Recency Bias: Why Yesterday's Returns Mislead You

When markets have gone up for two years, almost everyone expects them to keep going up. When they have fallen for two years, the expectation flips — even though nothing about long-run odds has changed. This pattern has a name, recency bias, and it is one of the most expensive habits in investing. This article works through what the bias is, the arithmetic that quantifies its cost, and practical structures that keep recent history from dictating your next move.

What recency bias is

Recency bias is the tendency to give recent events outsized weight when judging the future. Memory is easier than statistics, so the last few quarters feel more "real" than decades of data. In investing it shows up in four reliable ways:

• Extrapolating returns. A fund that returned 20% last year gets inflows; the same fund after a -15% year gets outflows — often the same fund, the same managers, the same holdings, different cash flows.
• Forecast anchoring. Year-ahead outlooks written during good times skew optimistic; the same analysts in drawdowns skew pessimistic.
• Portfolio drift. Investors gradually concentrate into whatever recently worked, which usually means owning more of an asset after its price has risen.
• Panic in drawdowns. After a crash, "this time is different" feels obvious, and cash feels safe — precisely when expected future returns tend to be higher.

It is related to but distinct from the broader behavior gap between investment returns and investor returns; recency bias is one of the main engines that drives the gap.

The arithmetic of chasing and fleeing

That example is simplified — real sequences are messier and longer — but the mechanism scales. Industry studies repeatedly find the gap between fund returns and the returns actually earned by fund investors runs to several percentage points per year in volatile funds, largely from timing driven by recent performance.

How good are the forecasts you are anchoring to?

The base rates on year-ahead market forecasts are humbling. Institutional strategists' aggregate index targets have historically missed the actual year-end level by wide margins in most years, and the errors are not systematically conservative or optimistic — they cluster around whatever just happened, which is exactly what recency bias predicts. Individual stock analysts' price targets show similar behavior: targets adjust after price moves more than they anticipate them.

None of this says forecasts are worthless — they bundle assumptions worth reading. It says the confidence implied by a single number a year out is not supported by its track record. A plan that requires the forecast to be right is a fragile plan.

Structures that neutralize recency bias

Since the bias is a feature of human psychology rather than a knowledge gap, the fix is procedural, not motivational:

1. Write an investment policy statement. One page: target allocation, what you buy on a schedule, and the specific rules under which you would deviate. Decide during calm weather what you will do in storms, then obey your past self.
2. Automate contributions. Dollar-cost averaging on autopilot removes the decision point where recency bias operates. You buy in down markets automatically, which is the behavior everyone endorses and few execute manually.
3. Rebalance by bands, not by feeling. A written band rule — for example, act only when an asset class drifts more than 5 percentage points from target — turns "the market fell, should I buy?" into a mechanical question with a mechanical answer. Our rebalancing bands guide works the arithmetic.
4. Check performance rarely, and never in isolation. Review annually, and compare against your own plan rather than last year's leaderboard. If a holding decision requires believing that the last three years will repeat, treat that as a red flag, not a thesis.
5. Pre-commit to a news diet. Daily market narratives are ~99% recency. Weekly or monthly consumption is enough for a long-horizon investor and leaves less residue.

A quick self-test

Ask yourself these questions and note the gut answers:

• If your largest holding had returned +35% over the last 12 months, would you add more? Now: if it had returned -25%, with no change in the business, would you sell?
• What is your honest guess for next year's market return, and what evidence — beyond the last two years — supports it?
• Look at your last three transactions. Were they purchases of things that had recently gone up, or sales of things that had recently fallen?

If the pattern is obvious, you are in good company — and the procedural fixes above exist precisely because willpower does not solve this.

The long view

Recent history is one input, and a noisy one. Decades of market data, the arithmetic of valuations, and your personal timeline are all more stable guides. The investors who compound successfully are rarely the best forecasters; they are the ones whose process makes good forecasting unnecessary. A sensible next step is reviewing your target stock-bond allocation and writing it down, so that the next exciting twelve months — in either direction — finds you with a plan instead of a mood.