Why a Few Chart Patterns Can Work Even If Markets Are Mostly Rational
Technical analysis, predicting future price movement from past price and volume charts, sits uneasily next to the idea that markets are efficient. Behavioral finance offers a plausible bridge between the two: if certain biases are shared across many investors and slow to correct, they can leave faint but statistically real footprints in price data, footprints that shrink dramatically once real costs are applied.
The mechanism: underreaction, overreaction, and momentum
Pure efficient-market theory predicts that technical analysis should not work at all, since any exploitable pattern in past prices ought to be arbitraged away the moment enough traders notice it. Yet a small number of technical effects, most notably momentum, the tendency for securities that have performed well over the past three to twelve months to modestly continue outperforming over the next few months, have been documented repeatedly enough, across enough independent markets and time periods, that dismissing them as pure data-mining artifacts has become difficult to defend. Behavioral finance supplies a coherent explanation for why this specific pattern, among the thousands of chart patterns technical analysts claim to have found, is one of the few with real statistical staying power.
The proposed mechanism runs through two related, well-documented biases operating on different timescales. Over shorter horizons, investors and even professional analysts tend to underreact to new information, adjusting their expectations toward good or bad news more slowly and incompletely than a fully rational, immediate updating process would predict, a pattern consistent with anchoring on prior beliefs. This gradual underreaction means a stock's price continues drifting in the direction of genuinely good or bad news for weeks or months after the news first arrives, rather than jumping immediately to its new fair value, which is exactly what produces a momentum effect in the data. Over longer horizons, three to five years, the opposite bias, overreaction, tends to dominate: investors extrapolate a long run of good or bad results too far into the future, pushing prices further from fundamental value than is justified, a pattern that helps explain the separate, longer-horizon tendency for extreme past winners and losers to partially reverse.
A useful way to picture underreaction concretely is to imagine a company reporting earnings meaningfully above expectations. A fully rational market should immediately reprice the stock to reflect the entire implication of that surprise, all at once, on the announcement day. What the evidence actually shows, discussed in more depth in the article on event studies elsewhere on this site, is that a portion of the price adjustment leaks out gradually over the following weeks rather than completing immediately, a pattern known as post-earnings-announcement drift and one of the clearest, most direct pieces of evidence for underreaction as a real, measurable phenomenon rather than a purely theoretical possibility. Momentum, in this framing, is close to the aggregate, portfolio-level echo of many individual instances of this same slow-drift pattern playing out across thousands of securities simultaneously.
The math: what survives after costs, and the crash risk
Worked example 1: how much of the theoretical momentum edge survives trading costs and taxes. Suppose historical testing shows stocks in the top decile of trailing six-to-twelve-month performance outperform the bottom decile by an average of 4% over the following year, before any costs, on a $100,000 portfolio, an annual gross edge of $100,000 x 4% = $4,000. Capturing this in practice requires quarterly rebalancing to keep rotating into current winners, four rebalances a year at a round-trip trading cost of 0.25% each, a total annual drag of 0.25% x 4 = 1.0%, or $100,000 x 1.0% = $1,000. Because the strategy trades frequently, its gains are realized as short-term rather than long-term capital gains; assume the difference between the investor's ordinary marginal tax rate and the long-term capital gains rate they would otherwise pay is 17 percentage points, and that the full $4,000 gross edge is realized as short-term gain: the incremental tax cost versus a buy-and-hold approach is 17% x $4,000 = $680. The net surviving edge is $4,000 - $1,000 - $680 = $2,320, or 2.32% of the portfolio, meaning roughly 58% of the theoretical gross edge survives once realistic trading costs and the incremental tax cost of frequent trading are subtracted, and this figure does not yet include the added risk discussed in the next example.
Worked example 2: what a momentum crash actually looks like in dollars. Momentum strategies are typically implemented, in their purest academic form, as long-short portfolios: long recent winners, short recent losers, in roughly equal dollar amounts, to isolate the momentum effect from the market's overall direction. Consider $50,000 long recent winners and $50,000 short recent losers, $100,000 of gross exposure. During a sharp, sudden market reversal, a pattern documented repeatedly following steep market declines, previously beaten-down stocks, the ones held short, can rally very sharply as the broad market recovers, while the prior winners, now comparatively expensive and often lower-beta, lag behind. Suppose the shorted losers rally 25% while the long winners rise only 5% over the same short window. The loss on the short leg is -25% x $50,000 = -$12,500 (a loss, because the shorted stocks rose in value), while the gain on the long leg is 5% x $50,000 = $2,500. Net portfolio result: $2,500 - $12,500 = -$10,000, a 10% loss on $100,000 of gross exposure in a single short window, concentrated specifically around the kind of sharp reversal that momentum, by construction, is poorly positioned to handle.
