The Portfolio That Sells Its Winners
A discretionary manager runs $420m. Year to date the fund is +6.2% against a benchmark of +9.1%. Trade records show realized gains have an average holding period of 4.1 months, while positions closed at a loss average 19.3 months. Winning trades turn over at 140% annually; the book overall turns over 85%. In a recent review the manager defended a position held at −38% from cost by saying 'it has to come back - the business is worth at least what I paid.' He also sold a position at +12% that subsequently rose another 60%, explaining that he 'didn't want to give back the gain.' After a strong October he raised gross exposure from 110% to 145%.
1. Name the behavioral mechanisms at work and cite the specific data points in the setup that evidence each one. Distinguish what is genuinely a bias from what could be legitimate process.
Read the worked answer
1. Disposition effect. The 4.1-month average holding period for winners versus 19.3 months for losers is the textbook signature: sell what feels good, hold what hurts. Turnover data sharpens it - 140% on winners vs 85% on the book as a whole means the manager's activity is concentrated in realizing gains.
2. Anchoring to the purchase price. 'Worth at least what I paid' is a reference point that has nothing to do with value. The purchase price is a sunk fact; the only relevant question is whether the forward return justifies the capital today. This is anchoring plus loss aversion producing a get-evenitis hold.
3. Regret aversion / prospect-theory risk seeking in gains. 'Didn't want to give back the gain' at +12% reflects the concave value function over gains: a sure small gain is preferred to a gamble with higher expected value. Selling a winner early is risk-averse in the domain of gains.
4. House-money effect / overconfidence after success. Raising gross from 110% to 145% after a good October, with no stated change in opportunity set, is classic. The prior gain has changed perceived risk budget even though the capital base and the opportunity set have not.
What is not necessarily a bias: long holding periods for losers can be correct if there is a documented thesis with a catalyst and the position is sized properly. The bias claim is supported here because the justification offered is the cost basis, not the fundamentals - and because the asymmetry in holding periods is systematic, not position-specific.
2. Design three concrete process rules that would dampen these behaviors without stripping the manager of discretion. Explain the mechanism by which each one works.
Read the worked answer
Rule 1 - Thesis review independent of cost basis. Each position carries a written pre-entry thesis with a falsifiable catalyst and a dated review trigger (e.g. two quarters, or a defined event). At review, the only permitted questions are 'would I buy this today at this price?' and 'what evidence has changed?' P&L versus cost is displayed only in the risk report, not on the position screen. Mechanism: removes the anchor from the decision frame, so loss aversion has nothing to grip.
Rule 2 - Percentage-based, pre-committed exit discipline instead of discretionary realization. Define per-position invalidation (thesis break, price stop, or time stop) at entry, and require a written override to hold past it. For winners, replace 'sell at +X%' with a trailing rule or a scaling schedule (e.g. trim one-third at a defined valuation band, let the rest run with a stop). Mechanism: converts the gain-domain decision from an emotional realization into a pre-specified rule made in a cold state, and prevents the asymmetric truncation of the right tail that is costing the fund (the +60% winner sold at +12%).
Rule 3 - Exposure changes tied to a stated signal, not to recent P&L. Any change in gross exposure above a threshold (say 10 percentage points) requires documenting the input that changed - opportunity set, volatility, signal count - and is approved against a rule written in advance. Mechanism: breaks the link between recent success and risk appetite, converting a house-money impulse into an auditable decision.
Supporting measurement: report hit rate, average win/loss, and holding period for winners vs losers quarterly. The manager can argue with a rule; it is harder to argue with a 4:19 holding-period ratio printed next to the benchmark shortfall.