Overconfidence and trading frequency: the cost of conviction

In 2000, Brad Barber and Terrance Odean published "Trading Is Hazardous to Your Wealth" in The Journal of Finance. They studied trading records from 66,465 households at a US discount broker between 1991 and 1996. Households in the most-active quintile turned over their portfolios at an annualized rate above 250% and earned net returns of 11.4% per year — 6.5 percentage points below the 17.9% market return over the period, and roughly 7 points below the least-active quintile. Gross returns across activity quintiles were nearly identical; the entire gap came from transaction costs incurred by trading on signals that did not have edge.

The follow-up paper, "Boys Will Be Boys: Gender, Overconfidence, and Common Stock Investment" (Quarterly Journal of Economics, 2001), tested whether overconfidence was the mechanism. Drawing on a well-documented finding from psychology that men are more overconfident than women in tasks perceived as masculine (Lundeberg, Fox, and Punccohar, Journal of Educational Psychology, 1994), Barber and Odean showed that men traded 45% more than women and underperformed them by approximately 0.94 percentage points annually as a result. Single men traded 67% more than single women and underperformed by 1.44 points. The trading-cost gap tracked the overconfidence gap.

The mechanism, formalized by Kent Daniel, David Hirshleifer, and Avanidhar Subrahmanyam in "Investor Psychology and Security Market Under- and Overreactions" (The Journal of Finance, 1998), is that overconfident investors overweight the precision of their private signals and underweight public information. Each signal feels actionable. The aggregate result is a high-frequency stream of trades on noise, paid for at the bid-ask spread.

The effect is not confined to retail. Markus Glaser and Martin Weber, in "Overconfidence and Trading Volume" (The Geneva Risk and Insurance Review, 2007), surveyed 215 online-brokerage investors and found that those who scored higher on a calibration-based overconfidence measure traded more — and that the relationship held controlling for age, wealth, and experience. The investors who knew the least about the limits of their own knowledge traded the most.

The implication is uncomfortable: most discretionary trading after the initial position is taken is destroying value, and the investor cannot tell in the moment which trades are the exception. The signal-to-noise ratio of any individual conviction is low enough that the cost of acting on it usually exceeds the edge it carries.

The structural defense is to bind trading decisions to pre-specified conditions rather than to in-the-moment conviction. If a position is exited only when an invalidation condition trips, or scaled only when a revise-up condition trips, the rate of trading collapses to the rate at which the world actually delivers new evidence — which is far lower than the rate at which a confident investor feels they have new evidence.

Tensile bounds position changes to pre-committed conditions on the thesis, so the trading rate is set by evidence rather than by conviction.