Key takeaways
- Of the 505 top-quartile US domestic equity funds of 2021, 35.05% were still top-quartile a year later, 0.40% after two years, and none at the end of 2024 or 2025.
- Across two non-overlapping five-year periods, 11.79% of top-quartile US equity funds stayed on top, while 31.22% fell all the way to the bottom quartile.
- Morningstar's study of nearly 9,000 funds puts three-year and five-year top-quartile repeat odds close to the 25% that pure chance would produce.
- Failure is stickier than success: 23% of bottom-quartile US equity funds were merged or liquidated within five years, against 10% of top-quartile funds.
- The pattern crosses the Atlantic: 3.75% of the UK equity funds that were top-quartile in 2021 were still top-quartile at the end of 2025.
The five-year answer: past winners almost never repeat
Does past fund performance predict future performance? It's the question sitting under every five-star rating, every best-buy list and every factsheet table. Two large datasets, S&P Dow Jones Indices' US Persistence Scorecard with data to December 2025 and Morningstar's October 2025 persistence study, put numbers on it. The numbers are stark.
The S&P scorecard tracked the 505 actively managed US domestic equity funds whose 2021 returns placed them in the top quartile of their peers. One year on, 35.05% were still there. After two years, 0.40%. At the end of 2024, and again at the end of 2025, none remained. The 164 large-cap funds in that cohort didn't even take that long: 20.12% survived the first year and none survived the second.
Morningstar ran the same test category by category across nearly 9,000 funds holding roughly $24 trillion, about two-thirds of the US fund market as of June 2025. In 11 of the 20 categories it examined, every top-quartile fund from 2021 had dropped out of the top quartile before three years had passed.
Note what the question is and isn't. This piece isn't about whether active funds beat the index; the SPIVA scorecard covers that ground. Persistence asks something narrower: whether funds that beat their own peers keep doing so. On the evidence, they don't, and the rest of this guide is about how we know, and where the exceptions live.
Fund performance persistence in one table: the quartile transition matrix
The cleanest test of fund performance persistence is a quartile transition matrix. Rank every fund over one period, sort them into quarters, then rank the survivors again over the next, non-overlapping period. If skill drives results, top-quartile funds should cluster in the top quartile again. If luck drives results, each starting group should scatter roughly evenly, about 25% to each quartile, before accounting for funds that die.
S&P's five-year matrix is the chart above. Of the 458 US domestic equity funds in the top quartile for the five years to December 2020, 11.79% were top-quartile again over the five years to December 2025. Some 16.81% landed in the second quartile, 8.73% in the third, and 31.22% fell to the bottom quartile. Another 9.83% were merged or liquidated, and 21.62% changed style. A top-quartile fund was almost three times as likely to finish in the bottom quartile as to stay on top.
Morningstar's matrices, built on non-overlapping periods in data running to December 2024, tell the same story with less drama. Prior-year top-quartile active funds repeated 24.7% of the time, and landed in the bottom quartile 27.3% of the time. Over three-year periods the repeat rate was 22.7%; over five-year periods, 19.9%. Morningstar's own reading is that the probability of a top-quartile active fund repeating over these horizons is close to random chance, with the shortfall from 25% mostly explained by funds that closed or changed category.
Three-year windows look kinder, and momentum explains most of the kindness
Shorter windows flatter past performance. Of the top-quartile large-cap funds of 2023, 29% were still top-quartile two years later, against the 0% recorded for the previous cohort. Of the top-half large-cap funds from 2023, 49% were still top-half at the end of 2025, roughly double the 25% random chance would suggest.
Morningstar finds the same short-horizon glow: over one-month, three-month and six-month windows, a top-quartile fund's chance of staying top-quartile ran just shy of 30%, then decayed beyond a year. That decay profile matches the momentum effect in the underlying stocks, where recent winners keep winning for a while and then stop.
That reading has a long pedigree. Mark Carhart's 1997 Journal of Finance study, built on a survivorship-bias-free sample, concluded that common factors in stock returns and fund expenses "almost completely explain persistence in equity mutual funds' mean and risk-adjusted returns", and that the earlier "hot hands" result was mostly one-year stock momentum wearing a manager's name. A fund that owns last year's winning stocks scores as a persistent winner for exactly as long as those stocks run. Money tends to arrive at precisely that point, which is the subject of our guide to performance chasing and what happens after it.
What actually persists is failure
Carhart's other finding was that the only persistence his model couldn't explain away was concentrated in strong underperformance by the worst funds. Twenty-eight years later, both datasets still show it.
In the S&P data, 23% of bottom-quartile US domestic equity funds were merged or liquidated within the following five years; for top-quartile funds the figure was 10%. In Morningstar's five-year matrix, a bottom-quartile fund's probability of staying in the bottom quartile or being merged or liquidated rose above 50%, and bottom-quartile funds were more than twice as likely to shut down as funds from the top two quartiles.
Fees sharpen the picture. Morningstar splits its matrices by cost and finds a poorly performing expensive fund carries around a 13% higher chance of staying in the bottom quartile or closing within a year than a cheap one. Over three years, cheap funds' odds of staying among the top performers run about 6% higher than expensive funds'. Past returns say little about future returns, but the charge you pay is contractual, which is why fund fees predict outcomes in a way past performance doesn't.
