# Overtrading and Poor Stock Selection Reduce Individual Investor Returns

Cite as: Gorak, R. (2026). Overtrading and Poor Stock Selection Reduce Individual Investor Returns. Tradicted. https://www.tradicted.com/research/odean-overtrading-1999/
Paper: Terrance Odean — *Do Investors Trade Too Much?*
Published in: American Economic Review (1999)
Original: https://faculty.haas.berkeley.edu/odean/papers%20current%20versions/doinvestors.pdf

Key finding: Over a one-year horizon, the securities purchased by discount brokerage investors underperformed those they sold by 3.31 percent (January 1987–December 1993), even before accounting for round-trip trading costs of approximately 5.9 percent.

---
Overtrading occurs when investors trade more than is justified by their information, paying unnecessary transaction costs in pursuit of gains they never realize. An investor switching stocks must earn roughly 6 percent more on the new position just to break even after commissions and spreads. In *Do Investors Trade Too Much?*, Odean (1999) analyzed 97,483 trades across 10,000 U.S. discount brokerage accounts from January 1987 through December 1993. Purchased securities underperformed sold securities by 3.31 percent over the following year—even before counting transaction costs.

## What the Study Found

When trading was filtered to exclude liquidity demands, tax-loss selling, and rebalancing, the return gap widened to 5.82 percent over one year. The bottom 90 percent of investors by trading activity showed a 4.28 percent return gap over one year. The most active 10 percent of investors showed a narrower gap of 2.69 percent over one year. Market-adjusted returns confirmed poor stock selection: purchases trailed sales by 3.22 percent over 252 trading days. A calendar-time portfolio using the Fama-French three-factor model produced a buy-minus-sell alpha of −0.249 percent per month, significant at the 1 percent level.

## Methodology

The data came from an anonymous U.S. discount brokerage that provided trade and position records for 10,000 randomly selected accounts. The 162,948 trade records span January 1987 through December 1993, with 97,483 trades matched to CRSP daily return data. Statistical significance was estimated by bootstrapping a 1,000-observation empirical distribution, with replacement securities matched by CRSP size decile and book-to-market quintile. Robustness checks used calendar-time portfolios with CAPM and Fama-French three-factor regressions across 4-, 12-, and 24-month formation periods.

## Key Statistics

| Metric | Finding | Context |
|---|---|---|
| Return gap: purchases vs. sales (84 trading days) | −1.36% | All transactions, full sample |
| Return gap: purchases vs. sales (252 trading days) | −3.31% | All transactions, January 1987–December 1993 |
| Return gap: purchases vs. sales (504 trading days) | −3.32% | All transactions, full sample |
| Return gap, excluding liquidity/tax/rebalancing (252 days) | −5.82% | Purchases within 3 weeks of profitable complete-position sales |
| Return gap: bottom 90% of traders (252 days) | −4.28% | Least active investors by trade count |
| Return gap: top 10% of traders (252 days) | −2.69% | Most active investors by trade count |
| Market-adjusted return gap (252 days) | −3.22% | Excess of CRSP value-weighted index; all transactions |
| Fama-French alpha, buy-minus-sell portfolio | −0.249%/month | 4-month formation period; significant at 1% |
| CAPM alpha (Jensen alpha), buy-minus-sell portfolio | −0.311%/month | 4-month formation period; significant at 1% |
| Average round-trip trading cost | ~5.9% | Average purchase commission 2.23% + average sale commission 2.76%; includes bid-ask spread estimate |
| Average monthly portfolio turnover | 6.5% | Full sample; estimated from position statements |
| Market timing regression (Eq. 4): R_mt = α + β·(Buys_{t-1} / (Buys_{t-1} + Sells_{t-1})) + ε | t = −0.4, R² = 0.0 | Lagged order imbalance does not predict market return |
| Average return to purchases (Eq. 1): R_{P,T} = (Σ∏(1 + R_{j,t})) / N − 1 | Computed over T = 84, 252, or 504 days | N = number of purchases; R_{j,t} = CRSP daily return for security j |

## Why This Matters

Retail investors who actively select stocks appear to misread information direction, not merely overestimate its precision. Reducing trade frequency would improve net returns for active retail investors, since each round-trip trade erodes principal before any return can accumulate. Attention-driven buying—concentrating purchases in securities that have moved dramatically—amplifies exposure to high-momentum names prone to reversal. The results challenge the assumption that retail investors hold useful private information about the securities they choose to trade. Turnover only looks harmless until you total it. A [trading journal](/trading-journal/) records every entry so your own frequency becomes a number rather than an impression.

## FAQ

### How much worse did investors perform when trading specifically to improve returns?

5.82 percent: the one-year return gap between sold and purchased securities after filtering out liquidity, tax-loss, and rebalancing trades. Over two years the gap reached 8.91 percent. The filtered sample required purchases within three weeks of profitable complete-position sales in securities of equal or smaller size, removing the most common non-profit-seeking motives.

### What did calendar-time portfolio tests show about the buy-minus-sell return gap?

−0.249 percent per month: the Fama-French three-factor alpha for the buy-minus-sell calendar-time portfolio using a 4-month formation period. Raw returns and CAPM alpha produced similarly negative results across all three formation periods (4, 12, and 24 months). Results for the 4- and 12-month periods were significant at the 1 percent level; the 24-month period was significant at the 5 percent level.

### How do the three calendar-time performance measures compare for the 12-month formation period?

−0.207 percent per month: the Fama-French three-factor alpha for the 12-month formation period buy-minus-sell portfolio. The CAPM alpha for the same period was −0.234 percent per month; the raw monthly return was −0.225 percent per month. The 24-month formation period Fama-French alpha was −0.136 percent per month.

### Did individual investors show any ability to time market entry and exit?

t = −0.4, R² = 0.0: the result of regressing the current month's CRSP return on the prior month's order imbalance. Market-adjusted returns to purchases trailed sales by 3.22 percent over 252 days; the raw return gap over the same period was 3.31 percent. The paper attributes both gaps to poor stock selection, with market timing contributing negligibly.

## Source

Odean, T. (1999). Do Investors Trade Too Much? *American Economic Review*, 89(5), 1279–1298.

[Read the full paper →](https://faculty.haas.berkeley.edu/odean/papers%20current%20versions/doinvestors.pdf)