Recency bias is the tendency to give recent information disproportionate weight in decisions. In investing, it shows up as the pattern of buying what just went up and selling what just went down.
It is one of the most studied biases in retail behavior, and one of the most reliably expensive. Money flows into the asset class that performed best in the last twelve months. Money flows out of the one that performed worst. Both decisions are typically wrong, because asset-class returns mean-revert across longer windows than the twelve months retail investors anchor on.
This is a plain-language walkthrough of what recency bias is, where it shows up in retail decisions, the math that makes it expensive, and the procedural defenses that work.
Why we anchor on the recent past
The human brain processes recent information more vividly than older information. The 2008 financial crisis felt more present in 2010 than the dot-com bust felt in 2007, even though both were equally part of the historical record at the time. The pattern is consistent across all kinds of decisions, not just investing.
For investing specifically, the recent-past anchor combines with two other forces to make recency bias particularly destructive.
Available memory. Information you remember vividly feels more probable than information you do not. A market drawdown that you lived through feels more likely to repeat than one you read about in a history book.
Confirmation seeking. Once you have anchored on "last year was good for tech," you start noticing news that supports the pattern and discounting news that contradicts it. The anchor self-reinforces.
Social proof of performance. Funds and strategies that performed well last year get more press coverage. The press coverage reinforces the anchor. The reinforcement attracts more inflows. The inflows can drive prices higher in the short term, further reinforcing the anchor. Eventually the cycle reverses.
The result is a pattern where retail money systematically flows into the asset class, sector, or strategy that has performed best in the last twelve months, often near the peak of that performance.
The math of mean reversion
Asset returns mean-revert across most observable time horizons. The exact mechanism varies by asset class, but the pattern is consistent.
Equity sector returns mean-revert over five-to-ten-year windows. The sector that led one decade is rarely the sector that leads the next. Technology dominated the late 1990s, lagged the 2000s, dominated the 2010s. Energy was the worst-performing sector for most of the 2010s and the best-performing sector in 2022.
International versus US equity returns mean-revert over multi-year windows. The US dominated international equities from 2008 to 2022. Before that, international dominated US from 2002 to 2007.
Small-cap versus large-cap returns mean-revert over multi-year windows. Each takes turns leading.
Value versus growth styles mean-revert over multi-year windows.
Across nearly every dimension you can slice the equity market, recent winners tend to be middle-pack or laggard performers over the next five years. Reversion is not perfectly predictable in timing, but the directional pattern is reliable.
Bonds, real estate, commodities, and most other major asset classes show similar reversion patterns over their own characteristic time scales.
A worked example
Consider an investor who, in early 2022, looked at the previous five years of returns. Large-cap technology stocks had compounded at roughly 20% per year. Small-cap value had returned roughly 8%. International had returned roughly 6%.
The recency-bias decision: shift the portfolio toward large-cap technology. The recent past was clearly telling them where the returns were.
The actual outcome: 2022 saw technology drop 30+% while small-cap value held up better and international caught up. By the end of 2023, a five-year window-rolling forward looked very different from the one the investor anchored on.
The mistake was not in noticing the past returns. The mistake was in projecting them forward without recognizing that the same returns that produced the impressive five-year window were also creating the conditions for reversion.
The same pattern played out in the opposite direction in 2009. Investors who looked at recent returns saw equity down 50%+ over 18 months. The recency-bias decision was to move to cash and bonds. The mean-reversion-aware decision was to maintain or increase equity exposure precisely because returns had been bad. The latter dramatically outperformed.
How recency bias compounds across decisions
Recency bias does not produce a single bad decision. It produces a sequence of decisions that compound into substantial underperformance.
The first decision is the initial chase: buying the asset class that just went up. The second decision is the panic exit: selling that same asset class after it has dropped 30%. The third decision is the next chase: buying the new asset class that has now had two years of strong returns. The fourth decision is the next panic exit. And so on.
Each decision is rational in isolation if you anchor on the most recent twelve months. The compounded sequence is catastrophic. The pattern is sometimes called "performance chasing" and it is the largest single source of the gap between investor returns and fund returns measured in industry data.
DALBAR's annual Quantitative Analysis of Investor Behavior study has documented this gap for decades. The average equity-fund investor underperforms the average equity fund by roughly 2 to 4 percentage points per year. The gap is not because investors pick bad funds. It is because they buy and sell funds based on recent performance, capturing the bad runs and missing the recoveries.
The procedural defenses
Defending against recency bias requires structural commitments that override the in-the-moment reaction to recent performance.
Pre-commit to your asset allocation in writing. Decide what mix of equity, fixed income, international, and alternative exposure you want, based on your goals and time horizon. Rebalance back to those targets quarterly or annually. The rebalancing mechanically forces you to sell what has run and buy what has lagged.
Ignore one-year returns when evaluating performance. Any meaningful evaluation of an investment strategy requires at least three years of data, ideally five or more. A strategy that did well in the last twelve months is not necessarily a good strategy. A strategy that did poorly in the last twelve months is not necessarily a bad one. Single-year noise is dominated by mean-reversion patterns at longer horizons.
Hold positions for at least a full business cycle. Most equity strategies do not show their true character in fewer than five years. Picking a strategy and then bailing on it after two years of underperformance is almost always more about recency bias than about the strategy's actual quality.
Read history. The most reliable defense against recency bias is exposure to longer historical context. Every market regime that felt unprecedented in real time has happened before in some form. Every "this time is different" claim has been wrong. Reading market history at the half-century scale calibrates expectations downward and prevents anchoring on recent windows.
Use disciplined methodology. Strategies with explicit, transparent methodologies (like model portfolios) are easier to evaluate on long horizons because the rules are visible. A model that has worked across multiple market regimes has higher claim to durability than one that worked beautifully for the last three years and has not been tested elsewhere.
A specific application: rebalancing
The simplest defense against recency bias in a practical portfolio is mechanical rebalancing. Set target weights for asset classes (e.g., 60% equity / 40% bonds, or 70% US / 30% international). When market movements push the weights more than 5 percentage points off target, rebalance back.
This sells what has run (the recently performing asset) and buys what has lagged (the recently underperforming asset). It is the exact opposite of what recency bias would have you do. Over decades, the rebalancing premium adds 50 to 100 basis points of annualized return for portfolios that maintain discipline.
Most retail investors fail to rebalance precisely because the action feels uncomfortable. Selling the winner and buying the laggard feels wrong every time. The discomfort is the bias talking. The math says do it anyway.
The bottom line
Recency bias is the tendency to anchor on the most recent twelve months as if they were the whole story. They are not. Asset returns mean-revert across longer windows than retail investors typically consider.
The defense is not willpower; it is structure. Pre-committed asset allocation. Ignore one-year returns. Hold strategies through full cycles. Read history. Rebalance mechanically.
These five disciplines, applied consistently, recover most of the 2-to-4 percentage point annual underperformance that DALBAR has documented in retail accounts. The math compounds across decades into life-changing dollar amounts.
This is educational content, not personalized investment advice. The right asset allocation and rebalancing approach for your portfolio depends on your specific situation, time horizon, and goals. Consult a financial advisor for guidance tailored to you.