UK and European funds tell the same story
None of this is a quirk of the US market. S&P's Europe Persistence Scorecard, also with data to December 2025, found that just 16% of the 408 funds whose five-year record placed them in the top half of the Global Equity category at the end of 2020 stayed top-half over the next five years. In no category did more than half of the funds repeat top-half status, against the 50% chance would produce.
For a UK reader the sterling numbers are the sharpest. Of the 80 UK equity funds in the top quartile for 2021, 18.75% were still there a year later and 3.75% at the end of 2025. The UK five-year transition matrix is bleaker than the US one: 2.90% of top-quartile UK equity funds repeated, 46.38% dropped to the bottom quartile, and 23.19% were merged or liquidated.
The best case for persistence: calm markets, cheap funds and bond categories
The counter-case deserves its numbers too, because the same scorecards contain them. In Europe, the share of 2021's top-half equity funds still compiling a top-half record ran 53% after two years, 31% after three, 23% after four and 16% after five. Random chance would give 50%, 25%, 13% and 6%. At every horizon, more funds kept winning than luck alone explains. S&P's own suggested reason is momentum: its momentum indices posted strongly positive excess performance over recent five-year periods across the S&P Global LargeMidCap, S&P 500 and S&P Europe 350, so managers who happened to hold persistent stocks persisted with them.
Market regime matters as well. Morningstar plots repeat rates against the VIX volatility index and finds that between 2016 and 2019, with volatility subdued, the one-year probability of a top-quartile fund repeating frequently exceeded 30% and occasionally approached 40%. When the VIX spiked in early 2020, it dropped to around 15%.
Asset class matters too. In Morningstar's data, the intermediate core bond category was the standout: six of the 35 top-quartile funds of 2021, about 17%, were still top-quartile at the end of 2024. S&P's fixed income tables agree at the two-year horizon, where 31% of top-quartile investment-grade intermediate funds from 2023 repeated in both following years. Even there, the five-year rates fall to single digits.
So the strongest honest claim for fund performance persistence is conditional: over horizons under three years, in calm markets, in cheaper funds, and in some bond categories, past rankings carry more signal than chance. The referee's problem is what that signal is. Carhart's decomposition attributes it to momentum and expenses, not manager skill, and S&P notes that large-cap repeat rates over five years actually fell below what a random distribution would produce, which it reads as evidence that outperformance, when it occurs, tends to be luck.
Why every factsheet carries the same warning
This evidence is why the warning on your factsheet is not boilerplate. Since January 2021, FCA rules in COBS 4.6.2R have required that any promotion showing past performance carries "a prominent warning that the figures refer to the past and that past performance is not a reliable indicator of future results", covers the preceding five years where possible, and uses complete 12-month periods. The SEC's investor education site is blunter: it's not uncommon for a fund to post better-than-average performance one year and mediocre or below-average performance the next, which is why the SEC requires funds to tell investors that past performance does not necessarily predict future results.
What this evidence cannot tell you
These are relative rankings inside categories, not statements about what you earned. A fund can sit in the top half of its category for a decade while the whole category trails its index; in 2025 alone, 79% of active large-cap US equity funds underperformed the S&P 500. Persistence data can't tell you a category was worth being in.
The windows are specific, not eternal. The five-year cohorts above start in 2021 or span 2015 to 2025, periods containing a pandemic crash, a violent rate cycle and an unusually narrow bull market. Morningstar's volatility work shows repeat rates are regime-dependent, so a different decade could produce different tables, in either direction.
The buckets hide detail. In S&P's five-year matrix, 21.62% of top-quartile funds "changed style", which is a reclassification, not a failure, and it mechanically shrinks every repeat rate. Quartile cut-offs are arbitrary too: a fund missing the top quartile by a basis point counts the same as one that collapsed.
And the referees have interests. S&P Dow Jones Indices sells the indices active funds are judged against, and Morningstar sells fund data and ratings. The tables are reproducible, and the two firms' methods reach the same answer independently, but neither is a disinterested party, and Morningstar's figures run to December 2024 while S&P's run to December 2025.
What would change the conclusion
Three things would genuinely reopen the question. First, repeat rates holding above chance across cohorts, not within one: if the 29% two-year large-cap figure recorded for the 2023 cohort showed up again for the 2024 and 2025 cohorts, and again after that, the luck explanation starts to strain. The scorecards are annual, so this is checkable every year.
Second, persistence surviving Carhart's controls. The current result is that momentum and expenses absorb nearly all of it. A study finding repeat winners after stripping both out would be evidence of the thing investors are actually paying for, which is skill.
Third, the failure signal weakening. The most reliable pattern in all of this data is that bad, expensive funds stay bad or disappear. If bottom-quartile funds started recovering at rates chance can't explain, the case for using past performance at all would strengthen, just from the opposite direction to the one fund marketing assumes. Until then, the tables reward reading a fund's costs over its trophy cabinet; LedgerTouch shows the ongoing charge on every holding beside its returns, which is the order this evidence suggests reading them in.