Stock Market Mistakes, Ranked by What They Cost
Key takeaway
Ordered by measured cost rather than by how often they are named, the expensive stock market mistakes are the quiet recurring ones: paying a percentage point of fees you do not need, trading often, arriving after the rise and selling into the fall. The dramatic ones, single stocks and day trading, cost less on average and more in the tail.
Every list of stock market mistakes is ordered the same way: by how easy the mistake is to name. Lack of research comes first because it sounds like the root of everything, emotional investing second because everyone has felt it, and fees somewhere near the bottom because a fee is not a story. None of that ordering has anything to do with what the mistakes cost. Several of them have been measured, on real accounts and over stated periods, and when you put the measurements in a column and sort by it, the list rearranges itself. The mistakes that cost most are the quiet, recurring ones that nobody experiences as a mistake at all.
There is a second thing the lists skip, and it is the reason a single ranking is impossible. Some mistakes are a small, near-certain charge every year, and compounding does the rest. Others are a one-off bet with a bad distribution: most of the time they cost nothing, and occasionally they cost the account. A percentage point of fees and a concentrated position in one company cannot be placed on the same scale, so this article uses two, and says which is which. The one thing it never does is add the numbers together — our companion piece on the six measured biases explains why that sum is meaningless, and this page inherits the rule.
Recurring mistakes: what each one costs per year
These are the mistakes that show up as a gap in annual return. Where the figure comes from a study, the table says what was compared and on which sample; where it is our own arithmetic, it says so and states the assumptions.
| Mistake | What was measured | Cost per year | Source and scope |
|---|---|---|---|
| 1. Trading often | Net annual return of the households that traded most vs those that traded least | 7.1 points (11.4% vs 18.5%) | Barber & Odean (2000), 66,465 US brokerage households, Feb 1991–Jan 1997. Almost all of the gap is transaction cost, not stock selection |
| 2. Paying for management you do not need | Equal-weighted average expense ratio of active US equity funds vs a broad index fund at 0.03% | 0.97 points (1.00% vs 0.03%) | Morningstar, 2026 US Fund Fee Study, 2025 data; own arithmetic for the difference |
| 3. Arriving after the rise, leaving after the fall | Return earned by the average dollar in US funds vs the funds’ own total return | 1.2 points (8.7% vs 9.9%) | Morningstar, Mind the Gap 2026, roughly 23,000 US funds and ETFs, decade to 31 Dec 2025 |
| 4. Selling into a fall and staying out for months | S&P 500 annual return with a rule that sells after any −4% day and returns after 60 or 120 sessions, vs staying invested | 0.8 to 1.8 points, depending on the window | Own calculation on daily closes of the S&P 500 price index, 1950–2025 and 2006–2025 (details below) |
| 5. Holding cash for years | Purchasing power lost at 3% inflation | 3 points of real value a year: 25.6% of purchasing power in ten years, 44.6% in twenty | Own arithmetic at a constant 3%; the actual rate is whatever your country’s index says |
Two things about that order. The largest number, trading frequency, was measured in the 1990s when commissions were a large part of the friction; commissions on US shares have since been competed away, so the 7.1 points should be read as an upper bound, not a current figure. The bid-ask spread, the tax triggered by each realised gain, and the tendency to buy what has just caught your attention are all still there. And the smallest study-measured number, the 1.2-point timing gap, is the one most often quoted at investors, usually inflated to three to five points by a different and looser comparison. Both points are argued in full in the companion piece.
Tail mistakes: what each one costs when it goes wrong
These do not produce a steady gap. They produce a distribution with a long bad side, and the honest way to state their cost is as a probability or as the size of the hole.
| Mistake | What was measured | The figure | Source and scope |
|---|---|---|---|
| 6. Owning one stock instead of the market | Share of US common stocks whose lifetime buy-and-hold return beat one-month Treasury bills | 42.6%; the best-performing 4% of companies account for the entire net gain of the US market | Bessembinder (2018), roughly 25,300 CRSP-listed companies, 1926–2016 |
| 7. Day trading | Outcome of every individual who began day trading Brazilian equity futures and persisted at least 300 days | 97% lost money; 0.4% out-earned a bank teller; the best of them made US$310 a day with a standard deviation of US$2,560 | Chague, De-Losso & Giovannetti (2019), Brazil, cohorts of 2013–2015. Taiwan, 1992–2006: fewer than 1% of day traders earn positive returns net of fees predictably (Barber, Lee, Liu & Odean, 2014) |
| 8. A position too large to survive | Gain required to recover a loss | −20% needs +25%; −50% needs +100%; −60% needs +150% | Arithmetic, not a study. The full recovery table is in our guide to the risk hierarchy |
Notice what the two tables say together. The mistakes people fear — picking the wrong company, being a bad trader — are in the second table, where the average cost is often small and the tail is where the damage lives. The mistakes people do not think of as mistakes at all — a fee, a habit of checking prices, a decision to wait for things to calm down — are in the first, where the cost is modest every single year and never stops.
