Investor psychology Lesson 7 of 15
Articles · Behavioral finance
Recency bias and performance chasing: buying what just went up
Why the last year feels like a forecast, what surveys and fund flows show about investors chasing returns, and what SPY and ARKK data say about how well it works.
On this page
Recency bias is giving the most recent stretch of experience too much weight when judging what comes next. After a strong year investors expect a strong year. After a crash they expect another. Performance chasing is what the belief looks like in an account: money moving into whatever has just done best.
It is one of the few biases where the cost can be read directly from public data, because fund companies report both what a fund earned and, through its flows, what its investors earned. The two differ, and the difference has a consistent sign. This lesson covers the survey evidence on what investors expect, the fund-flow evidence on what they do, a test on 32 years of SPY data, and one fund whose history shows the whole pattern.
What investors expect
For decades economists argued about expectations without measuring them. In 2014 Robin Greenwood and Andrei Shleifer gathered six surveys of investor expectations, from the Gallup poll of individual investors to a survey of chief financial officers, covering 1963 to 2011.
They found three things. The surveys agreed with each other, so they were measuring something real. Expected returns rose after the market had gone up and after its level was high. And the surveys were negatively related to what financial models said future returns would be, and to the returns that followed. When investors were most optimistic, subsequent returns tended to be lower.
Experience over a lifetime works the same way. Ulrike Malmendier and Stefan Nagel studied decades of the US Survey of Consumer Finances and found that people who had lived through poor stock returns were less willing to take financial risk, less likely to own stocks and held less in them when they did. Recent years counted more than distant ones, especially for the young, who have fewer years to average over.
What investors do
Erik Sirri and Peter Tufano showed in 1998 that mutual fund flows respond to past performance asymmetrically. Funds that performed very well in the previous period receive disproportionately large inflows. Funds that performed badly are not abandoned to the same degree. Investors chase winners more eagerly than they flee losers.
Andrea Frazzini and Owen Lamont followed the money one step further, into the stocks the funds held. Stocks owned by funds with heavy inflows went on to underperform stocks owned by funds with outflows. They called this the "dumb money" effect and estimated that by reallocating across funds, investors lowered their own returns by roughly 1% a year.
Morningstar measures the result each year in its Mind the Gap study, which compares a fund's reported total return with the investor return, a dollar-weighted figure that accounts for when money actually entered and left.
| Ten years ended December 2025 | Investor return | Fund total return | Gap |
|---|---|---|---|
| All US funds and ETFs | 8.7% | 9.9% | −1.2 |
| US equity funds | 12.8% | 13.3% | −0.4 |
| Allocation funds | 7.8% | 8.4% | −0.7 |
| Sector equity funds | 9.4% | 10.6% | −1.2 |
| Alternative funds | 2.6% | 4.1% | −1.6 |
Source: Morningstar, Mind the Gap 2026. Gaps are as published and reflect rounding.
Two readings are worth drawing out. The gap was smallest for plain US equity funds and all-in-one allocation funds, where investors tend to contribute steadily and leave the money alone. It was largest for narrow and volatile categories, the kind that are bought because of a recent run. And a gap of 1.2 points a year is about an eighth of the total return, lost without any fee being charged.
A test on real data: does last year predict next year?
If the recent past were a forecast, the last 12 months' return would tell you something about the next 12. The figure plots every month from March 1994 to November 2025 for SPY, with dividends included: the return over the year before on the horizontal axis and the return over the year after on the vertical axis.
The cloud has no shape. The correlation is 0.07.
| After a 12-month return of… | Cases | Average next 12 months | Next 12 months negative |
|---|---|---|---|
| Any | 381 | 12.4% | 18% |
| More than +20% | 117 | 14.3% | 14% |
| 0% to +20% | 194 | 13.7% | 13% |
| Below 0% | 70 | 5.7% | 40% |
| Below −15% | 29 | 16.3% | 21% |
The table needs careful reading, because it does not say what either the chaser or the contrarian would like. After a strong year the next year was about average. After a mildly negative year it was weaker, because the two long bear markets in the sample (2000–2002 and 2007–2009) each lasted more than a year. After a deeply negative year it was above average, because those are the months near the bottoms. The windows overlap heavily, so these 381 points contain only about 32 independent years and two major bear markets. The honest conclusion is the weak one: for a broad index, the last year tells you very little about the next.
One fund, the whole pattern
The ARK Innovation ETF (ARKK) is the clearest recent example of the gap between a fund's return and its investors' experience. The figure shows its month-end price and, underneath, the dollar value of shares traded each month, a rough gauge of how much attention and money the fund was attracting.
