Overconfidence Bias: Why Feeling Sure You're Right Makes You Trade More and Earn Less
Overconfidence bias is the tendency to overestimate your own skill at picking winners and timing markets — and the largest study ever done on it found the most confident, most active traders earned the worst returns. What the research found, why crypto makes it worse, and how to check your own numbers.
Overconfidence bias is the tendency to overestimate your own skill, knowledge, or ability to predict what a market will do next — and in investing, it has a specific, measurable signature: it makes people trade more often than the evidence justifies. The largest study ever done on individual investor behavior found that the most active traders, not the most patient ones, earned the worst returns, and the gap wasn’t small. If you’ve ever felt certain enough about a call to size up beyond your normal rules, or found yourself checking charts every few minutes convinced you could catch the next move, you’ve felt this bias operating in real time.
It matters more than most biases covered in this space because it doesn’t feel like a mistake while it’s happening — it feels like insight. Every other bias on this site distorts a decision; overconfidence distorts your estimate of how good your decisions are, which is why it’s so hard to catch from the inside.
The study behind it
The most cited empirical test is Brad Barber and Terrance Odean’s 2000 paper, “Trading Is Hazardous to Your Wealth,” published in The Journal of Finance. They analyzed real trading records from 66,465 households at a large U.S. discount brokerage between 1991 and 1996 — the average household turned over about 75% of its common-stock portfolio per year. The households that traded the most earned an average annual return of 11.4%, while the market itself returned 17.9% over the same period; the average household across the full sample earned 16.4%. Trading costs — commissions and the bid-ask spread — ate the difference. The paper’s own framing of the mechanism is direct: overconfident investors mistake chance successes for skill, trade on that false signal, and pay for it in fees they didn’t need to generate.
A follow-up paper by the same authors, “Boys Will Be Boys: Gender, Overconfidence, and Common Stock Investment,” published in 2001 in The Quarterly Journal of Economics, used the same brokerage-record approach across roughly 35,000 households from February 1991 through January 1997, cross-referenced against psychology research showing men, on average, report higher confidence than women in finance-related judgments. The trading data lined up with that confidence gap almost exactly: men traded 45% more than women, and single men traded 67% more than single women. Trading reduced men’s net annual returns by 2.65 percentage points on average, versus 1.72 percentage points for women — a difference of about 1 percentage point a year, driven entirely by how much more the more-confident group traded, not by which stocks either group picked.
Neither paper is about who’s a better investor. They’re both about the same mechanical chain: higher confidence in your own judgment produces more trades, and more trades produce more costs, and those costs are what actually shows up as a lower number in your account — regardless of whether any individual trade in the sequence was a good one.
Crypto isn’t an exception — if anything, it’s a better environment for this
A study examining who invests in cryptocurrency among American investors found that investors who scored higher on overconfidence measures were meaningfully more likely to hold crypto — on the order of 8 percentage points more likely to be a crypto owner, and roughly 10 points more likely to be a probable future one. Crypto’s structure feeds the bias in ways a stock account with market-hours trading doesn’t: a market that never closes gives you far more opportunities to act on a feeling, price swings large enough that a lucky short-term call feels like proof of skill, and a social feed built around people loudly claiming they called the move, which makes your own correct guesses feel more meaningful by comparison than a quieter, more boring stock ticker ever would.
None of that changes the underlying mechanism from the Barber and Odean research. It just means the same trade-more-because-you-feel-sure loop has more hours in the day, more volatility to misread as signal, and more social reinforcement to run on.
Why it shows up hardest exactly when you’re recovering
Overconfidence rarely peaks at the top of a bull run when everyone feels smart together — it tends to spike for one specific person at one specific moment: right after they’ve clawed back part of a loss. A drawdown followed by even a modest bounce creates a tidy personal narrative — “I held through the bottom, I bought the dip, I called it” — that reads as evidence of skill, even when the bounce was broad market movement that would have lifted almost any position you were still holding. That’s the same psychological terrain covered in House Money Effect vs. Break-Even Effect: the closer you get back to even, the more that recovered confidence pushes you toward a bigger, faster move instead of a more careful one, and a string of correct short-term calls during the recovery is exactly the kind of “chance success mistaken for skill” the original 2000 study describes.
It’s also one of the quieter engines behind revenge trading, covered in 5 Signs You’re Revenge Trading — the size-up after a loss usually comes wrapped in a story about a specific insight you supposedly have this time, not a straightforward admission that you’re trying to make the loss back faster. Overconfidence is often what supplies that story.