The fee is the mistake with the most certain price
A fee is the only entry in either table that is paid in every market, every year, with no dispersion. That certainty is what makes it expensive. Compound a 7% gross return over thirty years on a $10,000 start, and the difference between an index fund at 0.03% and a fund at the equal-weighted active average of 1.00% is $75,485 against $57,435 — the dearer fund delivers 23.9% less. At 0.60% the shortfall is 14.8%; at a 2.00% all-in cost, which is what an active fund plus an advisory wrap can come to, it is 42.7%.
| Annual cost | After 10 years | After 20 years | After 30 years | Shortfall vs 0.03% at 30 years |
|---|---|---|---|---|
| 0.03% | $19,616 | $38,480 | $75,485 | — |
| 0.60% | $18,596 | $34,581 | $64,306 | 14.8% |
| 1.00% | $17,908 | $32,071 | $57,435 | 23.9% |
| 2.00% | $16,289 | $26,533 | $43,219 | 42.7% |
$10,000 invested, 7% gross annual return, cost deducted annually. Own arithmetic; the return is an assumption, not a forecast.
Morningstar’s fee study makes a further point worth stating precisely. The asset-weighted average expense ratio across all US funds was 0.32% in 2025, which means most money is already in cheap funds. The 1.00% figure is the equal-weighted average of active US equity funds, which is what the average fund charges rather than what the average dollar pays. The mistake, in other words, is not universal — it is concentrated in the investors still paying the list price, and for them it is the most reliable drag in this article.
Missing the best days is the wrong way to describe the timing mistake
The version of the market-timing mistake that circulates most is a chart: stay invested and $10,000 grows to some figure, miss the ten best days and it grows to less than half. J.P. Morgan’s Guide to Retirement runs it on the S&P 500 total return index from 2 January 2006 to 31 December 2025: $80,619 fully invested, $35,866 without the ten best days, an annual return of 11.0% against 6.6%. The calculation is correct. The lesson usually attached to it is not, and an earlier version of this page repeated it without a figure or a period.
We ran the experiment on daily closes of the S&P 500 price index, which excludes dividends and therefore sits below J.P. Morgan’s levels. Over the same twenty years, staying invested returned 8.80% a year and missing the ten best days returned 4.48%: $53,953 against $24,023, a final pot 55% smaller. Over 1950–2025 the same ten-day exclusion takes the return from 8.24% to 7.06% — a cost of 1.2 percentage points a year, not half of anything. The shorter the window, the more dramatic the number, because a window containing 2008 and 2020 contains days the rest of the record does not.
Then run it the other way. Miss the ten worst days of 2006–2025 and the return rises from 8.80% to 13.62%. Miss the ten best and the ten worst, and it is 9.12% — ahead of staying invested. The chart proves that a handful of days decide everything, in both directions; it does not prove that selling is always wrong.
What the chart does establish, and what survives, is where the best days live. All ten of the best days in 2006–2025 arrived with the index at least 10% below its previous closing high, and six of the ten fell within ten trading sessions of one of the ten worst days. The same holds for 1950–2025. The best days are inside the crashes, usually days after the worst ones — so the moment that most makes an investor want to sell is the moment most likely to be followed by the days they cannot afford to miss.
That converts into a measurable mistake once you describe the behaviour honestly: not “missing ten days” but “selling after a bad day and waiting for calm”. Take a rule that sells the S&P 500 after any day of −4% or worse and buys back after a fixed wait. Over 2006–2025, coming back after 20 sessions returns 8.97% a year, marginally ahead of the 8.80% of staying put; after 60 sessions, 7.13%; after 120 sessions, 8.02%. Over 1950–2025 the figures are 8.21%, 6.66% and 6.43% against 8.24%. The cost of the panic is real and it is on the order of one to two points a year, and it comes from being out for months, not from missing ten days.
The practical consequence: the timing mistake is not a failure to predict; it is a failure to have decided in advance what you would do on the worst day. A written rule that fixes the answer — hold, or rebalance on a schedule, or a threshold — costs nothing and removes the decision from the day on which it is most likely to be made badly. Our guide to how often to rebalance covers what the rule should say.
