Key takeaways
- Morningstar's Mind the Gap 2025 found the average dollar in US funds and ETFs earned 7.0% a year over the decade to 31 December 2024, against an 8.2% buy-and-hold return — a gap of 1.2 percentage points a year.
- The famous 4% figure is DALBAR's: 3.49% a year for equity fund investors against 7.81% for the S&P 500 over the 20 years to 2011. DALBAR's own dollar-cost-averaging benchmark returned 3.17% — the average investor beat it.
- A 2026 Financial Analysts Journal paper rebuilt that calculation on the same sample and found poor timing cost investors only 0.10% a year. Most of the measured gap never touched anyone's wealth.
- The gap is concentrated, not uniform: 1.8 points a year in the most-traded funds against 0.8 in the steadiest, and 1.5 points in sector equity funds against 0.1 in all-in-one allocation funds.
- Among 66,465 brokerage households, the most active traders earned 11.4% a year while the market returned 17.9%.
The honest number is closer to 1% a year than 4%
If you have ever bought after a run-up, watched the thing fall, sold near the bottom and quietly wondered what that habit has cost you over a lifetime, this is the research that tries to answer it. The answer is smaller than the number you have probably heard.
The figure that circulates is around 4% a year, sometimes more. It comes from DALBAR, and it is almost certainly wrong. The most defensible estimate is Morningstar's Mind the Gap 2025, which looked at more than 25,000 US open-end funds and ETFs over the decade to 31 December 2024. The average dollar invested in those funds earned 7.0% a year. The funds themselves returned 8.2% a year to someone who bought at the start and never touched them. That difference — 1.2 percentage points a year — is the gap.
It is real money, and about a quarter of what the folklore claims. A serious challenge published in 2026 argues that even 1.2 points wildly overstates what bad timing costs.
The whole argument lives inside one distinction: dollar-weighted versus time-weighted
This is the part nobody explains, and it is the whole ballgame. When a fund reports "we returned 8.2% a year", that is a time-weighted return. It assumes one lump sum went in on day one and sat there. Every year gets equal billing, no matter how much money was in the fund that year.
A dollar-weighted return, what Morningstar calls the investor return, asks a different question: how did the average dollar do? Years when lots of money was invested count for more. If most people's money arrived after the good years, the average dollar earns less than the fund did, and a gap opens up.
Here is the catch, and it is fatal to the simple reading. A gap can open without anybody making a single bad decision. Harry Sit, in a guest post on Michael Kitces' site, works the example cleanly. Suppose a market doubles in year one, then goes flat for nine years. The ten-year time-weighted return is about 7.2% a year. Now suppose you did nothing clever or stupid, just invested $1,000 every year into an index fund. At the end you hold $11,000, and your dollar-weighted return is 1.7% a year. You lag the market by five and a half points, and you did nothing wrong. You didn't have much money invested during the one year that mattered.
Run the same market backwards, nine flat years then a double, and the same saver earns 12.3% a year, beating the market handsomely. Same behaviour. Same fund. Opposite verdict.
Morningstar says this out loud in the study itself: even "laudable practices like investing a portion of every paycheck or regularly rebalancing can open a gap", and the results should not be read as "a parable of 'dumb money'".
DALBAR's 4% is mostly a story about when the 1990s happened
DALBAR's Quantitative Analysis of Investor Behavior is the source of the scary statistic. The edition Kitces and Sit examined reported that over the 20 years to 2011, equity fund investors earned 3.49% a year while the S&P 500 returned 7.81%, a shortfall of 4.32 points a year, presented as the price of human frailty.
But look at the window: 1992 to 2011. Strong returns in the first decade, poor returns in the second — precisely the shape that mechanically produces a big negative gap for anyone who saves gradually. DALBAR compared a dollar-weighted investor return against a time-weighted index return, and attributed the whole difference to behaviour.
The killer detail is in DALBAR's own study. It also computed what a robotic investor who put a fixed sum in every year would have earned: 3.17% a year. The allegedly panic-prone average investor earned 3.49%. Against the right benchmark, the average investor beat mechanical dollar-cost averaging. That result is in DALBAR's data and not in DALBAR's headline.
The strongest critique says even Morningstar's 1.2% is mostly not timing
You could stop here, conclude that Morningstar's method has salvaged a smaller but honest number, and feel settled. In May 2026, four finance academics — Jon Fulkerson, Bradford Jordan, Timothy Riley and Qing Yan — published a paper in the Financial Analysts Journal titled "Bad Timing Does Not Cost Investors 15% of Their Funds' Returns". They pulled the same funds over the same decade and replicated Morningstar's gap almost exactly (they get 1.4 points; Morningstar reports 1.2).
Then comes the decomposition, using a method from a 2014 paper by Hayley. A dollar-weighted return is sensitive to cash flows in two directions at once. Money arriving today increases the weight on future returns: that is timing, and it genuinely affects how much you end up with. But it also retrospectively decreases the weight on past returns. That second effect, the hindsight effect, moves the reported number without moving a single dollar of anyone's wealth. If investors tend to buy funds that have just done well (and the evidence says they overwhelmingly do), the gap goes negative even if their purchases have no bearing on what happens next.
