Recency Bias in Investing: Why Your Brain Overweights the Last Few Weeks
Recency bias is the tendency to treat the most recent price action as more informative than it actually is — extrapolating a short rally or a short crash into a permanent trend. What the research found, what it costs investors in real dollars, and how to catch it before your next decision.
Recency bias is the tendency to give recent events far more weight than older ones when forming an expectation about what happens next — even when the recent stretch is a small, unrepresentative slice of a much longer history. In investing, it shows up as two mirror-image mistakes: assuming a short rally is the start of a durable trend, and assuming a short crash means the asset (or the market) is broken for good. Neither belief is actually supported by three weeks of price action, but both feel like conclusions rather than guesses while you’re inside them.
It’s one of the more dangerous biases covered on this site precisely because it doesn’t feel like bias — it feels like updating on new information, which is normally a good thing to do. The problem is the weighting: recent data isn’t just one more data point among many in your head, it quietly becomes the dominant one, crowding out a year or three years of history that would tell a very different story.
The mechanism: why recent data hijacks your beliefs
The clearest empirical look at this comes from Zhengyang Jiang and coauthors’ paper in The Quarterly Journal of Economics, “Investor Memory and Biased Beliefs: Evidence from the Field”. The researchers surveyed a nationally representative sample of more than 17,000 Chinese retail investors and matched their self-reported recollections of past returns against those investors’ actual, recorded trading histories. Two things stood out. First, recall showed a strong recency effect on its own — investors disproportionately remembered recent periods, especially the most recent month, over equally significant periods further back. Second, recall wasn’t only about recency: sharp crashes and dramatic rallies got remembered disproportionately regardless of when they happened, meaning memory is shaped by how dramatic an experience was, not just how new it is. Investors’ stated expectations about future returns tracked these biased recollections closely, and those expectations predicted what they actually did with their money next.
That finding builds on an earlier, widely cited result: Ulrike Malmendier and Stefan Nagel’s 2011 QJE paper, “Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?” They found that people who personally lived through a period of poor market returns report lower risk tolerance and invest less in stocks for years afterward, even when the underlying economic outlook has nothing left to do with that earlier period — and that the most recent experiences carry disproportionately more weight in this effect than older ones. Put the two papers together and the picture is consistent: your brain doesn’t average your investing history evenly. It builds your expectations mostly out of whatever happened most recently, or whatever hit hardest, and quietly discounts the rest.
What it actually costs, in dollars
This isn’t just an interesting quirk of memory — it has a measurable price tag, because biased expectations drive real buying and selling decisions. Morningstar’s “Mind the Gap” 2025 study compared the returns investors actually earned in U.S. mutual funds and ETFs to the returns those same funds reported, over the ten years ending December 31, 2024. The average fund returned about 8.2% annually, but the average dollar invested in those funds earned only about 7.0% — a gap of roughly 1.2 percentage points a year, equal to about 15% of the funds’ total return over the decade. The gap wasn’t random: it was driven by the timing of investor purchases and withdrawals, and it was substantially worse for more volatile funds. The most volatile category of funds showed a 1.8-percentage-point annual gap, more than double the 0.8-point gap in the most stable category.
That pattern is recency bias with a receipt attached. A volatile fund produces sharper, more recent swings to overweight — a hot run pulls in money right as the run is ending, and a sharp drawdown pushes money out right as it’s bottoming. The fund’s actual long-run return was available the whole time; what moved the money was the last few months of it.
Why it hits hardest during a recovery
Recency bias is loudest exactly when you’re least equipped to correct for it: right after a loss. Three red weeks in a row after you’re already down don’t just feel discouraging — they feel like proof that the position (or the market, or you) is broken, in a way that overrides however many green months came before it. The reverse is just as common on the way back up: three green weeks after a drawdown can feel like confirmation the worst is over, even when three green weeks is a tiny, statistically unremarkable sample. Neither feeling is wrong to notice. Both are wrong to treat as a conclusion.
