Key takeaways
- Selling one £5,000 UK holding and buying another cost £36.50 on rate cards read on 19 August 2026, or 0.73% of the position. Stamp duty is £25 of that.
- Turn a £50,000 portfolio over once a year and the dealing bill is £365. Compounded for 20 years inside an ISA it costs £19,260, or 13.0% of the ending pot.
- Commission is the layer that collapsed. On a broker charging no commission the same round trip still costs £26.50, and 94% of that is the 0.5% duty.
- Tax drag runs backwards below 74% turnover: 20 annual exempt amounts of £3,000 beat one, so moderate turnover ended £7,756 ahead of buy and hold.
- Barber and Odean's most active households turned over more than 250% a year and netted 11.4% against 18.5% for the least active, from 1991 to 1996.
The cost of frequent trading is 0.73% of every pound you move
You want to know what trading often actually costs, in pounds, before anything is said about whether the trades were any good. Here's the arithmetic from UK rate cards read on 19 August 2026. Sell a £5,000 holding in a UK share and buy a different one, and the round trip costs £36.50. That's 0.73% of the position.
It breaks into two dealing commissions of £5 each on AJ Bell's schedule, which prices an online share deal the same in its ISA and its dealing account, plus £25 of stamp duty on the purchase and £1.50 of bid-offer spread. Turnover simply multiplies it. Replace a quarter of a £50,000 portfolio in a year and you've spent £91.25. Replace the whole thing and you've spent £365, which is 0.73% of the portfolio.
That relationship is linear and dull, and the dullness is the point. The cost of frequent trading is knowable in advance to the penny, while the returns it chases are not knowable at all. The tax that sits on top behaves differently, and not in the direction most people assume.
Stamp duty is 68% of a UK share round trip
Here is the same £5,000 round trip, itemised.
| Component | Cost | Share of the round trip |
|---|---|---|
| Dealing commission, one sale and one purchase | £10.00 | 27% |
| Stamp Duty Reserve Tax on the purchase | £25.00 | 68% |
| Bid-offer spread | £1.50 | 4% |
| PTM levy, not payable below £10,000 | £0.00 | 0% |
| Total | £36.50 | 0.73% of the position |
The duty is the largest line and the one nobody shops around for. GOV.UK puts it plainly: "When you buy shares, you usually pay a tax or duty of 0.5% on the transaction." Buy electronically and it's Stamp Duty Reserve Tax. It lands on the purchase only, so a round trip pays it once, and a portfolio that never buys anything never pays it at all.
The spread figure comes from Vanguard, which publishes the bid-offer spread on its exchange traded funds as a one-off cost, estimated on a rolling 30-day average. Its FTSE 100 fund showed 0.03% as at 2 July 2026. On £5,000 that's £1.50.
The PTM levy is the one line that switches on with size rather than frequency. The Takeover Panel sets it at "150p per contract where the total consideration of the relevant trade is greater than £10,000". A £5,000 position stays under it. Trade in £20,000 blocks instead and you add £3.00 per round trip, which is real money but the smallest thing on this page.
Commission is the only layer that has collapsed
Dealing charges have fallen hard, and the schedules now reward frequency explicitly. Hargreaves Lansdown charges £6.95 a share trade at "0 - 19 trades last month" and £3.95 at "20+ trades last month". AJ Bell charges £5 per online share deal, dropping to £3.50 "where there were 10 or more shares deals in the previous month". Interactive investor charges £3.99 a UK share trade on a Core plan costing £5.99 a month. Freetrade advertises "no commission charges per trade and no subscription fee on the Basic plan".
So the explicit price of trading gets cheaper the more you do it. That's a genuine change, and it's the strongest reason to distrust older writing on this subject. It also runs out quickly. Set commission to zero and the £5,000 round trip still costs £26.50, because the duty and the spread don't move. At that point 94% of what you pay is the 0.5%, which no broker in Britain can discount.
The contrast with funds is sharper still. GOV.UK lists the cases where no duty is charged, including when you "buy units in a unit trust from the fund manager". AJ Bell charges £1.50 per online fund deal, and Vanguard states that its mutual funds "do not incur one-off costs" in the form of a spread. A £5,000 switch between two funds on that route costs £3.00, against £36.50 in shares. That's the same decision, priced twelve times apart, on the same platform.
Compounded for 20 years, 100% turnover costs £19,260 inside an ISA
Start with the simpler container. An ISA pays no capital gains tax at all, since GOV.UK's rule is that "You can save tax-free with Individual Savings Accounts", with a £20,000 subscription limit for the 2026 to 2027 tax year. So inside one, the only cost of turnover is the dealing bill, compounding against you every year it's paid.
The growth rate used below is 5.60% a year, the annualised return of the iShares Core FTSE 100 exchange traded fund from its launch on 27 April 2000 to 31 July 2026, on a net asset value basis with income reinvested. It's one fund over one stretch of one market, and it's history rather than a forecast. It's here to make the arithmetic concrete, not to predict anything.
| Annual turnover | After 10 years | After 20 years | After 30 years |
|---|---|---|---|
| None, buy and hold | £86,220 | £148,679 | £256,382 |
| 25% | £84,742 | £143,623 | £243,417 |
| 50% | £83,286 | £138,731 | £231,087 |
| 100% | £80,442 | £129,418 | £208,213 |
| 250% | £72,426 | £104,912 | £151,968 |
Replacing the portfolio once a year costs £19,260 over 20 years, which is 13.0% of what the untouched pot reached. Over 30 years it's £48,169, or 18.8%. At the 250% turnover of the busiest households in the research below, 20 years costs £43,767, or 29.4%.
