Key takeaways
- Across every category in Morningstar's Active/Passive Barometer, an investor picking an active fund at random faced a 60% chance of trailing its average passive peer, on data to 30 June 2026.
- Active US large-blend funds beat their passive peers 31.6% of the time over one year, 20.4% over five years and 10.5% over ten. The odds thin as the horizon lengthens.
- Goyal and Wahal studied 3,400 plan sponsors from 1994 to 2003. Managers those sponsors fired went on to earn excess returns no different from the managers hired to replace them.
- The FCA found that buying a UK best buy list fund and holding it 3 or 5 years worked about as well as tracking the list, on data from 2003 to 2015.
- Outside a wrapper the switch is a disposal. UK gains above the £3,000 annual exempt amount are taxed at 18% or 24% from 6 April 2026.
The fund you'd switch into faces almost the odds you're leaving behind
Your fund has trailed its benchmark for three years. The one at the top of the category table hasn't. Should you sell the first and buy the second?
On the evidence, the swap buys you very little, because the fund you'd move into faces roughly the same forward odds as the one you'd sell. Morningstar's US Active/Passive Barometer, published on 6 August 2026 with data to 30 June 2026, covers nearly 9,226 funds holding approximately USD 29 trillion, or about 67% of the US fund market. Its summary of the long horizons is blunt: "an investor selecting an active manager at random would face a 60% probability of them underperforming their average passive peer."
That 60% is the figure to hold on to, because it applies to the fund you're buying just as much as to the fund you're selling. Over the 12 months to June 2026, just over 40% of active US funds survived and beat their asset-weighted average passive composite, an increase of 7 percentage points on the year before. Neither number moves because one particular fund has had a bad run.
So the cost of switching funds isn't really the dealing charge. It's that you pay something certain, in tax and in time out of the market, to buy a change in odds the data doesn't show.
Fund performance persistence, measured by horizon
The chart above plots one category, US large blend, across six horizons ending June 2026. Take it for what it is. Each column is the share of active funds that survived that horizon and beat their average passive peer, not the same funds tracked forward from one good year, so it shows how the base rate moves as the window lengthens rather than whether individual winners repeat. Active funds cleared the bar 31.6% of the time over one year, 26.1% over three years, 20.4% over five, 10.5% over ten, 3.2% over 15 years and 4.9% over 20.
Read the ladder rather than any single rung. Short horizons flatter active funds, because one good year is reachable by luck. Stretch the window and the survivors thin out. US large growth is starker: 17.2% over one year, 5% over ten, and 0.8% over 20.
One more caution. Because each horizon is a separate cohort, the 20-year rate sits slightly above the 15-year one, which is a quirk of the cohorts rather than a revival. And the pattern isn't universal. In intermediate core bond, 66% of active funds beat their passive peers over one year and 43.1% over ten, so the decay is much gentler than it is in US equity.
Institutions ran this experiment on 3,400 plan sponsors
The cleanest test of switching funds after underperformance wasn't run on retail investors. It was run on pension plans, which have to document what they do and why.
Amit Goyal and Sunil Wahal examined "the selection and termination of investment management firms by 3,400 plan sponsors between 1994 and 2003", in the Journal of Finance in 2008. Sponsors hired managers after strong runs and fired them after weak ones, which is the institutional version of the decision in front of you. On the hiring side they found that "this return-chasing behavior does not deliver positive excess returns thereafter."
The firing side is the part that bears on this. Goyal and Wahal matched the sacking and the replacement into single round trips. They wrote that "if plan sponsors had stayed with fired investment managers, their excess returns would be no different from those delivered by newly hired managers." The switch was, on average, a lateral move that cost money to make.
Nearly a decade later, Bradford Cornell, Jason Hsu and David Nanigian pushed harder, in a Journal of Portfolio Management article published on 31 July 2017. Studying "portfolios constructed using the typical three-year evaluation periods employed by most pension funds", they found "that investors who chose managers with poor recent performance earned higher benchmark-adjusted returns than those who chose managers with superior recent performance." They don't oversell it. Their own stated implication is "that asset owners should focus on factors other than past performance."
The FCA tested a UK best buy list, and holding did as well as following it
UK evidence on the same question exists, and it comes from the regulator. In Annex 4 of its Asset Management Market Study interim report, published in November 2016, the FCA compared funds recommended on direct-to-consumer platform best buy lists with funds that weren't, using Morningstar Direct net returns "over the 2003-2015 period."
Two results matter. The first: "non-recommended share classes underperformed their benchmarks significantly, while recommended products did not significantly outperform their benchmarks." The list picked funds that beat the also-rans. It didn't pick funds that beat the index.
The second is closer to the switching question, because the FCA ran the analysis twice. Once assuming you keep your portfolio in line with the list as it changes, which means selling whatever drops off it. Once assuming you buy a fund when it first appears and hold it "for either three or five years, irrespective of whether the share classes in question are de-listed over this horizon." The verdict: "we find qualitatively similar results in the bottom panel as for the top panel." Keeping up with the list added nothing over ignoring it after the first purchase.
The magnitudes are small, in the way monthly regression constants are. Non-recommended share classes ran at -0.10 monthly percentage points against benchmark, recommended products at 0.01. The FCA notes that "a monthly figure of 0.01 in the table equates to a net excess return of approximately 12 basis points on an annualized basis", which puts the non-recommended drag at roughly 1.2% a year.
The FCA's final report, published in June 2017, gave the industry-wide version. Active funds for sale in the UK "outperformed benchmarks before charges were deducted, but underperformed benchmarks after charges on an annualised basis by around 60 basis points".
The cost of switching funds is mostly tax, and the wrapper decides it
Everything above concerns the odds. The cost of switching funds is a separate quantity, and it's the part you can compute before you act.