Concentration and day trading: small average cost, unbounded tail
The average stock does fine; the median stock does not. Bessembinder’s finding is that over 1926–2016, only 42.6% of the roughly 25,300 companies that appeared in the CRSP database produced a lifetime buy-and-hold return higher than one-month Treasury bills, and that the best-performing 4% of them account for the entire net wealth creation of the US market — the rest, collectively, matched Treasury bills. The index return is an average dominated by a few extreme winners. An investor who holds one or two names has a better-than-even chance of holding one of the ones that did not beat cash, and no particular chance of holding one of the 4%. That is not a cost in points a year; it is a coin toss with a bad tail, which is why it belongs in the second table.
Day trading has the same shape with worse odds. The Brazilian study followed every individual who began day trading equity futures in 2013–2015 and persisted for at least 300 days: 97% lost money, 0.4% earned more than a bank teller, and the single best performer made US$310 a day with a standard deviation of US$2,560. The Taiwanese records, covering 1992–2006, find that fewer than 1% of day traders earn positive returns net of fees in a way that persists from one year to the next. There are always profitable day traders this year; the question the studies answer is how many of them are profitable next year as well.
Position size is the mistake that turns every other tail into a terminal one, and its arithmetic is the least negotiable thing in this article: a 50% loss needs a 100% gain to get back to even, and a 60% loss needs 150%. Every mistake above is survivable at 3% of a portfolio and none of them is at 40%. That is the whole case for sizing positions before choosing them, and it is why nobody who has done the maths runs full Kelly.
What is not on the list, and why
The mistakes that head most lists — insufficient research, no exit plan, impatience, ignoring tax — are absent here because nobody has put a number on them that survives inspection, not because they are free. “Lack of research” has no measured cost because no study has managed to separate the investors who researched from those who did not; what the brokerage records do show is that conviction, expressed as activity, was expensive. Tax is real and jurisdiction-specific, and any figure would be true for one reader in five. Impatience is a description of the timing gap, already counted. A list ordered by cost has to leave out what has not been costed, and say so, rather than pad the bottom with items that sound wise.
Two mistakes that are genuinely on the list are covered by their own articles, because their costs come from the same arithmetic. Holding cash while waiting for a better entry is the subject of our lump sum versus dollar-cost averaging piece, which puts a number on the wait; holding cash for years because it feels safe is the subject of why savings lose value every year.
The order to fix them in
- Cost first. It is the only mistake with a certain price, it is fixed in an afternoon, and the fix never needs repeating. Find the expense ratio of everything you own; anything near 1% needs a reason.
- Frequency second. Decide how often you are allowed to transact — a rebalancing date, a threshold — and let the rule refuse the rest. The largest measured gap in the first table came from frequency, not from judgement.
- Write down the worst-day rule. What you will do after a −4% day, decided now. The cost of not having it is one to two points a year, incurred on the days you are least able to think.
- Count your positions and size them. Fewer than a handful of names is a bet on the 4%; any single position large enough to need a 100% recovery is a bet on nothing.
- Automate arrival. Fixed contributions on fixed dates remove the 1.2-point gap by construction. If you do not know how much risk you can hold through a fall, the risk profile questionnaire is the place to start, and our guide to investor mindset covers why the rule has to exist before the reason to break it does.
Frequently asked questions
What is the most expensive mistake in stock market investing?
Measured per year, trading frequently: the most active US brokerage households earned 11.4% a year net against 18.5% for the least active in 1991–1997, almost entirely through transaction costs. Measured by certainty, an unnecessary fee: a 1.00% fund instead of a 0.03% index fund delivers 23.9% less after thirty years at 7% gross, in every market. Measured by the size of the hole, position size: a 50% loss needs a 100% gain to repair.
How much do fees really cost over the long run?
On a $10,000 start at 7% gross, thirty years at 0.03% ends at $75,485; at 1.00% it ends at $57,435, a shortfall of 23.9%; at 2.00%, $43,219, or 42.7% less. The 1.00% is Morningstar’s equal-weighted average for active US equity funds in 2025; the asset-weighted average across all US funds was 0.32%, so the mistake is concentrated in the investors still paying the list price.
Is missing the best days really as costly as the charts say?
