Recency bias

Recency bias is the tendency to weigh recent events or data much more heavily than older information when forming judgments or making decisions. The last thing you saw feels more important than the average of everything that came before, even when the older information is more statistically reliable or more representative of the underlying reality.

The mechanism

Recency bias happens because recent events are psychologically more available — easier to recall and more vivid in your mind. If a stock has risen 30% in the past month, that recent success is salient and easy to retrieve from memory. The fact that it fell 50% over the previous three years is harder to recall in the moment.

This availability then gets misinterpreted as importance. “This happened recently, therefore I should weight it heavily.” But recency is not a sound basis for probability or expected value. A stock’s recent performance tells you almost nothing about its future return; the fact that it happened last week rather than last year does not change that.

Recency in market cycles

Recency bias is most visible (and most dangerous) at turning points in market cycles.

At the peak of a bull market, after years of rising prices, investors become convinced that rising is the normal state. Recent evidence — all positive — confirms this. They buy aggressively, extrapolating the recent uptrend. Valuations soar.

At the trough of a bear market, after months of falling prices, investors become convinced that falling is normal. Recency says the market will keep going down, so they sell in panic. Valuations become absurdly cheap.

In both cases, recency bias causes investors to do the opposite of what a longer-term view would suggest. They buy high (when recent performance is best) and sell low (when recent performance is worst).

Why the long average matters

A classic illustration: an investor looks at a stock that has returned 25% annually over the past five years but 6% annually over the past 20 years. If recent performance is in her mind — and it will be, thanks to recency bias — she might expect 25% going forward. But the longer history is more reliable. Over 20 years, luck and cycle effects have mostly averaged out; five years is barely enough time to distinguish signal from noise.

Recency bias causes you to treat the last few data points as if they are more representative than they are. You end up extrapolating a recent run and missing reversion to the long-term mean.

Recency and earnings per share forecasts

One of the clearest examples is how analysts forecast earnings. Recent results feel more real and more predictive than they are. A company misses earnings for two quarters, and the consensus forecast is cut sharply. But if the company’s long-run trajectory suggests the miss was temporary, recency bias in the analyst community will cause forecasts to undershoot, creating a setup for positive surprises later.

Conversely, after a quarter of strong beats, analysts push forecasts up too much, extrapolating the recent good fortune. This is where recency bias and overconfidence bias often work together.

Recency and portfolio rebalancing

In asset allocation, recency bias suggests you should buy whatever has recently done best and sell whatever has recently done worst. This is the opposite of the disciplined approach: rebalancing means selling your winners and buying your losers (mean reversion).

An investor who rebalances quarterly or annually can fight recency bias mechanically. The investor who adjusts holdings based on recent performance is likely falling into the bias trap.

Distinguishing recency from other biases

Recency bias is about when you weigh evidence (recent stuff too much). Availability heuristic is about how easy to retrieve something is (which often correlates with recent, but not always). Anchoring bias is about dependence on a specific starting value. They can all occur together.

Recency is also not the same as hot-hand fallacy, though they often co-occur. Hot hand is the belief that a streak will continue. Recency is the overweighting of the streak when forming any judgment.

Defenses against recency bias

• Use trailing averages. When evaluating a stock or fund, look at 10-year returns, not recent returns. The longer history is less affected by recency.
• Establish rules before the bias hits. Decide on your asset allocation and rebalancing schedule when you are calm, not after a strong recent run has made you excited or a recent decline has made you scared.
• Check the history. When you feel certain a trend will continue (bull market, rising earnings, strong momentum), look at how long past trends actually lasted. This quickly deflates recency bias.
• Use a process, not judgment. A mechanical rebalancing rule or a diversification policy removes the opportunity for recency-driven emotional decisions.

See also