Inside a stocks and shares ISA or a pension, a switch is close to free. The ISA subscription limit for the 2026 to 2027 tax year is £20,000, and gains inside the wrapper aren't taxed, so selling one fund and buying another triggers nothing beyond dealing costs and the days your money spends in cash.
Outside a wrapper it's a disposal. Gains above the annual exempt amount, which GOV.UK gives as £3,000, are charged at 18% for a basic rate taxpayer and 24% for a higher or additional rate taxpayer on gains made from 6 April 2026. On a holding sitting on a £10,000 gain, a higher rate taxpayer would face £1,680 of tax on the £7,000 above the allowance. That's a certain 16.8% of the gain, paid up front, to move into a fund whose forward odds the evidence puts close to the one just sold.
Which is why selling an underperforming fund is a different decision in a pension from what it is in a general investment account. Same fund, same disappointment, materially different arithmetic.
The strongest counter-case: fees persist even when returns don't
There's a serious objection to all of this, and the same Morningstar data supports it.
The active fund success rate is not uniform, and the thing that moves it most is cost. Over the ten years to June 2026, active US large-blend funds in the cheapest fee quintile beat their passive peers 23.1% of the time. The most expensive quintile managed 9.1%. Morningstar puts it plainly: "the success rates of the cheapest funds are more than double those of their most expensive active peers", and in several categories a low-cost active fund crosses 50%.
That matters because it separates two decisions people run together. Switching because a fund lagged is acting on a signal the data says is weak. Switching because you're paying for something available far cheaper is acting on a signal that has held up across every horizon in the table. One is a bet on persistence, the other is arithmetic. Our piece on the total cost of owning a fund takes that layer by layer.
A second objection comes from the regulator rather than the industry. The FCA's finding was that "there is little evidence of persistence in outperformance in academic literature, and where performance persistence has been identified, it is persistently poor performance." Chronic laggards do repeat, in other words, even where chronic winners don't. Someone leaving a fund that has trailed for a decade is acting on the one form of persistence the FCA did find, and its final report also noted that "we found some evidence of persistent poor performance of funds."
Third, investors are not as hopeless at this as the folklore suggests. Morningstar found the asset-weighted return for active managers beat the equal-weighted return in 16 of the 20 categories it covers, meaning money sat in the better funds rather than the worse ones. Two of the four categories where investors chose badly, global real estate and corporate bond, both recorded 10-year success rates above 44%.
What the trading itself costs, measured rather than assumed
Morningstar's Mind the Gap study puts a number on what investors give up by transacting. Over the decade to 31 December 2024, "the average dollar invested in US mutual funds and exchange-traded funds earned 7.0% per year", against an aggregate total return of 8.2% for the same funds. That 1.2 percentage point gap is, in Morningstar's words, "equivalent to around 15% of funds' aggregate total return over this period".
The link to trading shows up directly. Funds with the most stable cash flows, meaning the least investor trading, lagged their own total return by 0.8% a year, "which was 1-percentage-point narrower than the gap for funds with the most volatile cash flows." The busiest quintile therefore gave up roughly 1.8% a year against the funds it held.
Morningstar is careful about the cause, and doesn't pin all of it on panic. Its own framing is that "there can be debate about what causes investor return gaps", that the gap "appears to reflect a number of factors", and that whatever the cause it behaves like "a rather persistent cost, not unlike fund expense ratios". We look at the flow side of the same question in our piece on performance chasing.
What this evidence cannot tell you
Start with the sample. The Barometer and Mind the Gap are US datasets. The FCA work is UK, but it ends in 2015 and it studies platform best buy lists rather than individual switching decisions. Goyal and Wahal studied institutions between 1994 and 2003, and a pension committee firing a segregated mandate is not a person selling a fund in an ISA. The direction of the finding holds across four independent datasets. The magnitude travels less well.
None of it is a forecast either. A success rate is a base rate over a window that has already closed. It says what share of funds cleared the bar in a period that has happened, and it doesn't identify which fund clears it next.
Age is a limit in itself. All three of the named academic and regulatory studies are at least nine years old, and one of them covers 1994 to 2003. Their samples predate a good deal of what has happened to fund charges since, and the fee-quintile split above is the reminder that charges move these odds more than anything else measured here.
The persistence table is also a category average, and it hides enormous spread. Morningstar's point about US large growth is that the 5% who won didn't win by much, while "the potential downside from picking a poor performer was enormous."
And none of these studies model your tax position, your platform's charges or the reason you bought the fund in the first place. The SPIVA scorecard and the Barometer both count funds, not people.
What would change the conclusion
If the fund changed rather than its luck. Every study above measures a fund whose strategy stayed put while its returns wobbled. A manager departure, a style drift or a fee rise is new information about the fund, not another draw from the same distribution. None of this evidence speaks to that case.
If the categories keep diverging. The distance between intermediate core bond at 43.1% over ten years and US large growth at 5% is not noise. If the bond and property numbers hold at those levels through another cycle, the idea that active funds mostly lose stops being a fair description of the whole market.
If switching stopped costing anything. Rates of 18% and 24% from 6 April 2026 are what make the untaxed wrapper matter, and they apply only outside it. If gains tax on funds went to nothing, the cost of switching funds in a general investment account would look like the cost inside an ISA, and the whole question would collapse back to the odds, where the evidence says there isn't much in it.
The number worth watching isn't your fund's rank in its category, which on this evidence is a poor guide to the decade ahead. It's how far the fund has drifted from the job you bought it for, and what a change of horse costs you after tax. LedgerTouch tracks the second. The first is a question only you can answer, and our guide to fund fees covers the one input that has predicted anything at all.