Investors Sell Winners Faster Than Losers, and Pay for It

The disposition effect is usually presented as a psychological curiosity - investors dislike admitting a mistake. It is more specific than that: it is a measured, costly pattern that runs directly opposite to what tax efficiency rewards, and the alternative explanations have been tested and rejected.

Hersh Shefrin and Meir Statman named the effect in 1985. What follows concerns the evidence for it, drawn from Terrance Odean’s direct test on real brokerage records, and what it implies for anyone managing a taxable portfolio.

Ten Thousand Accounts, One Consistent Direction

Odean’s “Are Investors Reluctant to Realize Their Losses?”, published in the Journal of Finance in 1998, examined trading records for 10,000 accounts at a large discount brokerage. The measure he introduced compares the proportion of available gains actually realised against the proportion of available losses actually realised - the denominators matter, because an investor holding mostly winners will sell mostly winners without any behavioural bias at all.

Corrected for that, investors realised gains at roughly one and a half to two times the rate at which they realised losses. The direction was consistent, not marginal.

The Obvious Innocent Explanations Were Tested and Failed

This is the part that turns an observation into a finding. Portfolio rebalancing would produce a similar pattern, and does not explain it. Avoiding the higher trading costs of low-priced stocks would produce a similar pattern, and does not explain it either.

Nor was the behaviour vindicated by what happened next. The winners investors sold went on to outperform the losers they kept, so the pattern was not skill in disguise. On taxable accounts it was straightforwardly suboptimal and produced lower after-tax returns.

The Instinct Runs Backwards From the Tax Code

A position sitting on a loss is the one worth selling first on tax grounds - it banks a deduction against other gains. A position sitting on a gain is the one worth holding absent another reason, since selling crystallises a liability that could have been deferred.

Investors on average do close to the opposite: they realise the tax bill early and defer the deduction. The behaviour is not a rounding error against the returns involved, because the cost compounds - every year the deduction is deferred is a year its value sits unused while the crystallised gain has already been paid on.

Loss Aversion, Not Miscalculation

The mechanism is not arithmetic error. Selling a loser converts a paper loss into a realised one and closes off the possibility that the position recovers - it is an admission that a specific decision was wrong. Selling a winner converts a paper gain into a banked one and feels like vindication at the moment of sale.

Both moves feel good for reasons unconnected to what either position is likely to do next. That is why the effect survives in investors who know about it: it is generated by how the outcome is experienced, not by what the investor believes about probabilities.

What This Means for Allocators

Measure the ratio rather than trusting the impression. Proportion of gains realised against proportion of losses realised is computable from any transaction history, and it is the diagnostic Odean used.

Separate the tax question from the conviction question explicitly. A losing position can be worth selling because the thesis broke, and a winner worth holding because it did not - the error is letting the realised or unrealised status of the position do that reasoning silently.

And treat comfort as a signal to check. The trade that feels easiest to make and the one that feels easiest to avoid both run in the direction the portfolio’s tax efficiency penalises.