Why Traders Overtrade, and What It Costs

The short answer

Overtrading is trading more often than your edge can pay for. It is usually written about as an emotional problem — impatience, boredom, the need to be doing something — and the emotional account may well be true. But it is not what makes it expensive, and it is not what the evidence is about.

What the evidence is about is a bill. Every trade pays a spread, usually a commission, sometimes financing, and in some markets a transaction tax. Those are charged per decision, not per year. Being right about direction has to cover them before it covers anything else. Double the number of decisions and you double the bill. You do not double the quality of your reasoning.

Two studies have measured what that costs across entire populations of investors, and the answer in both is: more than most people would guess, and in a way that has very little to do with picking badly.

What was measured

Sixty-six thousand households, and the difference frequency made

Barber and Odean examined the common stock accounts of 66,465 households at a large American discount broker over the six years ending January 1997. The average household turned over more than 75 per cent of its portfolio a year and earned an annual return of 16.4 per cent, against a market return of 17.9 per cent.

Sorted into five groups by how much they traded, the picture sharpens considerably:

Group	Net annual return
Highest turnover fifth	11.4%
The market	17.9%
Lowest turnover fifth	18.5%
Average household	16.4%

The households that traded least matched the market. The households that traded most gave up something close to six and a half percentage points a year relative to them. Same broker, same market, same period. The variable is frequency.

A whole market, and where the money actually went

The second measurement is the one that settles what the cost is made of. Barber, Lee, Liu and Odean used the complete trading history of every investor in Taiwan from the start of 1995 to the end of 1999 — not a sample, the whole market. The aggregate portfolio of individual investors carried an annual performance penalty of 3.8 percentage points, which the authors put at 2.2 per cent of Taiwan's gross domestic product, or 2.8 per cent of total personal income.

Then they broke the losses down. Four categories:

• trading losses — being on the wrong side — 27%
• commissions — 32%
• transaction taxes — 34%
• market-timing losses — 7%

Read that twice. Roughly two-thirds of what individual investors lost was not selection at all. It was the cost of transacting. Over the same period, institutions in the same market earned an annual performance boost of 1.5 percentage points, after their own commissions and taxes.

This is why "work on your entries" is such poor advice to give someone who is overtrading. The entries were a minority of the problem.

Why people do it

The academic answer is overconfidence, and it was tested in an unusually clean way.

Theoretical models predict that overconfident investors trade excessively. Barber and Odean needed a group that differed in overconfidence for reasons unrelated to trading, so they split the account data by gender, on the basis of psychological research showing that men are more overconfident than women in domains such as finance. Across more than 35,000 households from February 1991 to January 1997, men traded 45 per cent more than women, and trading reduced men's net returns by 2.65 percentage points a year against 1.72 points for women. Among single account holders, where the attribution is cleanest, single men traded 67 per cent more than single women and gave up a further 1.44 percentage points a year.

That design does not prove overconfidence causes overtrading in any individual. What it does is make the prediction risky — a theory that said nothing about who trades more would not have anticipated the split — and the prediction held.

Note what the explanation is not. It is not a claim about greed, boredom or thrill-seeking. Those may be real, but they were not what was tested. There is a difference between an explanation and a name, which is the subject of why "fear and greed" explains nothing.

What overtrading looks like from the inside

It very rarely feels like overtrading. It feels like a series of individually reasonable decisions:

• The setup you waited for did not appear, so you took a worse one rather than nothing.
• A position closed at a small loss and the reason for it still looks valid, so you re-entered.
• The market moved without you and getting in late felt better than missing it entirely.
• You were already watching, and watching without acting is uncomfortable.
• You widened what counts as a setup, gradually, without ever deciding to.

None of those is an emotional collapse. Each is a small relaxation of a rule, and collectively they are the entire mechanism. This is why the counter is not composure. It is a written definition of what you will act on, made when nothing is at stake, which is the subject of what a plan has to settle in advance.

What to actually count

The useful thing about frequency is that, unlike a mood, it is a number you already have. Every platform will export it. Four counts are worth keeping:

1. Trades per week, plotted over time. Not an average — the shape. Most overtrading is episodic and clusters after losses.
2. The share that met your own written criteria. This requires having written them down before, which is the whole argument for recording the decision rather than the outcome.
3. Total transaction costs as a share of the result. The Taiwanese breakdown is a warning about what this ratio can look like without anyone noticing.
4. Trades taken within an hour of closing a loser. That cluster has its own literature, covered in what happens after a loss.

A worked illustration, with hypothetical figures, of why the third one matters: suppose a hypothetical account pays a round-trip cost of 0.1 per cent of position value and trades twice a week, so roughly 100 round trips a year. The cost line is then about 10 per cent of position value annually before anything else happens. Raise it to four trades a week and that doubles. Nothing in that arithmetic depends on being right, and nothing about being right makes it go away. These numbers are illustrative only; your own costs depend entirely on your broker and market.

The honest limits of this evidence

• Both studies are of equity investors — American discount-brokerage households and the Taiwanese stock market. They are not studies of leveraged retail products, where costs work differently and are usually higher.
• They are old: 1991–1996 and 1995–1999. Commissions have fallen a great deal since. The transaction-tax component in particular is specific to that market.
• They describe populations, not people. Some households in the high-turnover group did well. The finding is about the average, and about the direction of the relationship.
• Lower commissions do not remove the result, because the spread and the selection effects remain, but they do change its size, and we have not seen an equivalent whole-market study on recent data.

What survives all of those caveats is the structure of the thing: a cost charged per decision, paid by a decision process that is not reliably better than the market it is betting against. That structure is what the wider evidence on trader behaviour keeps returning to, and it is why the first question this publication asks about any trading habit is how often it makes you decide.

Nothing here is advice, and no figure above is a prediction about any individual account.