Applying Hayley's method forced them to rebuild the calculation — it requires estimating terminal portfolio assets, and on that amended basis the gap they decompose is 0.17 points a year, not 1.4. Of that, 0.07 points is hindsight and 0.10 points is genuine timing: roughly 1.2% of the funds' total return, not 15%. Across fund categories, about half of the gaps they estimate is hindsight. They are careful to say their timing estimate is "not derived from a direct deconstruction of the Return Gaps reported by Morningstar". Their conclusion is unusually direct regardless: "investors' timing has an economically insignificant impact on their aggregate wealth."
They also show how much the answer depends on plumbing. Morningstar changed its calculation method between the 2023, 2024 and 2025 editions. Holding the period constant at 2015–2024 and running all three, the average reported gap falls from 2.14 points, to 1.64, to 0.93, purely on how the benchmark portfolio is weighted and rebalanced. The 2025 method is the one the critics endorse.
Where the gap does show up, it is concentrated in the funds people trade
Aggregate numbers hide the useful finding. Sort funds by how volatile their cash flows are, a proxy for how much investors trade in and out, and the gap trends wider, from 0.8 points a year for the steadiest quintile to 1.8 points for the most-traded. That is the chart above. Sort by the volatility of the funds' returns instead and the shape repeats: 0.4 points for the calmest quintile, 2.0 for the wildest.
By fund type, the spread is starker. Allocation funds — the all-in-one, target-date sort that rebalance themselves and give you nothing obvious to react to — posted a gap of just 0.1 points a year. Sector equity funds, the concentrated thematic bets people buy after a hot run — the AI-infrastructure trade is the current specimen — posted 1.5 points. Broad US equity funds sat at 0.6 points. ETFs, for all their low costs, showed a wider gap (1.7 points) than open-end funds (1.2) — a symptom, Morningstar suggests, of how easy they are to trade.
Note what this survives. Even if you accept the critics entirely and treat the level of the gap as unreliable, the cross-section — systematically worse in the volatile, specialised, easily-traded stuff, from sector funds to a commodity sleeve like gold — still says something about where the temptation lives. It is not showing up in the boring products where there is nothing to time.
The behavioural cost is real; it just isn't measured well by fund flows
None of this means investors are rational. It means fund-flow arithmetic is a bad instrument for proving they aren't. Better evidence comes from watching real accounts. Brad Barber and Terrance Odean examined 66,465 households at a large US discount broker between 1991 and 1996. The average household earned 16.4% a year and turned over 75% of its portfolio annually. The households that traded most earned 11.4% a year, while the market returned 17.9%. That is a 6.5-point annual penalty on the heaviest traders, measured against the stocks they held, with no dollar-weighting sleight of hand available to explain it away.
As for why people trade at all, Shlomo Benartzi and Richard Thaler's "myopic loss aversion" offers the cleanest mechanism. Losses hurt roughly twice as much as equivalent gains feel good, and people evaluate their portfolios far more often than their horizon requires. Their simulations found the historical equity premium is consistent with investors behaving as though their time horizon were about one year, even when saving for a retirement decades away. Someone checking a 30-year portfolio through a 12-month lens sees far more losses than someone checking it once a decade, and loss aversion does the rest.
What this evidence cannot tell you
All the fund studies measure investors in aggregate. Fulkerson and his co-authors are explicit that their result says nothing about whether particular people, or the investors in any one fund, time badly. Plenty do. An aggregate of roughly zero is compatible with a lot of individual pain that cancels out.
The Morningstar data is US funds only, over one decade that held an unusually strong US equity market. Barber and Odean's households traded in the 1990s, before zero-commission apps, and held stocks directly rather than funds. Benartzi and Thaler ran simulations, not a natural experiment.
What would change the conclusion
Three things would move this. First, if a study linked investors' actual trades to their subsequent returns directly — rather than inferring behaviour from a dollar-weighted residual — and still found a large penalty, the folklore would be partly rehabilitated. Fulkerson's paper explicitly calls for that work. Second, a decade with the opposite return sequence to 2015–2024 would flip the hindsight component and could make the measured gap vanish, which would tell you the metric was always more about market shape than human weakness. Third, if the gap in automated, low-volatility products ever widened to match sector funds, the "temptation is concentrated" reading here would be wrong.
What survives is modest and worth having: the cost of your own timing is probably not 4% a year, may not even be 1%, and is concentrated in the volatile, specialised things you are most tempted to trade. The strongest version of the finding is not that investors are foolish. It is that the products which give you nothing to react to are the ones where nobody reacts. Whether in a spreadsheet or in something like LedgerTouch, the number worth knowing about yourself is how often you transact in your most volatile holdings — because that is the only place the research consistently finds a cost.
This content is for informational purposes only and does not constitute financial advice. Always do your own research or consult a qualified financial adviser before making investment decisions.