What a position size decides before any view is tested
Position size decides how much a wrong view costs, how many wrong views in a row an account can absorb, and whether a single bad morning is a setback or the end of the account. Two traders can hold the same option on the same day and have opposite experiences, because one holds a contract and the other holds ten. This page sets out the sizing method the library's worked examples use, then tests it: on a run of losses, on a £50,000 book whose every price falls at once, on an uncovered put hit by a gap, and on a position priced in dollars.
Everything here is measured in pounds per contract, so the contract conventions come first. An ICE standard UK stock option covers 1,000 shares and is quoted in pence, so each 1p of premium is £10. An ICE mini covers 100 shares, so 1p is £1. A FTSE 100 index option (ESX) is £10 a point. A US equity option covers 100 shares and is quoted in dollars, so each $1 of premium is $100, converted to pounds at a stated rate. The basics page has the full table of contract sizes. Reading a per-share figure as the per-contract figure understates the commitment by the multiplier, so every figure below is the per-contract amount.
For a vertical spread the most that can be lost is (width − credit) × multiplier, plus commission. The HSBC October 1,450/1,400 bull put spread collects 12.00p on a 50p width: (50 − 12) × £10 = £380.00 a contract, £382.80 with the opening commission. Level 2's defined-risk table gives the same formula for condors, butterflies and calendars.
The sizing framework used on this site
The library's worked examples measure every position in pounds before it is opened and compare the figure with a set of lines drawn as a share of the account. The lines are the library's own round numbers. They are not a regulatory or industry standard, and no study derives them; the methods page records where each teaching convention comes from and what evidence exists for it. What this page adds is the arithmetic: what each line allows in pounds, what it rules out, and where it fails.
| What is measured | How it is worked out | Line in the worked examples | On a £50,000 account |
|---|---|---|---|
| Speculative long premium (far out-of-the-money options, positions held over one event) | Maximum loss: premium paid plus commission | 1% of the account | £500 |
| Directional long options and defined-risk spreads | Maximum loss: debit plus commission, or (width − credit) × multiplier plus commission | 2% of the account | £1,000 |
| Cash-secured put | Collateral: strike × contract size, with the loss if the shares went to zero shown beside it | 5% of the account in collateral | £2,500 |
| Covered call | No new capital. The shares carry their own loss; the call's delta is counted with them | Measured with the share holding | Exposure line below |
| Undefined-risk position (uncovered put or call on margin, short straddle or strangle, ratio spread) | Two figures: the broker's initial margin, and the loss in an instant 20% move against the position | Margin no more than 5% of the account; gap loss no more than 10% | £2,500 and £5,000 |
| Capital committed across the book | Premiums paid, spread risk, collateral and margin, added up | 25% of the account | £12,500 |
| Exposure to one underlying | Delta-£ (share-equivalents × share price) across every position in that name, shares included | 10% of the account | £5,000 |
| Exposure to one group of related names (banks, oil, miners) | Delta-£ added up across the group | 15% of the account | £7,500 |
The first three lines limit what one position can lose, and the capital line limits how much of the account is tied up at once. The last two limit how much of the account depends on one price or one kind of news, which a loss line cannot see: a call that can lose only its premium can still move like several thousand pounds of shares. The undefined-risk line uses two figures because neither is enough alone. Margin is what the broker holds today and can raise tomorrow; the gap loss is what the account pays if the price jumps before anything can be done. The position-size tool on the planner turns a loss line into a number of contracts, with commission, a collateral cap and a stress figure for positions with no maximum loss.
The Tesco test: what a 5% collateral line allows
A cash-secured put ties up the strike's full value, which is large on a 1,000-share contract. The October 430 put on Tesco, written at 8.25p on 17 August 2026 (model value 8.30p), needs £4,300 of collateral on one standard contract: 8.6% of a £50,000 account, beyond the 5% line. Under that line one standard contract needs an account of £86,000. Tesco is one of the 22 UK names that ICE also lists as a 100-share mini (code 8TC), where the same put ties up £430 and fits the line from an account of £8,600. On £50,000 the line allows five minis, £2,150.00 of collateral. The mini is listed on ICE; whether a given broker offers it and quotes a two-way price has to be checked on the broker's own platform. The fixed commission weighs more on a small premium: an assumed £1.70 commission per mini contract is 20.6% of an 8.25p premium on 100 shares.