This is closely related to, but distinct from, two other biases covered on this site. The gambler’s fallacy is the belief that a streak makes the opposite outcome more likely — “it’s been down this long, it’s due.” Recency bias is closer to the opposite error: believing a streak makes the same outcome more likely to continue, simply because it’s what just happened. And it feeds directly into the question covered in Dead Cat Bounce or Real Recovery? — the four-part framework there (breadth, catalyst, volume, retest) exists precisely because “it went up for a week” isn’t, on its own, evidence of anything. Recency bias is why that week feels like more evidence than it is.
A worked example
Say a position is down 35% from its high. Over the last three weeks, it’s climbed 18% off the recent low. Recency bias whispers two contradictory things depending on which window you’re looking at: if you’re anchored to the last three weeks, it feels like the recovery is confirmed and you should add back in at size. If you’re anchored to the fact that it’s still down 35% overall, the last three weeks can feel like noise not worth acting on either way.
The check that cuts through both: pull the asset’s full price history, not just the window that happens to match your current mood, and ask what a reasonable prior would have been before the last three weeks started. If nothing about the underlying thesis changed in that window — no new information, no resolved uncertainty — then the three-week move is a data point to note, not a thesis to build a position around. If something concrete did change, the recent window is worth more weight, but the reason to weight it should be the news, not the fact that it’s recent.
Recent window vs. longer window: what each one is actually good for
| What you’re deciding | What the recent window tells you | What a longer window tells you |
|---|---|---|
| Is the underlying thesis still intact | Very little on its own | The base rate you should be updating from |
| Position sizing / risk right now | Useful — recent volatility is relevant to current risk | Less relevant; conditions can shift |
| Whether to change your overall belief about an asset | Feels decisive, usually isn’t | Where the actual evidence lives |
| Timing of an entry or exit within an existing thesis | Genuinely useful | Too slow to help with timing |
A five-minute check before you act on a streak
- Write down what you believed about this position or the broader market three months ago, before the current streak started.
- Ask what, specifically, changed — a real catalyst, or just the direction of price.
- Look at a chart zoomed out to at least one year, and place the current streak inside it. Does it look decisive at that scale, or like normal noise?
- If you can’t name a concrete reason beyond “it’s been going this direction,” treat the streak as data to log, not a signal to act on.
FAQ
Isn’t paying attention to recent price action just… paying attention to the market? There’s a real difference between using recent data as one input and treating it as the whole picture. Recency bias isn’t noticing that price moved — it’s the automatic upweighting of that move relative to longer-run data, so that three green weeks quietly out-argue three years of history in your head without you deciding that should happen. The fix isn’t ignoring recent price action; it’s deliberately putting it next to a longer window before it gets to set your expectations on its own.
How is this different from the gambler’s fallacy? They point in opposite directions. The gambler’s fallacy says a streak makes reversal more likely — “it’s down five days in a row, it has to bounce.” Recency bias says a streak makes continuation more likely — “it’s down five days in a row, so it’s just going to keep falling.” Both are wrong for the same underlying reason: an asset’s price has no memory of its own recent path, so neither a reversal nor a continuation is owed to you by the streak itself. It’s worth reading both, because which one you’re prone to often depends on whether you’re currently up or down.
Does this mean I should ignore short-term price action entirely and only look at long-term charts? No — short-term data is genuinely useful for some decisions, like managing risk on a leveraged position or timing an entry within a thesis you already hold. The problem is specifically letting a short window set your beliefs about where the asset or the market is headed, or how much risk to take, without deliberately checking it against a longer one. Use short-term data for execution; use longer-term data for conviction.
If this pattern is showing up as a bigger problem than any one decision — a habit of resizing risk based on whatever just happened rather than a stable plan — Overconfidence Bias covers the closely related habit of treating a recent string of correct calls as proof of skill, and how to check whether that’s actually true.