5.10 Why most retail intraday traders lose: base rates

Motivation

Before believing you have an edge, look at what happened to everyone else who believed it.

5.10.1Day-trader studies by country#

Evidence

Taiwan. Barber, Lee, Liu and Odean, "The cross-section of speculator skill: Evidence from day trading", Journal of Financial Markets 18 (2014), 1–24: fewer than 1% of day traders are able to predictably earn positive abnormal returns net of fees; of roughly 360,000 individuals who day-traded, about 1,000 a year predictably outperformed. Barber, Lee, Liu, Odean and Zhang, "Learning, fast or slow", Review of Asset Pricing Studies 10(1), 2020: day traders' aggregate performance is negative; traders with a history of losses generate 74% of day-trading volume; about 97% of day traders are likely to lose money in future day trading; and the behaviour fits overconfidence and biased learning better than rational learning about ability (abstract, RePEc).
Brazil. Chague, De-Losso and Giovannetti, "Day trading for a living?" (working paper, 2019; SSRN 3423101): of all individuals who began day-trading Brazilian mini-index futures between 2013 and 2015 and persisted for more than 300 days, 97% lost money; only 1.1% earned more than the Brazilian minimum wage and about 0.5% more than a bank teller's starting salary (0.4% in one version); there was no evidence of learning.
United States. Barber and Odean, "Trading is hazardous to your wealth", Journal of Finance 55(2), 2000: among 66,465 households at a discount broker (1991–96), the most active traders earned 11.4% a year against 17.9% for the market.
India. SEBI's FY23 study (PR 13/2024, 24 Jul 2024): 71% of individual intraday traders in the equity cash segment made net losses (65% in FY19, 69% in FY22); among those with more than 500 trades a year, 80%; among loss-makers, trading costs added a further 57% to their trading losses. In F&O, 87.7% of individuals lost money in FY26 (M01 · 1.11).

5.10.2Mechanisms of loss#

Mechanisms, mapped to this module

Costs (5.9): the hurdle $c/\sigma_t$ is large at short horizons.
Adverse selection (5.3–5.4): limit orders are filled preferentially when informed or faster traders want the other side; market orders pay the spread that compensates liquidity providers for this.
Speed: in a price-time-priority book (M01 · 1.4), the first to react to public information captures it; retail is last.
Behaviour (M14): overconfidence, the disposition effect (selling winners early, holding losers), preference for lottery-like payoffs.
Statistical illusion: short samples of intraday P&L are noisy; a lucky month looks like skill.

5.10.3What a track record is worth#

Worked examplehow much is a good track record worth?

Take the Taiwanese base rate: a prior probability of 1% that a day trader is skilled. Suppose a skilled trader has a true net per-trade Sharpe of 0.1, an unskilled one 0. After 200 trades, the skilled trader's t-statistic has mean $0.1\sqrt{200} = 1.41$ . The probability of observing $t \ge 2$ is $P(Z \ge 0.59) = 0.279$ if skilled and $P(Z \ge 2) = 0.023$ if not. By Bayes' rule the posterior probability of skill is $0.01 \times 0.279/(0.01 \times 0.279 + 0.99 \times 0.023) = 0.11$ . A "statistically significant" 200-trade record leaves an 89% chance that the trader has no edge. After 500 trades with $t \ge 2$ the posterior is about 0.21. Base rates dominate small samples.