None of the lines is advice about how much of an account belongs in options at all, and none makes a trade safe.
Capital at risk and market exposure are different numbers
Model inputs. Monday 17 August 2026. BP at 530p, a model level (BP closed at 519.6p that day; price data: Yahoo Finance); IV 26%; Bank Rate 3.75%; BP's third-quarter dividend assumed at 6.39p going ex on Thursday 12 November 2026 (the date was in BP's 2026 financial calendar on the model date; the amount assumes the second-quarter 8.66 US cents at the model rate of $1.356 per £1). The December 500 call has 123 days to Friday 18 December and is American, so it is priced on the binomial tree: model value 49.35p, filled at 49.25p. Commission an assumed £1.40 a contract. BP is used as a model underlying; this is not a view on BP.
The call costs 49.25p × 1,000 shares = £492.50, and £493.90 with commission: that is the most it can lose, just inside the 1% line of £500. It does not behave like £493.90 of anything. Its delta is 0.70, so the contract moves like 702 BP shares: about £7.02 for each 1p move in BP. Those 702 shares are worth £3,722.92, which is the call's delta-£, or market exposure: 7.5 times its maximum loss and 7.4% of the account. For a small move, the call and £3,722.92 of BP shares gain or lose about the same amount; for a large fall the call's loss stops at the premium and the shares' does not. The Greeks page sets out delta and the other position Greeks in these units.
The covered call is the mirror image. Writing the BP October 560 call against 1,000 shares already held brings in £120.00 (12.00p; model value 11.97p) and commits no new capital, yet the holding can lose £5,181.40 if BP falls to zero: (530 − 12) × £10 plus commission. The call trims the exposure only a little, to 660 share-equivalents or £3,495.41, because a call 30p out of the money has a delta of 0.34. A cash-secured put sits between the two: the five Tesco minis tie up £2,150.00 of collateral but carry £699.92 of exposure at the start (156 share-equivalents), and that exposure grows as Tesco falls, approaching the full 500 shares once the put is well in the money.
Both numbers matter, for different reasons. The loss lines stop any one position from doing serious damage by itself. The exposure lines catch the case the loss lines miss: several positions, each small by its maximum loss, that all lose on the same move.
One book, three numbers: a £50,000 account on 17 August 2026
The teaching book holds four positions, each taken from a library page and priced on the same standard assumptions, so the figures here match those pages. It is a device for showing the measures at work, not a portfolio anyone is being shown to hold.
Model inputs. As above for BP. HSBC at 1,530p, a model level (HSBC closed at 1,531.0p on 17 August 2026; price data: Yahoo Finance), IV 25%, no ex-dividend date before 16 October; the October 1,450 put written at 26.00p (model 26.04p) and the 1,400 put bought at 14.00p (model 13.99p), two contracts. Tesco at 450p, a model level (Tesco closed at 447.8p), IV 22%, an assumed 5.08p interim dividend going ex on Thursday 15 October 2026 (the interim had not been declared on the model date; Tesco declared 5.05p on 8 October 2026, ex 15 October, and the example is not re-priced); five October 430 mini puts written at 8.25p. All October options have 60 days to Friday 16 October 2026 and are American, priced on the binomial tree (the BP October call by Black-Scholes-Merton, as no ex-date falls in its life). Commission £1.40 a standard contract and £1.70 a mini. Delta-£ values each position's share-equivalents at the model share price. Betas are each share's sensitivity to the FTSE 100, estimated from the 250 trading days of daily returns to 17 August 2026 (price data: Yahoo Finance, checked 28 September 2026).