- From its 2014 launch to the end of 2019 the fund's price went from about $20 to $50. Only 0.6% of all the dollars ever traded in it changed hands in those five years.
- In 2020 the price rose 149%. Attention followed the return: monthly dollar volume peaked at about $47 billion in March 2021, two months after the price peaked at $137.44.
- In 2021 the fund lost about 24% and in 2022 another 67%, closing that year at $31.24.
- About 40% of all dollars traded in the fund's history changed hands in the two years from July 2020 to June 2022, across the peak and the fall.
Trading volume counts both sides of every trade, so it is not the same as money flowing in. Morningstar measured that directly. In December 2021 it estimated ARKK investors' five-year dollar-weighted return at 9.9% a year, far below the fund's reported total return over the same five years, and a 12% loss for the average dollar over the latest twelve months, when the fund itself had lost 4.3%. By mid-2023 it reported the fund's annualized total return since inception was still 9.6%, while the dollar-weighted return since inception was sharply negative: the fund had made money and the average dollar in it had lost money.
The fund did what it said it would. The loss came from the timing of the money, and the timing came from extrapolating a spectacular year.
What it looks like in an account
An illustrative investor, Jordan, reviews a retirement account each January and moves 20% of it into the prior year's best-performing fund on the plan's menu.
| January | Best fund last year | Its return last year | Its return in the year Jordan held it |
|---|---|---|---|
| Year 1 | Technology | +48% | −31% |
| Year 2 | Energy | +57% | −4% |
| Year 3 | Technology | +55% | +22% |
Jordan was right once in three years and lost about 19% in total on the slice that was moved (0.69 × 0.96 × 1.22 = 0.81, against the menu's balanced fund, which was up over the same years). These numbers are illustrative, but the shape is the one in the Morningstar table: the decision rule "buy what led" guarantees buying after the rise, and the sectors that lead by 50% in a year are the ones that can fall 30% in the next.
Is momentum not real?
It is, and the difference matters. Narasimhan Jegadeesh and Sheridan Titman showed in 1993 that stocks that had outperformed over the past 3 to 12 months kept outperforming over the next 3 to 12, by about 1% a month before costs. Werner De Bondt and Richard Thaler had shown in 1985 that over three to five years the pattern reverses: past losers beat past winners by 24.6% over the following three years.
Both findings are about large, diversified portfolios of stocks rebalanced by rule, measured before trading costs, and with long losing stretches. Neither describes an individual moving savings into last year's best fund. A performance chaser typically acts on a horizon of a year or more, after the story has reached the news, in one concentrated holding. That is closer to the horizon where returns reverse than to the one where they persist. 52-week-high momentum covers what a rule-based version involves.
What helps
No intervention has been shown to remove recency bias. The patterns in the data do suggest where the damage is smallest.
- Automate contributions. The smallest gaps in the Morningstar study are in funds held through regular, automatic investing. See dollar-cost averaging.
- Decide the allocation first, then rebalance to it. Rebalancing moves money from what has risen to what has lagged. It is the mechanical opposite of chasing.
- Read a full cycle. Before buying anything because of its one-year return, look at its worst year and its return over a period that includes a bear market.
- Wait. A written rule such as "nothing I first heard about this month" costs little when the idea is good and saves a great deal when it is not.
How Finabase helps
The comparison tab lists 1-week, 1-month, 3-month, 6-month, year-to-date, 1-year and 2-year returns for several symbols in one table, which makes it plain how different the same fund looks depending on the window you pick. The sector page ranks sectors by return over several periods, and the leaders over one period are often not the leaders over the next. The SPY chart set to MAX shows the full history rather than the stretch that is freshest in memory.
Frequently asked questions
What is recency bias in investing?
Recency bias is expecting the near future to resemble the recent past and weighting the last few months more than the long record. It leads investors to buy after prices have risen and sell after they have fallen.
What is the behavior gap?
It is the difference between the return a fund reports and the lower return its investors actually earned, because of when they bought and sold. Morningstar put it at about 1.2 percentage points a year for US funds over the ten years to 2025.
Is chasing performance the same as momentum investing?
No. Momentum strategies buy a diversified basket of recent winners by rule, hold for months and rebalance, accepting trading costs and sharp reversals. Performance chasing is an undiversified, late and unplanned version that tends to arrive near the end of a run.
Does past performance predict future returns?
For a broad index, barely: SPY's last 12 months explained almost none of its next 12. For actively managed funds, top rankings rarely persist once costs are counted. Past performance says more about an investment's risk than about its next year.
