The Behaviour Gap

Two returns for the same fund

A fund reports a **time-weighted** return: the performance of a dollar invested at the start and left alone. An investor experiences a **dollar-weighted** return, which weights each period by how much money was actually there. When contributions cluster after good years and withdrawals cluster after bad ones, the second number is worse than the first — and the difference is entirely the timing of the household’s own decisions. The mechanism needs no forecasting ability to be damaging. Money tends to arrive after returns have been good, because that is when the fund is visible and comfortable, and it tends to leave after a fall, because that is when the pain is real. Every one of those decisions is individually reasonable and collectively expensive: the investor buys more shares at high prices and sells them at low ones, which is the mirror image of what the plan wants. Worth noting is what the gap is **not**. It is not the expense ratio (PF6), and it is not a market-timing skill the household lacks. It is the cost of a decision made in the middle of a strong year or a frightening month, and its size therefore depends on how much of the plan is left to that kind of decision. The fund against the investor in it — Fund, +25% then −20%: 0% over two years · Investor: $10,000, then $10,000 more: $20,000 in, $18,000 out · Investor’s total return: −10% · Annualised: about −5.1% ← Both numbers are true at once, which is the point: the fund delivered a flat two years and the household lost money inside it.

The evidence, including the part that disagrees

The most-cited estimate of the gap is Morningstar’s Mind the Gap series, which compares a fund’s time-weighted return with the dollar-weighted return of the average dollar invested in it. In the 2025 edition the average fund dollar earned about 7.0% a year against 8.2% for the funds themselves — a gap of roughly 1.2 points, and the report attributes a large part of it to poor timing. That figure is contested, and the contest is worth teaching because it shows how to read evidence rather than what to believe. Fulkerson, Jordan, Riley and Yan, in the Financial Analysts Journal in May 2026, re-examined the same and better data and concluded that poor timing by mutual fund investors costs them only about 0.10% a year — an order of magnitude smaller. Their argument is about measurement: the cash-flow data behind the larger numbers mixes investor choices with things investors do not control, such as automatic contributions, rebalancing and plan-level flows. The honest reading sits between the two, and the direction is not in doubt. Trading frequency is associated with worse outcomes: Barber and Odean’s study of 66,465 households found that the most active traders earned 11.4% a year against 17.9% for the market over the same period. Whether the gap for a given household is a tenth of a point or more than a point, it is caused by decisions that can be pre-empted, and the fixes are the same either way. The claims, side by side — Morningstar, Mind the Gap 2025: fund dollars earned ≈ 7.0% vs 8.2% for the funds · Fulkerson et al., FAJ May 2026: poor timing costs ≈ 0.10% a year on the same sample · Barber & Odean (66,465 households): most active traders 11.4% vs 17.9% for the market ← A study that measures the average investor does not predict any individual. What it establishes is that the mechanism is real and the fix is structural — not that a particular household is doomed to lose a point a year.

What closes the gap: machinery, not information

The gap is not caused by ignorance, so it is not fixed by learning more — a household that has understood everything in this subject can still sell at the bottom. What closes it is machinery that removes the decision from the moment: **automatic contributions** on a fixed schedule, a **written investment policy** that states the allocation, the bands, the wrapper order and the guardrails, and a **review cadence** of quarters rather than mornings. Two further devices do disproportionate work. The first is a **decision checklist** held in advance: what would make me sell this, what would make me buy more, and what evidence would change the plan? A written answer makes an exception visible when it is proposed. The second is **journaling** — the same discipline the trading curriculum asks for (Trading & Charts T19) — recording the thesis, the risk and the reason before a decision, so that later review compares what was expected with what happened rather than with what feels true in hindsight. The last device is calibration, which is a habit rather than a tool: before each decision, write down how confident you are and what would prove you wrong, then check the record. The plan’s predict-then-reveal segments are the small version of this, and the point is to notice whether confidence is tracking accuracy or merely tracking the strength of the feeling. Four pieces of machinery — Automation: contributions and withdrawals happen on a schedule, not a mood · A written policy: allocation, bands, wrapper order, guardrails — decided once · A checklist and a journal: the exception becomes visible before it is taken · A quarterly cadence: a plan reviewed every morning gets traded ← None of this requires a forecast, and none of it improves the expected return of the assets. It improves the return the household actually receives from them.

What the gap costs in dollars

A point of return sounds small because it is quoted as a rate and not as money, so it is worth converting once and remembering the answer. Take $100,000 invested at the funds’ own return of about 8.2% a year against the same money earning the investor return of 7.0%, and compound both for thirty years. The first grows to a little over a million dollars and the second to roughly seven hundred and sixty thousand — a difference of about three hundred thousand dollars, from a gap of just over one point a year. That conversion is the point: the behaviour gap is not a rounding error in a rate, it is a sum of money large enough to change what a retirement looks like. The scaling is worth seeing because the number grows faster than most people expect. At a tenth of a point — the size the 2026 critique suggests — the same thirty years on the same balance costs tens of thousands rather than hundreds of thousands. That is the honest reason to teach both estimates: the mechanism is real, and the size of the price depends on how much of the plan is left to judgement. A household that automates everything and reviews quarterly is buying the tenth-of-a-point version; one that acts on the news is exposed to the larger one. There is a related statistic that circulates widely and deserves its caveat. The claim that missing the ten best days of a decade destroys much of the return is arithmetically true, but the best days cluster near the worst ones, so a household invested through the declines had a good chance of being invested through the rebounds as well. The useful version of the point is not that being out of the market on the best days is catastrophic; it is that the days producing large gains are the ones that follow large falls, which is exactly when the impulse to be out is strongest. The statistic is a warning about behaviour rather than a proof about timing. The last conversion is the frequency of looking. Attention is the fuel of the mechanism, and the evidence on trading frequency points the same way as the horizon evidence: accounts that watch and act more often do worse than accounts that do not, for the mechanical reason that every look is an opportunity to react to noise. That does not make information harmful; it makes a daily cadence harmful, because a daily cadence guarantees that every drawdown is experienced as a decision rather than as an entry in a record. Reducing the cadence is a return input that requires no forecast at all. • Around 1.2 points a year over thirty years on $100,000 is roughly $300,000 — the gap in money, not in basis points. • The tenth-of-a-point version of the same gap is tens of thousands rather than hundreds of thousands. • Missing the best days is a real effect with a real caveat: those days cluster near the worst ones. • The impulse to be out is strongest exactly when the rebound days arrive, which is why the statistic is behavioural. • Look less often: attention is the fuel of the mechanism, and a daily cadence turns every drawdown into a decision. Convert every rate in this lesson into money before acting on it. A gap of one point and a cost of three hundred thousand dollars are the same fact stated twice, and only the second one changes behaviour.

Sources

• The gap between fund returns and investor returnsMorningstar — “Mind the Gap 2025” (fund dollar-weighted 7.0% vs 8.2% time-weighted)
• A critique of the size of the timing costFulkerson, Jordan, Riley & Yan, Financial Analysts Journal, May 2026 (poor timing costs ≈ 0.10% a year on the same sample)
• Trading frequency and household performanceBarber & Odean — “Trading Is Hazardous to Your Wealth” (66,465 households)
• Pre-commitment, automation and implementation intentionsBehavioural finance literature on commitment devices and planning