| Position | Capital committed | Maximum loss | Delta (share-equivalents) | Delta-£ (exposure) | Vega (£ per vol point) | Beta to the FTSE 100 | Beta-weighted £ |
|---|---|---|---|---|---|---|---|
| A. 1,000 BP shares held, October 560 call written | None new (shares worth £5,300) | £5,181.40 | +660 | £3,495.41 | −£7.88 | −0.13 | −£454.40 |
| B. BP December 500 call, 1 contract | £493.90 | £493.90 | +702 | £3,722.92 | £10.33 | −0.13 | −£483.98 |
| C. HSBC October 1,450/ | £765.60 | £765.60 | +202 | £3,086.54 | −£10.10 | 1.58 | £4,876.73 |
| D. Tesco October 430 put, 5 minis, cash-secured | £2,150.00 collateral | £2,117.25 | +156 | £699.92 | −£3.18 | 0.59 | £412.95 |
| Options positions B to D | £3,409.50 (6.8%) | £3,376.75 (6.8%) | |||||
| Whole book | £8,558.15 (17.1%) | £11,004.79 (22.0%) | −£10.82 | £4,351.30 |
Maximum loss. The options commit £3,409.50, 6.8% of the account, well inside the 25% line, and could lose £3,376.75 if all three reached their maximum loss together. That total, sometimes called the book's heat, is a blunt figure: it assumes the worst for every position at once. Adding the BP shares, which can lose £5,181.40 on their own, takes it to £8,558.15, 17.1% of the account.
Exposure. Every position passes its own line, but the two BP positions together carry £7,218.33 of BP exposure, 14.4% of the account, above the 10% line for one underlying. The worked plan would hold one of them, or fewer shares, rather than both. This is the check the loss lines cannot make: the call is a 1% position by what it can lose and a 7.4% position by exposure, and it adds to the shares rather than offsetting them. The whole book behaves like £11,004.79 of shares, 22.0% of the account, for small moves.
Volatility. The book's vega is −£10.82 per volatility point: it loses about that much for each point by which implied volatility rises in all three names at once, and its time decay earns £2.55 a day. The two figures are linked; a book that earns from time passing usually loses when implied volatility rises.
Beta-weighted. Weighting each position's delta-£ by its beta turns the book into one FTSE 100 figure: £4,351.30, or £0.40 a point of the index, a small fraction of one ESX contract's £107,500 of index exposure. That figure is much smaller than the £11,004.79 of share exposure, and the reason is BP: over the year to 17 August its beta came out at −0.13, so the beta view counts both BP positions as a slight hedge. A one-year beta is a noisy estimate. It says how the book tends to move with the index on an ordinary day. It says nothing about BP on its own results day, 30 October, which is why the per-underlying line is measured in delta-£, not in beta-weighted pounds.
Five losses in a row are more likely than 1%
If 60% of trades win and each trade is independent, any particular five trades all lose 0.4 × 0.4 × 0.4 × 0.4 × 0.4 = 1.02% of the time, about 1 in 98. That figure answers a narrow question: whether the next five trades will all lose. The useful question is whether a run of five losses turns up somewhere in a trading record, and the answer is far higher, because a record of 100 trades contains 96 overlapping stretches of five. Over 10 trades the chance of at least one such run is 4.1%; over 20 trades it is 10.0%, over 50 25.6%, over 100 45.9% and over 200 71.4%. The 1% figure understates the chance over 100 trades by a factor of about 45.
| Share of trades that win | 20 trades | 50 trades | 100 trades | 200 trades |
|---|---|---|---|---|
| 50% | 25.0% | 55.2% | 81.0% | 96.6% |
| 60% | 10.0% | 25.6% | 45.9% | 71.4% |
| 70% | 2.8% | 7.7% | 15.3% | 28.6% |
| 80% | 0.4% | 1.2% | 2.4% | 4.9% |
A high win rate shortens the runs but does not remove them. Over 100 trades the chance of meeting a run of a given length is:
| Run of losses | 60% of trades win | 70% win | 80% win |
|---|---|---|---|
| 3 or more | 98.8% | 86.4% | 47.4% |
| 4 or more | 80.1% | 43.3% | 11.8% |
| 5 or more | 45.9% | 15.3% | 2.4% |
| 6 or more | 21.2% | 4.8% | 0.5% |
| 7 or more | 8.9% | 1.4% | 0.1% |
| 8 or more | 3.6% | 0.4% | <0.1% |
What five losses cost at 1%, 2% and 5%
How much a run costs depends on the stake. With the stake set as a fixed share of the account as it stands (fixed-fractional sizing), each loss is taken from a smaller balance, so five full losses at 5% leave 0.95 × 0.95 × 0.95 × 0.95 × 0.95 of the account: a fall of 22.6%, £11,310.95 on £50,000. Five losses of a fixed 5% of the starting balance would cost 25%. At 2% the five losses cost 9.6%, and at 1% 4.9%. Getting back to the starting balance needs a gain of 29.2% after the 5% run and 10.6% after the 2% run, because the gain is earned on the smaller balance. A full loss on a 5% cash-secured put would mean the shares had gone to zero, so the 5% row describes long premium or spreads sized at 5%, not the collateral line.