Over twenty years, yes: missing the ten best days of 2006–2025 cut the S&P 500 price return from 8.80% to 4.48% a year, roughly halving the final pot. Over 1950–2025 the same exclusion costs 1.2 percentage points a year, and missing the ten worst days would have helped more than missing the ten best hurt. What holds in every window is that the best days cluster inside crashes, within days of the worst ones, so the real cost belongs to selling after a fall and staying out for months: one to two points a year.
Is buying individual stocks a mistake?
It is a bet with a specific shape. Over 1926–2016 only 42.6% of US common stocks beat one-month Treasury bills over their lifetime, and 4% of companies account for all of the market’s net wealth creation. A diversified index owns the 4% by construction; a portfolio of two or three names has a better-than-even chance of owning none of them, which is why concentration is a question of position size rather than of stock selection.
Do 90% of day traders lose money?
The 90% has no identifiable source. The measured figures are narrower and worse: in Brazil, 97% of those who began day trading equity futures in 2013–2015 and kept going for at least 300 days lost money, and 0.4% out-earned a bank teller; in Taiwan over 1992–2006, fewer than 1% of day traders earned positive returns net of fees in a way that persisted from year to year.
Why can’t you add these costs together to get the total cost of bad investing?
Because they were measured on different people, in different decades, against different benchmarks, and several describe the same behaviour from different angles. Summing the first table’s gaps would take a long-run 8% return negative, which nobody has observed. What you can do is let one measured gap compound: 1.2 points a year, over twenty years at 8%, is about 20% of the final sum.
Sources
- Brad M. Barber and Terrance Odean, “Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors”, The Journal of Finance 55(2), April 2000. 66,465 households at a large US discount broker, February 1991 to January 1997; net annual return 11.4% for the most active quintile against 18.5% for the least active.
- Zachary Evens and Brendan McCann, 2026 US Fund Fee Study, Morningstar Manager Research, May 2026. Asset-weighted average expense ratio across US funds 0.32% in 2025; active US equity funds 0.58% asset-weighted and 1.00% equal-weighted. The 0.03% comparison figure is what the cheapest of the large S&P 500 index ETFs charge.
- Morningstar, Mind the Gap 2026. Roughly 23,000 US open-end funds and ETFs; for the decade to 31 December 2025 the average dollar earned 8.7% a year against 9.9% for the funds themselves.
- J.P. Morgan Asset Management, Guide to Retirement 2026, “Impact of being out of the market”: $10,000 in the S&P 500 Total Return Index, 2 January 2006 to 31 December 2025, $80,619 fully invested against $35,866 missing the ten best days; 11.0% against 6.6% a year.
- Hendrik Bessembinder, “Do Stocks Outperform Treasury Bills?”, Journal of Financial Economics 129(3), 2018. Approximately 25,300 companies in the CRSP database, 1926–2016; 42.6% with lifetime buy-and-hold returns above one-month Treasury bills; the best-performing 4% of companies explain the net gain for the entire US market.
- Fernando Chague, Rodrigo De-Losso and Bruno Giovannetti, “Day Trading for a Living?”, Department of Economics FEA-USP, Working Paper 2019-47 (revised June 2020). All individuals who began day trading Brazilian equity futures in 2013–2015 and persisted at least 300 days: 97% lost money, 0.4% earned more than a bank teller (US$54 a day), top individual US$310 a day with a standard deviation of US$2,560.
- Brad M. Barber, Yi-Tsung Lee, Yu-Jane Liu and Terrance Odean, “The Cross-Section of Speculator Skill: Evidence from Day Trading”, Journal of Financial Markets 18, 2014, pp. 1–24. Taiwan Stock Exchange, 1992–2006; fewer than 1% of day traders earn positive abnormal returns net of fees predictably.
- Own calculations on daily closing prices of the S&P 500 price index (Yahoo Finance, ^GSPC), 3 January 1950 to 31 December 2025 and 3 January 2006 to 31 December 2025, dividends excluded. Fully invested, ten best days removed, ten worst removed, both removed; drawdown from the prior closing high on each best day; the −4% sell rule with 20, 60 and 120-session waits. Fee and recovery tables: compound arithmetic at the stated assumptions.
Correction, 19 September 2026: an earlier version of this page said that missing a handful of the best days “removes a large share of the total return”, with no period and no figure attached. The figures, windows and method now stated above replace that sentence, as our corrections policy requires.
This article is educational and is not investment advice. We are not licensed advisers, and nothing here is a recommendation to buy, sell or hold any security. Figures are taken from the studies named above, on the samples and periods stated, or are our own arithmetic at assumptions we have made explicit; they are averages or illustrations, not forecasts of any individual result.
