
When researchers tested this signal on actual market data — after accounting for trading costs and taxes — they found that it did not outperform simply holding the stock. In many cases, it underperformed.
Momentum signals are another popular tool. The idea is that stocks that have gone up recently tend to keep going up. And indeed, there is a documented short-term momentum effect in financial markets. But here’s the problem: once you subtract trading costs — every time you buy or sell, you pay a small fee — the advantage disappears. And momentum signals fail catastrophically during market reversals. The strategy that worked fine in a rising market suddenly produces large losses when conditions change.
Pattern recognition — head and shoulders, double bottoms, flags, pennants — has been studied extensively. Malkiel summarizes the findings bluntly: researchers who have applied rigorous statistical tests to these patterns have found no consistent predictive power. When tested out of sample — on data that wasn’t used to identify the pattern in the first place — the results are no better than random.
The Chart Looks Like Signal. It’s Noise.
Here is the uncomfortable truth about charts: the human brain is very good at finding patterns. Frighteningly good. We see faces in clouds, animals in rock formations, trends in random data.
When you look at six months of price data, your brain will find patterns. It will find “head and shoulders.” It will find “the stock always bounces off $50.” It will construct a narrative.
But the stock doesn’t know $50 is supposed to be important. The price history has no memory. Each day’s movement, like the coin flip in Part 1, is influenced by new information — not by where the price was last Tuesday.
Malkiel describes an experiment in his book. He asked students to simulate a stock price by flipping a coin repeatedly. Heads meant the price went up by a fixed amount; tails meant it went down. Over time, the random sequence of heads and tails produced charts that looked remarkably like real stock charts — with apparent trends, patterns, and turning points that a chartist might interpret as meaningful signals.
Those patterns meant nothing. They were artifacts of randomness.
Real stock price movements are not purely random — companies release earnings, executives make decisions, economic conditions change. But the part of real price movements that is driven by public information that everyone can already see? That part has already been priced in. What’s left is close enough to random that chart patterns cannot consistently extract useful predictions from it.
One Place the Research Is More Complicated
There is a nuance worth mentioning honestly: short-term momentum.
Studies have found that stocks that have performed well over the past three to twelve months tend to continue outperforming slightly over the next few months. This is one of the most replicated findings in finance. It seems to challenge the efficient market view.
But several things temper this finding:
1. The effect is small and degrades quickly once transaction costs are included.
2. Momentum strategies fail badly during market downturns — they tend to buy things that are already extended and sell things that are already beaten down, right before reversals.
3. As momentum strategies have become more widely known and traded, the advantage has largely been arbitraged away.
This is exactly what the efficient market theory predicts: any exploitable pattern, once widely known, gets traded away. The momentum effect has been partially documented, partially exploited, and partially eliminated — all in sequence, just as the theory says it should be.
A Note to Luca and Lili
You are going to encounter charts. Software platforms will show you “signals” and “trend reversals.” Financial news channels will speak confidently about stocks that are “testing support” or “breaking resistance.”
None of this is necessarily dishonest. Many people genuinely believe in it. And the patterns look real — our brains are built to see them.
But the evidence is clear: studying charts does not give you a consistent, tradeable edge after costs. If it did, the edge would be competed away by everyone else using the same charts.
Your grandfather spent his career looking for genuine information advantages — things the market had not already priced in. That means going deeper than charts. It means understanding a business, estimating its cash flows, and being patient when the market prices it wrong. That is harder than following a signal. It is also more defensible.
We’ll keep building that case as this series unfolds.
— Papa
The One-Sentence Summary
The Weak Form of the Efficient Market Hypothesis holds that all information contained in past prices is already reflected in today’s price — which means chart patterns and technical signals cannot consistently predict future price movements any better than chance, a conclusion supported by decades of empirical research and confirmed by the self-defeating nature of any widely known pattern.
Next: Efficient Markets, Part 3 — Why Fundamental Analysis Is Harder Than It Looks. Technical analysis studies price history. Fundamental analysis studies the company — its earnings, its growth, its assets. That sounds like a stronger foundation. But efficient markets have a challenge for fundamental analysts too, and it’s one worth understanding before you believe any analysis is truly giving you an edge.
— Jim