Summary
Recency bias turns the last year into a forecast. Surveys show investors expect the most after markets have risen, flows show money moving disproportionately into recent winners, and Morningstar's study puts the resulting gap between fund returns and investor returns at about 1.2 points a year. On 32 years of SPY data the past year's return had a correlation of 0.07 with the next, and ARKK shows the mechanism in one fund: the price rose 149% in 2020, most of the dollars arrived afterwards, and the average dollar lost money in a fund that had gained. Automatic contributions and rebalancing to a chosen allocation are the habits associated with the smallest gaps.
Check your understanding
5 questions on this lesson. Each answer comes with the reason, and nothing is scored.
Surveys show investors expect the highest returns after the market has risen strongly. What did Greenwood and Shleifer find about those moments?
- Models of expected return, and the returns that followed, tended to be lower
- Returns that followed were reliably higher
- Investors’ expectations had no connection to past returns
Show the answer
A. Models of expected return, and the returns that followed, tended to be lower Expectations rose with past returns and were negatively related to model-based expected returns and to subsequent returns. Optimism peaked when prospects were weakest.
A fund reports a ten-year total return of 9.9% a year, but its investors’ dollar-weighted return was 8.7%. Where did the 1.2 points go?
- To the fund’s expense ratio
- To the timing of purchases and sales: more money was in the fund during its weaker periods than during its stronger ones
- To taxes
Show the answer
B. To the timing of purchases and sales: more money was in the fund during its weaker periods than during its stronger ones Total return assumes one investment held throughout. Investor return weights each period by the money actually in the fund, so buying after gains and selling after losses lowers it.
For SPY since 1993 the correlation between the last 12 months’ return and the next 12 months’ return is 0.07. What follows?
- A strong year makes a weak year likely
- A strong year makes another strong year likely
- For a broad index the last year carries almost no information about the next
Show the answer
C. For a broad index the last year carries almost no information about the next A correlation near zero supports neither the chaser nor the person calling a top. The sample also holds only about 32 independent years.
ARKK’s price rose from about $20 to $50 between 2014 and 2019, yet only 0.6% of all dollars ever traded in it changed hands in those years. What does that illustrate?
- The fund was closed to new investors
- Attention and money arrived after the spectacular return, not before it
- Trading volume has no relation to investor interest
Show the answer
B. Attention and money arrived after the spectacular return, not before it Most of the money met the fund near its peak, which is how a fund with a positive return since launch can leave its average dollar with a loss.
How does a rule-based momentum strategy differ from chasing last year’s best fund?
- It does not differ
- It holds a diversified basket chosen by rule over a 3 to 12 month horizon and rebalances, where chasing is one late, concentrated purchase
- Momentum strategies never lose money
Show the answer
B. It holds a diversified basket chosen by rule over a 3 to 12 month horizon and rebalances, where chasing is one late, concentrated purchase The research finding is about broad portfolios over short horizons before costs. Over three to five years the pattern reverses, which is closer to the horizon performance chasers act on.
Exercise
One task to do in Finabase. Nothing is graded or saved.
Open the Comparison tab for ARKK and add SPY and one fund you own or follow. In the table, read across the rows from 1-week return to 2-year return. For each symbol, note which window makes it look best and which makes it look worst. Which symbol would you have picked from the 1-month row alone, and would the 2-year row agree?
Open ARKK comparisonWhat you should find
The ranking usually changes from row to row, and a volatile fund changes most. If one window made a fund look far better than the rest, that window is the one a headline would quote. No row is the forecast: each is one sample of a past that could have been drawn differently.
Sources and further reading
- Greenwood and Shleifer — Expectations of returns and expected returns, Review of Financial Studies 27(3), 2014
- Sirri and Tufano — Costly search and mutual fund flows, Journal of Finance 53(5), 1998
- Frazzini and Lamont — Dumb money: mutual fund flows and the cross-section of stock returns, Journal of Financial Economics 88(2), 2008
- Malmendier and Nagel — Depression babies: do macroeconomic experiences affect risk taking? Quarterly Journal of Economics 126(1), 2011
- De Bondt and Thaler — Does the stock market overreact? Journal of Finance 40(3), 1985
- Jegadeesh and Titman — Returns to buying winners and selling losers: implications for stock market efficiency, Journal of Finance 48(1), 1993
- Morningstar — Mind the Gap 2026 (Ptak), ten years ended December 31, 2025
- Morningstar — Arnott, ARKK: an object lesson in how not to invest (December 2021) and Why ARKK shareholders are still underwater (June 2023)
- Finabase chart data — SPY monthly closes and dividends, February 1993 to October 2026; ARKK monthly closes and volume, November 2014 to October 2026
