Recency bias
What is Recency bias?
Recency bias is the tendency to give recent events disproportionate weight when judging probability, trend, or future return.
Historical dashboards should allow multiple horizons and mark unusual regimes, structural breaks, and data limitations. A default window can silently drive conclusions, so users should see whether a ranking survives different start dates and whether the recent observation is statistically or economically unusual within the longer record.
Portfolio forecasts should blend current conditions with long-run evidence transparently and show how much the conclusion depends on the selected weighting scheme.
How it works
Recent observations are easier to recall and can feel more representative than older evidence. Investors may extrapolate a rally, crash, inflation shock, low-volatility period, or manager streak beyond what the full record supports. Giving new information more weight can be rational when conditions changed; bias arises when the weight exceeds its relevance and reliability.
Portfolio consequences
Recency can drive performance chasing, selling after losses, shortening horizons, concentrating in recent winners, or abandoning diversification just before leadership changes. Risk models based on a short calm window can understate danger, while models dominated by a crisis can overstate normal risk. Fund flows and marketing rankings can reinforce the cycle.
Example
After three strong years for one market, an investor raises its expected return and reallocates heavily despite higher valuation. The return was real, but its recency does not prove persistence. Comparing longer history, valuation, earnings, currency, and alternative regimes may show that future reward is less certain than the recent chart suggests.
Decision controls
Use multiple horizons, rolling periods, valuation, base rates, scenario analysis, and explicit regime assumptions. Rebalance to policy bands rather than recent rankings. For forecasts, separate structural evidence from price momentum and state why a recent observation should persist. Avoid using an arbitrarily long history when technology, regulation, or market structure has genuinely changed.
Practical review
Test sensitivity to start and end dates, include difficult periods, and compare real-time with revised data. Ask whether the current recommendation would reverse if the last twelve months were hidden. Record expectations before observing the next outcome. Recent evidence belongs in analysis, but it should not silently replace the full probability distribution.
Sources and further reading
- Behavioral Patterns of U.S. Investors, U.S. Securities and Exchange Commission
- Ten Things to Consider Before You Make Investing Decisions, Investor.gov, U.S. Securities and Exchange Commission