It is useful to extend worked example 2 with the perspective of an investor who held the position through the entire episode rather than exiting at the trough, since a crash's severity is often only half the story. Suppose the reversal described above unfolds over one month, and over the following months the strategy recovers at its normal, modest historical pace, averaging roughly 0.4% net of costs per month based on worked example 1's annual figures. Recovering the $10,000 drawdown at that pace, applied to the reduced $90,000 base remaining after the crash, requires $10,000 / ($90,000 x 0.004) = $10,000 / $360 ≈ 28 months, well over two years, illustrating why a single sharp momentum crash can erase several years' worth of the strategy's typically modest annual edge, an asymmetry a simple average annual return figure completely fails to convey.
What the evidence shows across markets
Momentum has been documented not just in U.S. equities but across international stock markets, across commodity futures, across currencies, and across corporate bonds, a breadth of replication that is unusual among claimed market anomalies and is one of the stronger arguments that the effect reflects a genuine, if modest, behavioral pattern rather than a statistical artifact confined to one dataset. The long-horizon reversal effect, the tendency for extreme three-to-five-year winners and losers to partially revert, has a similarly broad replication record, though the two effects, momentum over months and reversal over years, coexist without contradiction precisely because they operate on the different timescales described above.
The crash risk illustrated in worked example 2 is itself a well-studied, documented feature of momentum, not a hypothetical worst case invented for this article. Momentum strategies have shown a repeated tendency to suffer their sharpest, most concentrated losses specifically during rapid market rebounds following steep declines, precisely when previously depressed, high-beta stocks recover fastest and prior winners, having become comparatively expensive and lower-risk, lag. This asymmetric crash risk is a large part of why momentum's headline average return overstates how attractive the strategy actually is on a risk-adjusted basis: an investor evaluating only the average annual return, without separately weighing the risk of a sharp, concentrated loss arriving at an unpredictable and inconvenient time, is working from an incomplete picture. Beyond momentum, the broader universe of classical technical chart patterns, head-and-shoulders formations, double tops, specific candlestick shapes, has fared far worse under the same rigorous testing standards, generally failing to show statistically reliable predictive power once researchers control for the fact that any sufficiently large dataset will produce some patterns that appear meaningful purely by chance.
Using this responsibly in a real portfolio
For most individual investors, the practical lesson is not to build a manual, chart-watching trading strategy around momentum, since the version of momentum with genuine, broadly replicated evidence behind it is a systematic, rules-based factor applied across a diversified basket of hundreds of securities, not a discretionary judgment about whether one particular stock's chart looks like it is trending. Low-cost momentum factor funds and exchange-traded funds now offer a way to access a diversified, systematically implemented version of this exposure without the manual trading costs and behavioral pitfalls of running it by hand, though even these funds carry the crash risk illustrated in worked example 2 and should be sized as a modest satellite allocation rather than a portfolio's core.
For an investor drawn to discretionary technical trading specifically, the honest framing is that the vast majority of individually chosen chart patterns have not held up under careful testing, and that the one pattern with genuine evidence behind it, momentum, only survives in a systematic, diversified, cost-controlled implementation, not in the hand-picked, single-stock version most retail technical trading actually resembles. A useful personal test before acting on any chart-based signal is to ask whether the specific pattern has been documented across many independent securities and time periods in a way resembling worked example 1's decile-spread framework, or whether it is simply a pattern that looked compelling on one particular chart after the fact.
Professionals with concentrated stock compensation, employer shares accumulated through vesting schedules or option grants, face a related and worth-naming variant of this same behavioral pattern, even if they never trade a chart in their life. A stock that has recently performed very well can trigger the same underreaction-driven momentum dynamic discussed above, giving a false sense that continued strong performance is the reliable, expected outcome rather than one possibility among several, precisely at the moment concentration risk in that single holding is often at its highest as a share of total net worth. Recognizing momentum as a modest, well-documented, but genuinely risky pattern, rather than as evidence that a recent winner is a safe bet to keep holding, is a useful mental check before deciding whether to diversify out of a concentrated position.
Actionable breakdown
- Treat momentum as documented but modest, cost-sensitive, and crash-prone.
- Access momentum through diversified, systematic funds, not manual picks.
- Size any momentum allocation as a small satellite, not a portfolio's core.
- Be skeptical of chart patterns without broad, independent replication.
- Account for trading costs and short-term tax drag before trusting a backtest.
Common pitfalls
- Confusing a pattern that worked once with a statistically reliable effect.
- Ignoring how transaction costs and taxes erode small, frequent trading edges.
- Evaluating momentum's average return without weighing its crash risk.
- Assuming a documented anomaly stays exploitable once it becomes widely known.
The bottom line
Behavioral biases give momentum a genuine, well-replicated statistical edge, but trading costs, taxes, and sharp, unpredictable crash risk consume enough of that edge that it belongs in a diversified, systematic sleeve of a portfolio, not a hand-picked chart-trading strategy.
All articles · The behavioral critique · Behavioral finance · Momentum · The Fama-French three-factor model