Notice how little separates buy and hold from 25% turnover: £5,055 over two decades, or 3.4%. Replacing a quarter of a portfolio a year is roughly what a disciplined rebalancing policy produces, and it barely registers. If you want the evidence on how often that's worth doing, our piece on rebalancing frequency works through Vanguard's own comparison. The damage in this table starts once turnover passes the portfolio's whole value each year.
Tax drag runs backwards until you turn over about 74% a year
Outside a tax wrapper the arithmetic changes shape, and it surprises people. Capital gains tax for the 2026 to 2027 year runs at 18% within the basic rate band and 24% above it, and GOV.UK sets the tax-free allowance at £3,000. The intuition says that realising gains early hands money to HMRC that could have compounded, so turnover must be worse in a dealing account than in an ISA. Half of that is right.
The chart with this piece runs the comparison over 20 years on a £50,000 dealing account at the 24% rate, taxing gains as they're realised and taxing whatever is left unrealised at the end. Buy and hold reaches £148,679, pays £22,963 of tax on the way out, and nets £125,716. Turnover of 25% a year nets £133,472, which is £7,756 more. Turnover of 50% nets £129,562. Only at 100% does the trader fall behind, at £121,581, and at 250% the gap opens to £25,952, or 20.6%.
The crossover sits at about 74% annual turnover. The mechanism is the allowance. A buy-and-hold investor who sells once uses one £3,000 exemption in 20 years. Someone realising gains every year uses 20 of them, and at moderate turnover the annual gain is small enough that a large share of it escapes tax entirely. Deferral is worth something and so is the allowance, and below 74% the allowance wins.
One rule stops this being free money. HMRC's "bed and breakfast" rule, at TCGA92/S106A, identifies a disposal against shares of the same class "acquired within the 30 days after the disposal". Sell and buy the same holding back on Monday and the base cost doesn't reset. Using the allowance means genuinely changing what you own, or staying out for 30 days, and both carry the round trip costed above. Our guide to the bed and ISA arithmetic follows the same trade-off into a wrapper.
The counter-case: the famous evidence is thirty years old
The study everyone reaches for is Barber and Odean's, published in 2000. They looked at 66,465 households at a large discount broker from 1991 to 1996. The households that traded most earned a net annualised 11.4%, against 18.5% for those that traded least, and against a market return of 17.9%. The average household turned over 75% of its portfolio a year; the busiest turned over more than 250%.
Their conclusion was blunt: "trading is hazardous to your wealth." The critics have a real point about applying it here, though. That sample is American, it predates the collapse in commissions, and it covers a single six-year bull market. A UK investor in 2026 pays a fraction of the commission those households paid.
What survives the objection is the part that was never about commission. Gross returns barely differed across turnover levels in that data, with the paper reporting "very little difference in the gross performance of households that trade frequently (with monthly turnover in excess of 8.8 percent) and those that trade infrequently". The active households weren't picking worse shares. They were paying more to hold the same ones. That mechanism is exactly what the table above prices, and in Britain the cost of frequent trading is dominated by a statutory 0.5% that has nothing to do with skill. The behavioural side of why turnover climbs is covered in our piece on overconfidence in investing.
Where this model breaks
Several things in it are softer than they look. The spread of 0.03% is Vanguard's published rolling average for a FTSE 100 tracker, which is about as tight as UK equity dealing gets. A mid-cap or small-cap share trades wider, and no UK broker publishes a median spread for individual shares, so this model understates the spread for most portfolios rather than overstating it.
There's no market impact in it either, which is fair for a £5,000 retail order and wrong for anything institutional. The 5.60% growth rate is one fund's realised history from 2000 to 2026 and carries no forecasting weight; a lower rate shrinks every pound figure here while leaving the percentages close to unchanged. Platform account fees sit outside the model entirely, because they're charged on the balance rather than on turnover, and our breakdown of the cost of investing in the UK ranks those separately.
The tax half is more fragile than the dealing half. It assumes gains rather than losses, a constant 24% rate, and a £3,000 allowance that stays put. That allowance is the whole reason the crossover sits at 74%, and it's a policy setting rather than a law of nature. The model also treats portfolio turnover as a choice, when takeovers, index changes and fund closures all force it.
What would change the conclusion
Three things would move it, and they're worth watching in that order. Abolishing or cutting stamp duty on shares takes 68% out of the round trip, and at £11.50 a turn the whole question shrinks to a rounding error below 100% turnover. Cutting the £3,000 annual exempt amount pushes the 74% crossover toward zero and turns any realisation in a dealing account into pure drag. A materially lower return assumption makes deferral worth less and pushes the crossover up.
The figure none of this tells you is your own. Most platforms report your balance and your charges, and none of them report the percentage of your portfolio you replaced last year, which is the number that drives every line above. LedgerTouch computes turnover from your transaction history, which at least puts the multiplier on the screen next to the costs it multiplies.