Every figure in this section assumes independent trades. Positions on the same underlying, or on related shares in one sell-off, are not independent: their losses tend to arrive together, which makes long runs more likely than the tables show. The correlation section below measures how closely UK banks, oil companies and miners moved together over the year to 17 August 2026.
Win rate, payoff and the chance of a 20% drawdown
A win rate means nothing without the payoff ratio, the average win divided by the average loss. Four combinations show the range. The Kelly fraction, the last column, is explained in the next section.
| Win rate and payoff | Expected result per £100 risked | Win rate needed to break even | Kelly fraction |
|---|---|---|---|
| 70% of trades win; wins equal losses (1:1) | +£40.00 | 50.0% | 0.40 |
| 70% win; losses four times the wins (payoff 0.25) | −£12.50 | 80.0% | −0.50 |
| 50% win; wins twice the losses (payoff 2) | +£50.00 | 33.3% | 0.25 |
| 30% win; wins five times the losses (payoff 5) | +£80.00 | 16.7% | 0.16 |
Credit spreads sit near the second row. The HSBC bull put spread keeps £117.20 if it expires worthless and loses up to £382.80, a payoff ratio of 0.31; treated as a simple win-or-lose bet it would need to win 76.6% of the time to break even, while its model probability of finishing above the 1,438p breakeven (risk-neutral, lognormal, IV 25%) is 73.3%. The bull put spread page shows why the true expected result is close to zero rather than negative: many losing outcomes land between the strikes and lose less than the maximum. A simple win-or-lose table is therefore harsher on spreads than their real payoffs, but it shows correctly how little room a small payoff ratio leaves.
How often the account falls 20% below its peak
The table below gives the chance that an account falls 20% below its highest value at some point within 100 trades, risking 2% of the current account on each trade. Every trade either wins the stated payoff or loses the whole 2%, independently of the others. Beside each chance, in brackets, is the expected result per £1 risked.
| Share of trades that win | Payoff 0.3 | Payoff 0.5 | Payoff 1 | Payoff 2 | Payoff 3 |
|---|---|---|---|---|---|
| 50% | >99.9% (−0.35) | 99.1% (−0.25) | 50.0% (0.00) | 6.6% (+0.50) | 2.6% (+1.00) |
| 60% | 98.4% (−0.22) | 68.6% (−0.10) | 5.9% (+0.20) | 0.5% (+0.80) | 0.2% (+1.40) |
| 70% | 55.3% (−0.09) | 8.1% (+0.05) | 0.1% (+0.40) | <0.1% (+1.10) | <0.1% (+1.80) |
| 80% | 1.9% (+0.04) | <0.1% (+0.20) | <0.1% (+0.60) | <0.1% (+1.40) | <0.1% (+2.20) |
The left-hand column is where a high win rate looks reassuring and is not. At 70% wins and a payoff of 0.3, the expected result is negative and the account has a 55.3% chance of falling 20% within 100 trades even at a 2% stake. Moving the payoff from 0.3 to 0.5 at the same win rate cuts that to 8.1%. The size of the average loss, not the win rate, decides which cell a strategy is in.