A worked example
Say your portfolio is up 22% over the past twelve months, and you feel that reflects real skill — good enough to justify sizing up your next position beyond your usual cap. Before acting on that feeling, run two checks the research above suggests: First, what would a simple, low-cost index fund tracking the same broad market have returned over the identical period? If it also returned somewhere near 22%, your result is telling you about the market’s direction, not your stock- or coin-picking ability. Second, count your actual trades over that window and estimate the total cost — spread, fees, any short-term tax drag from frequent selling. If a buy-and-hold version of the same starting capital, held untouched, would have outperformed your actively-traded result once those costs are subtracted, the trading itself was a net negative, dressed up as the thing that made you money.
Neither check tells you to stop trading. It tells you whether the specific confidence driving your next decision is backed by a result that survives comparison to a passive baseline — which is the same standard the professional research above is built on.
What overconfidence actually costs you
| Where it shows up | What the bias makes you do | What’s actually optimal |
|---|---|---|
| Trading frequency | More trades than your thesis actually requires, each one adding cost | Trade when the thesis changes, not when confidence spikes |
| Position sizing | Sizing up after a string of wins, treating recent luck as proven skill | Size based on a pre-set rule, independent of your last few results |
| Attribution | Crediting wins to skill, blaming losses on bad luck or manipulation | Evaluate wins and losses by the same process standard |
| Performance tracking | Remembering the calls that worked, forgetting the ones that didn’t | Compare actual net returns to a passive benchmark, in writing |
A five-minute self-audit
- Pull your total account value from exactly one year ago, and today.
- Calculate your actual return, net of every fee, spread, and any tax drag from short-term selling.
- Look up how a broad index fund covering the same asset class performed over the identical window.
- If your actively-managed number is close to or below the passive number, the trading itself didn’t add value — the market’s direction did the work you’re crediting yourself for.
- Count how many of your last 20 trades you can specifically remember losing on, without checking your records. If you can name most of your wins but had to go look up the losses, that’s the selective memory that keeps overconfidence funded.
If the honest answer is that a passive benchmark would have beaten your actual result, that’s not a reason for shame — it’s the exact finding the 2000 study describes, at the scale of one account instead of 66,465. The fix isn’t to stop having opinions about the market; it’s to stop letting a recent string of being right change how much you’re willing to risk on the next one.
FAQ
Isn’t confidence just a good thing? Don’t I need conviction to hold a position through volatility? Genuine conviction and overconfidence look similar from the inside but come from different places. Conviction is holding a position because you’ve stated a specific, falsifiable thesis and it still holds up under scrutiny. Overconfidence is overestimating how precisely you know something — how sure you are, not what you actually know. The test used elsewhere on this site works here too: can you state your thesis and what would prove it wrong, in plain terms, without referencing your own track record or gut feeling? If the honest answer is “I’ve just been right a lot lately,” that’s overconfidence borrowing conviction’s clothing.
Doesn’t more experience fix this over time? Not automatically, and this is one of the more uncomfortable findings in the research. Barber and Odean’s data span years, not a single lucky quarter, and the pattern of high turnover producing lower net returns held up across that whole window rather than fading as investors gained more trades under their belt. More trades mainly generate more raw material for selective memory — you remember the calls that worked and quietly reclassify the ones that didn’t as bad luck, external manipulation, or someone else’s fault, which is a separate, compounding bias (self-attribution bias) that keeps overconfidence topped up rather than correcting it.
Is the gender research saying men are worse investors than women? It’s describing one large, specific dataset — U.S. discount-brokerage accounts from February 1991 through January 1997 — not making a claim about any individual, in either direction. Plenty of individual men trade carefully and plenty of individual women trade recklessly; the study measured averages across roughly 35,000 households, not any one person’s behavior. What makes the finding useful here isn’t who it’s about — it’s that it’s one of the cleanest natural tests of the actual causal chain (higher self-assessed confidence in an area → more trading → higher costs → lower net returns), because it uses a trait that’s observable and correlated with the confidence measures researchers separately test, on top of the trading records themselves. The mechanism is what to take from it, not a verdict on either group.
How is this different from the gambler’s fallacy or the house money effect covered elsewhere on this site? They’re related but distinct failure points. The gambler’s fallacy is a specific misjudgment about probability — believing a losing or winning streak changes the odds of what comes next. The house money effect is about how a recent gain changes your appetite for risk with that specific pool of money. Overconfidence is broader than either: it’s a general overestimation of your own skill, knowledge, or precision, and it’s often the thing feeding both of the other two — you take the gambler’s fallacy bet or the house-money-fueled risk because some part of you believes you’re reading the situation more accurately than you actually are.