Overconfidence and the Cost of Overtrading

Ask a roomful of drivers whether they’re above-average behind the wheel and most hands go up — which is statistically impossible, but emotionally obvious. The same hand goes up in the investing world, where it’s far more expensive. Overconfidence is the bias that feels the best: it doesn’t sting like a loss or nag like regret, it just hums along quietly telling you that you’ve got an edge. And then it bills you for it — not in one dramatic blow-up, but in a slow drip of trading costs that turns a perfectly decent portfolio into a mediocre one. This lesson is about the bias that punishes you for acting, and why the cure is calibration, not timidity.

Guess first

Overconfidence comes in three flavours

“Overconfidence” sounds like one thing — thinking you’re great — but psychologists pulled it apart into three distinct mistakes, and they don’t always travel together. You can be humble about your raw ability while being wildly overconfident about how precise your forecasts are. Knowing which flavour is biting you matters, because each one warps a different investing decision.

Flavour one — overestimation. You overrate your absolute ability or your odds of success. The day-trader who thinks they’ll beat the market by 10% a year; the new investor sure their first stock pick will double. It’s a misjudgement of how good you are, full stop. Analogy: estimating you’ll finish the marathon in 3 hours when your training pace says 4:30.

Flavour two — overplacement (the better-than-average effect). You overrate yourself relative to others. This is the driving survey: the majority can’t be above average, yet the majority believe they are. In markets it’s lethal, because every trade has someone on the other side. If you’re buying because you’re sure the stock is going up, someone at least as informed is selling to you because they’re sure it’s going down. You can’t both be the smart money. Analogy: 80% of people rating themselves a better-than-average cook at the same potluck.

Flavour three — overprecision. You’re too certain your beliefs are right — your error bars are too narrow. You don’t just think the stock will rise; you’re sure it’ll rise between 8% and 12%, when the honest range is more like −20% to +40%. This is the most durable, best-documented flavour, and it gets its own section next.

When it matters

The three flavours hit different decisions. Overestimation makes you take positions that are too big for your actual edge. Overplacement makes you trade at all — every trade is a bet that you know something the person on the other side doesn’t. Overprecision makes you under-diversify and skip hedges, because if you know the range is narrow, why protect against a crash that “can’t” happen? Spotting which one is talking is the first step to muzzling it.

Overprecision — your 90% is really a 45%

Of the three, overprecision is the one with the cleanest, most embarrassing evidence — and the one that quietly underlies under-diversification and blow-ups.

The analogy. Imagine packing for a trip and being asked: “Give me a temperature range you’re 90% sure the weather will fall in.” A well-calibrated packer gives a wide range (so they’re rarely caught out) and is genuinely surprised only one trip in ten. An overprecise packer gives a tight, confident range — and gets soaked or frozen far more than one time in ten, because reality keeps falling outside their too-tidy guess.

The definition. A 90% confidence interval is a range you state such that you believe there’s a 90% chance the true answer lands inside it — and so a well-calibrated person should be wrong (truth falls outside) only about 10% of the time. Overprecision is the tendency to make those intervals far too narrow, so the truth escapes them far more often than it should.

The evidence. Across decades of studies (Alpert & Raiffa; Soll & Klayman and many others), when people give 90% intervals for unknown quantities, the truth lands inside only about 40–50% of the time — call it ~45%. Read that again: a range you’d swear has a 90% chance of containing the answer actually contains it less than half the time. Your “90% sure” is, in reality, closer to a coin flip.

The calibration chart below is the picture of this. The diagonal is perfect calibration — what you’d see if “90% sure” meant right 90% of the time. The real curve sags below it: the surer people feel, the bigger the gap between confidence and accuracy.

Worked example — the too-narrow interval

You’re 90% sure next year’s return on your favourite stock will land between +6% and +10%. That feels reasonable; that feels informed. But a single stock’s annual return has an enormous spread — historically a typical big stock can easily swing anywhere from −30% to +50% in a year. So your honest 90% interval should be something like −25% to +45%, perhaps fifteen times wider than the one you gave.

What does the tight interval cost you? It convinces you a 35% crash “can’t” happen, so you skip diversifying, you size the position too big, and you don’t build in any cushion. Then reality lands at −22% — comfortably inside the honest range, miles outside yours — and you’re blindsided by an outcome you should have called ordinary. Overprecision doesn’t just make you wrong; it makes you unprepared to be wrong.

Self-attribution and the illusion of control

If overconfidence were just a starting condition, experience would slowly cure it — you’d make forecasts, see them miss, and tighten up. Two mechanisms make sure that learning loop never closes.