The stake matters as much as the strategy. The same three profiles at four stakes:
| Risk per trade | 70% win, payoff 0.5 (expected +0.05 per £1) | 60% win, payoff 1 (expected +0.20) | 40% win, payoff 2 (expected +0.20) |
|---|---|---|---|
| 1% of the account | <0.1% / <0.1% | <0.1% / <0.1% | 2.3% / <0.1% |
| 2% | 8.1% / <0.1% | 5.9% / <0.1% | 41.3% / 0.1% |
| 5% | 77.8% / 3.3% | 77.3% / 2.1% | 99.3% / 25.6% |
| 10% | 99.7% / 46.8% | 99.9% / 43.2% | >99.9% / 90.8% |
Each cell shows the chance of a 20% fall, then of a 50% fall. Two profiles with the same expected result behave very differently: the 60%-win, even-payoff profile has a 5.9% chance of a 20% fall at 2% stakes, while the 40%-win profile that pays twice its losses has 41.3%. Long premium, which wins less often and more, has its own drawdown problem: long losing runs. For the first two profiles, moving from 2% to 5% a trade turns a 20% fall from uncommon into likely; for the third it becomes close to certain.
Kelly: the largest stake the arithmetic supports
The Kelly fraction is the share of the account that makes it grow fastest over many repeated bets, when the win rate p and the payoff b are known exactly: f* = p − (1 − p) ÷ b. For 70% wins at even payoff it is 0.7 − 0.3 ÷ 1 = 0.40. For 70% wins with losses four times the wins, it is 0.7 − 0.3 ÷ 0.25 = −0.50. A negative Kelly fraction means no positive stake makes the account grow: at those odds the bet loses money on average, whatever the size.
The full Kelly stake is also a harsh one. For the 70%, even-payoff bet, the long-run growth rate per trade (logarithmic) is 8.23% at the full Kelly stake of 0.40, 6.07% at half of it, and −7.14% at twice it: staking double the Kelly fraction shrinks the account over time even though every bet has a positive expected result. Half the stake keeps about three-quarters of the growth. And the formula assumes p and b are known. In options they are estimates, and an overestimate of the win rate pushes the stake above the true Kelly fraction, where growth falls away quickly.
Options add a further point. The methods page shows that, priced on the model and judged on the model's own probabilities, every option in the library is a fair bet before costs: the premium equals the discounted value of what the option is expected to pay. For a fair bet the Kelly fraction is zero, and after commission and the bid-ask spread it is negative; the HSBC spread's expected result is +£0.34 before costs and −£12.46 after them on its own page. A positive Kelly stake needs a reason to think the market's implied volatility is wrong, which is a claim about data, discussed on the implied-volatility page. That is why the framework above is built on what a position can lose, and not on an estimate of what it will make.
The book in four instant moves
A stress test asks what the book is worth straight after a stated move, before anything can be done. Each row below moves all three share prices together and shifts every implied volatility by the stated number of points, with each strike keeping its own volatility otherwise (sticky-strike, explained on the implied-volatility page). The method, and the gap history behind the size of the moves, are on the Level 3 page. Each option is revalued on the same pricing model as the entry prices, American options on the tree.
| Position | All three −20%, IV +10 points | −10%, IV +5 | No move, IV +5 | +10%, IV −3 |
|---|---|---|---|---|
| A. BP shares with the 560 call written | −£948.59 | −£444.30 | −£39.75 | +£280.21 |
| B. BP December 500 call | −£377.77 | −£245.28 | +£53.97 | +£410.86 |
| C. HSBC bull put spread, 2 contracts | −£635.81 | −£379.72 | −£44.47 | +£209.77 |
| D. Tesco 430 put, 5 minis | −£329.57 | −£133.75 | −£16.46 | +£38.58 |
| Options B to D | −£1,343.15 | −£758.75 | −£6.96 | +£659.20 |
| Whole book | −£2,291.74 (−4.6%) | −£1,203.05 (−2.4%) | −£46.71 (−0.1%) | +£939.41 (+1.9%) |
An instant 20% fall in all three shares, with implied volatility up 10 points, costs the book £2,291.74, 4.6% of the account. The largest line is the share holding, not an option. The long call loses £377.77, most of its premium, and the HSBC spread £635.81, less than its £765.60 maximum because 60 days still remain. A rise in volatility alone costs the book £46.71, with the long call's gain offsetting most of what the written options lose.
