Overconfidence and overtrading

Ask a room of drivers whether they are above average and most hands go up. Ask a room of traders the same question and the result is the same, except that in trading, the belief has a direct cost: it makes you trade more, and trading more is expensive.

The evidence

Brad Barber and Terrance Odean spent years with the same brokerage data set from the previous lesson, and two of their papers are the foundation for everything in this lesson.

"Trading Is Hazardous to Your Wealth" (2000). They sorted about 66,000 households by how often they traded. Before costs, the most active traders and the least active earned similar gross returns. After costs (commissions and the bid-ask spread), the most active fifth of households earned substantially lower net returns than the least active, and lower than a simple market index. The stocks people bought did not do better than the stocks they sold. Activity did not add edge; it added cost.

"Boys Will Be Boys" (2001). Psychology research had established that men tend to be more overconfident than women in domains they perceive as masculine, including finance. Barber and Odean used this as a way to test whether overconfidence causes overtrading. In their data, men traded roughly 45% more than women, and the extra trading reduced their net returns by more. Single men traded the most and had the worst net performance. The point of the paper is not a gender contest; it is that a known difference in overconfidence lined up with a difference in trading volume and, through costs, a difference in results.

A related paper, Odean's "Do Investors Trade Too Much?" (1999), found that the stocks investors bought subsequently underperformed the stocks they sold, even ignoring costs. The average trade was not just costly; it was, on average, slightly wrong.

Key idea: Barber and Odean's work shows that trading more does not produce more edge. Costs are certain, edge is uncertain, and overconfidence makes people forget the difference. The traders who did best in the data were the ones who did the least.

Why overconfidence is so sticky

Overconfidence in trading is fed by three things that are structural, not personal.

Noisy feedback. In most skills, feedback is quick and clean: you play the wrong note, you hear it. In trading, a bad decision often wins and a good decision often loses. You can do something wrong fifty times and be rewarded thirty of them. That is close to the ideal schedule for training a belief that is not true.

Hindsight. After every move, the chart looks obvious. The mind quietly rewrites "I thought it might go up" into "I knew it would go up." Kahneman's work on hindsight bias shows that people genuinely misremember their earlier uncertainty. A trader who reviews charts without a written record of what they thought at the time will conclude, honestly, that they are better than they are.

Self-attribution. Winning trades are skill; losing trades are bad luck, the market, the news, the algos. This one has been directly studied in traders, and it correlates with subsequent overtrading: the more a trader credits their wins to themselves, the more they trade afterwards.

What overtrading actually looks like

Overtrading rarely feels like overtrading. It feels like being engaged. Some of its forms:

• Taking B-grade setups because the A-grade one has not appeared and the session feels wasted.
• Re-entering the same market right after a stop-out, because the idea "was right, just early."
• Adding instruments to the watchlist mid-session.
• Scalping the lunch chop "to stay sharp."
• Trading the first five minutes on a news release because "that's where the money is."
• In crypto and forex, trading at 2 a.m. because the market is open, not because a setup is present.

Each of these is a trade whose cost is certain and whose edge is unproven. The Barber-Odean result says that, in aggregate, these trades are a transfer from you to your broker.

A cost-side sanity check

You can make overconfidence visible by pricing it. Suppose you pay $2 round-trip per futures contract in commission and lose one tick of spread on average, worth $12.50 on a common index contract. That is $14.50 per contract per trade. At ten trades a day, 250 days a year, one contract each, the cost is $36,250 a year. To break even, your decisions must produce that much in gross edge before you have made a cent. At three trades a day, the hurdle drops to $10,875.

Stocks, forex and crypto have different numbers but the same shape: costs scale with trade count, edge does not.

Try it: Count your trades for the last month. Multiply by your all-in cost per trade (commission plus spread, or your platform's fee). That is your monthly hurdle. Now look at your log and mark every trade that was not on your written plan's setup list. Multiply that count by the same cost. That second number is what overconfidence billed you last month.

Recap

• Barber and Odean (2000): the most active households earned the lowest net returns; gross returns were similar, costs made the difference.
• Barber and Odean (2001): men, who psychology research finds are more overconfident in finance, traded about 45% more than women and had lower net returns as a result.
• Odean (1999): the stocks investors bought went on to underperform the ones they sold, so extra trades were on average slightly wrong even before costs.
• Overconfidence persists because trading feedback is noisy, hindsight rewrites memory, and wins are credited to skill while losses are blamed on luck.
• Costs scale with trade count and edge does not; price your trade count in dollars and prune everything not on the plan.