Self-attribution bias. This is the habit of crediting your wins to skill and blaming your losses on luck (or the Fed, or a “rigged” market, or bad timing). The asymmetry is the whole problem: when you win you update toward “I’m good at this,” and when you lose you refuse to update at all. So a track record of 50% wins and 50% losses gets remembered as “mostly skill, with some bad breaks,” and your confidence ratchets up regardless of results. You’ve built a scoreboard that can only go up.

Analogy: a gambler who remembers every jackpot as proof of a “system” and every loss as the casino cheating. The system never gets falsified, because losses are never allowed to count as evidence against it.

Illusion of control. This is the sense that you can influence outcomes that are actually random or beyond you. The classic lab finding: people throw dice harder for high numbers and softer for low ones, as if effort bent the odds. In investing it shows up as the belief that watching the ticker more, reading more news, or clicking more buttons gives you more control over your returns. It doesn’t — the market doesn’t know you’re staring at it — but the activity feels like steering. More buttons, more dashboards, more trades: it all simulates control without delivering any.

Overtrading — the bill arrives

Put the pieces together — you overrate your edge (overestimation), you’re sure you know more than the person on the other side (overplacement), you’re too certain of your forecasts (overprecision), and you misremember your record as mostly skill (self-attribution) — and the behavioural output is a single, measurable habit: you trade too much. Overtrading is overconfidence made visible, and unlike a vague bias it leaves fingerprints all over your brokerage statement.

The evidence. Barber & Odean (2000), studying real US households at a discount broker, found that the most active quintile of traders earned net returns roughly 6.5 percentage points per year below the buy-and-hold market benchmark. The killer detail: their gross returns — before costs — were about the same as everyone else’s. They weren’t picking systematically worse stocks. The frantic trading itself, paid for in commissions and the bid–ask spread, drilled the hole. Trading was, in the paper’s words, hazardous to their wealth.

Their follow-up, “Boys Will Be Boys” (Barber & Odean, 2001), sharpened the point: men traded about 45% more than women and underperformed them by more — exactly what you’d predict if overconfidence (which surveys find runs higher in men, especially in finance) drives the trading. More confidence → more trades → more costs → worse net returns. The mechanism and the data line up.

A quick word on the hidden cost: the bid–ask spread is the small gap between the price you can buy at and the (lower) price you can sell at. Every round-trip trade — in and back out — pays that gap plus any commission. One trade, no big deal. Hundreds of trades a year, and the gaps compound into a serious drag, all of it invisible on any single ticket.

Worked example — turning 7% into less

Suppose your portfolio earns a 7% gross return in a year — that’s the return before any trading costs, the raw market gain on what you held. Now suppose your overconfidence has you churning the whole portfolio many times over, and each round-trip costs about 0.3% of the traded amount in commissions and spread. You turn the portfolio over roughly 8 times in the year (a very active, but real, level of churn).

Your total trading cost for the year is:

$\text{Cost} = 8 \times 0.3\% = 2.4\%$

So your net return — what actually lands in your account — is:

$\text{Net} = 7\% - 2.4\% = 4.6\%$

Your overconfidence just converted a respectable 7% into a mediocre 4.6%, handing 2.4 percentage points to your broker and the market-maker for the privilege of feeling busy. And we haven’t even counted taxes on the gains you realised early (a cost you met in Sunk Costs and the Disposition Effect). Crank the turnover higher and the gap easily reaches the Barber–Odean territory of ~6.5 points. Compound a 2.4-point annual leak over decades and it quietly devours a huge slice of your final wealth.

Same gross return, same starting stocks — the only difference is the trading, and it’s the trading that loses.

The fix is calibration, not cowardice

Here’s the subtle part the data is not telling you: be a coward. Some confidence is necessary to invest at all. If you waited for certainty you’d never buy anything, you’d sit in cash, and inflation would quietly eat you (the slow-motion error from Risk and Return). Acting under uncertainty requires enough conviction to pull the trigger. Zero confidence is its own failure mode.

The problem isn’t confidence — it’s mis-calibration: the gap between how sure you feel and how often you’re actually right. A well-calibrated investor can be confident and humble at once: confident enough to hold a thesis and act on it, humble enough to keep the error bars honest, diversify, and trade rarely. The villain of this lesson isn’t belief; it’s the too-narrow belief that masquerades as knowledge.

The whole picture

A mixed recap pulling from the whole lesson:

Key Takeaways