What the stress table leaves out: cash for assignment
The table shows values, not cash. After a 20% fall both HSBC 1,450 puts are deep in the money, and an American put can be assigned early (see early assignment of short puts). If both are assigned before the bought 1,400 puts are exercised, the account must pay for 2,000 HSBC shares: £29,000.00, 58% of the account, plus £145.00 of stamp duty reserve tax, which the put writer pays as the buyer of the shares. Exercising the bought puts sells the shares at 1,400p and caps the loss, but the cash has to be found first; one leg of a spread assigned works through the sequence. The five Tesco minis would take 500 shares for £2,150.00, cash already held as collateral, plus £10.75 of stamp duty reserve tax. A spread's risk is fixed only at expiry; the assignment page explains why. Brokers can also raise margin requirements after a fall, which is the start of the spiral described on the Level 3 page.
A spread can show half its maximum loss with two months to run
Sizing by maximum loss covers the worst case at expiry. Between entry and expiry the position is marked at its current price, and that mark can move a long way before the maximum loss is anywhere near certain. One HSBC 1,450/1,400 spread, per contract, the day after entry (59 days left), revalued on the model:
| What happens | Mark per contract | Share of the maximum loss |
|---|---|---|
| No move; implied volatility up 10 points | −£39.06 | 10% |
| HSBC 5% lower at 1,453.5p; IV up 5 points | −£100.64 | 26% |
| HSBC 10% lower at 1,377p; IV up 5 points | −£190.36 | 50% |
| Still at 1,377p with 30 days left; IV up 5 points | −£210.44 | 55% |
With HSBC 10% lower the day after entry, the spread shows half its maximum loss, 59 days before the maximum can be reached, and the mark moves towards the maximum only as time passes. A worked plan that closes at a set fraction of the maximum loss turns these marks into realised losses; one that holds accepts that the mark can reach the maximum. Either way the loss line was set on the right number, because the maximum is what a spread can cost; the mark only decides what the account shows along the way. Implied volatility alone moves the mark as well: with no change in the price, a 10-point rise costs £39.06 a contract, the part of short-premium risk that a price chart does not show.
Undefined risk: the margin line and the gap test, worked on a BP put
A written option with nothing bought behind it has no maximum loss that a 1% or 2% line could be applied to. The framework measures it twice instead: its worked examples keep the broker's initial margin to no more than 5% of the account's value, and the loss from an instant 20% move against the position to no more than 10%. Both are round numbers, and the 20% move is a stress size, not a forecast or a floor; the Level 3 page's gap history shows the one-day falls behind it.
Model inputs. Monday 17 August 2026: one BP October 500 put written at 9.00p (model value 9.04p; BP 530p, IV 26%, 60 days, no ex-date before expiry, American, binomial tree) in a margin account with permission to write uncovered puts (IBKR lists short puts at its Options Level 3; see account types and permissions). The margin is worked with the Cboe/FINRA strategy-based formula as a published yardstick: 100% of the premium plus 20% of the share price less the amount the put is out of the money, with a minimum of the premium plus 10% of the strike. For ICE contracts the broker sets its own requirement and can raise it; the Level 3 page has the formulas in full.
| Test | Per contract | Contracts that pass | At 2 contracts |
|---|---|---|---|
| Initial margin, no more than £2,500 (5%) | (9.00 + 106.00 − 30.00)p × £10 = £850.00 | 2 (£2,500 ÷ £850 = 2.94) | £1,700.00, 3.4% |
| Loss in an instant 20% fall to 424p, IV up 10 points, no more than £5,000 (10%) | Put worth 78.67p: (78.67p − 9.00p) × £10 = £696.69 | 7 (7.18; 7 contracts lose £4,876.82) | £1,393.38, 2.8% |
| Requirement after the fall | (78.67 + 84.80)p × £10 = £1,634.69 | Not a line: a consequence | £3,269.38, 6.7% of the account after the loss |
The margin line binds first, at two contracts; the gap line alone would allow seven. After the fall, the requirement for two contracts nearly doubles, to £3,269.38, on an account that has just lost £1,393.38 on this position alone (£48,606.62 left, before anything else in the book). Margin rises exactly when equity falls, which is why an undefined-risk line is set well below what the account could meet today.
Writing the same put cash-secured needs £5,000 of collateral per contract, 10% of the account, which is outside the 5% collateral line. Margin reduces the capital committed, not the risk: if BP fell to 265p, half its price, the put would cost £2,260.00 per contract at expiry whether it was cash-secured or margined, and £4,912.80 with the opening and assignment commissions if BP went to zero. A 20% gap is a test size, not the worst case.
An index example shows the other failure. The short strangle page writes one FTSE 100 October 10,200 put and 11,250 call for £905.00: its Cboe-formula requirement at entry is £12,030.00, 24.1% of a £50,000 account, and an instant 20% fall with implied volatility 20 points higher marks it at −£15,474.40, 30.9%. It fails both lines at one contract, and an ESX contract cannot be traded in fractions; at £10 a point, one contract's exposure is £107,500 of index.
HISTORY
Case file: March 2020, a fall beyond the gap test, taken in steps
On 11 March 2020 the Bank of England cut Bank Rate by 50 basis points to 0.25% and the UK countercyclical capital buffer to 0% (Bank of England, 11 March 2020). On 19 March a special meeting of the Monetary Policy Committee cut Bank Rate to 0.1% and raised the Bank's bond holdings by £200bn to a total of £645bn (Monetary Policy Summary, 19 March 2020). Price data (Yahoo Finance): the FTSE 100's lowest close came after both, on 23 March, 32.6% below its close on 21 February; its largest one-day fall in those 21 sessions was 10.9%, on 12 March.
This page's gap test applies one instant 20% move (the margin line and the gap test). No single day in March 2020 reached it, yet over the month the index fell further than that (price data), so a position that passed the test at entry could lose more than the test allows without any one day breaking it. On the two BP puts above, the requirement after a 20% fall nearly doubles, to £3,269.38 (the gap test table); in a fall taken in steps it rises at each step while the account's equity falls.
The gap test's move is a size for one test, not a limit on how far a market can travel while a position stays open, and here the low came after the Bank had acted twice.
Banks, oil companies and miners move in groups
The per-underlying line treats HSBC and Barclays as separate names. Their share prices do not behave separately. Over the 250 trading days to 17 August 2026, daily returns on the large UK banks, oil companies and miners moved closely within each group and much less across groups:
| Pair | Correlation | Group |
|---|---|---|
| Barclays and Lloyds | 0.83 | Banks |
| HSBC and Barclays | 0.73 | Banks |
| HSBC and Lloyds | 0.70 | Banks |
| BP and Shell | 0.82 | Oil |
| Rio Tinto and Anglo American | 0.76 | Miners |
| Glencore and Anglo American | 0.69 | Miners |
| Rio Tinto and Glencore | 0.62 | Miners |
| BP and HSBC | −0.14 | Across groups |
| HSBC and Tesco | 0.02 | Across groups |
Three equal positions in names that move with a correlation of 0.75 swing, together, about 91% as much as one position three times the size; if the three were unrelated, the combined swing would be about 58%. Three bank spreads that each pass the 10% line for one underlying can therefore add up to something close to a single 30% position in UK banks, which is what the 15% line for a group of related names is there to catch. The teaching book's three names sat in different groups, and over this window their cross-correlations were close to zero or slightly negative, which is part of why its beta-weighted figure is small. A book of HSBC, Barclays and Lloyds spreads would not have that cushion. A correlation measured over a year is an average; it is not a promise about any single bad day.
A stop on an option is a trigger, not a price
A stop order sells once the price reaches a level; it does not promise that level. Options make the gap between trigger and fill wider than it is for shares: the quoted spread is wide on many ICE series, half of it is paid on each trade, and an option's price can jump past the stop when the share opens lower or when implied volatility moves. Take the BP December 500 call bought at 49.25p, with a stop to sell at 24.50p, about half the price, meant to cap the loss at £247.50. If BP opens lower the next morning (122 days left), the call's model value at the open is:
| Overnight move | Call at the open | What the stop does | Loss per contract |
|---|---|---|---|
| BP −10% to 477p, IV unchanged | 19.25p | Fires at the open, below the stop level | £299.99 |
| BP −10% to 477p, IV up 5 points | 24.55p | Does not fire: the rise in volatility holds the call above the stop | £246.97 on the mark |
| BP −15% to 450.5p, IV up 5 points | 14.57p | Fires at the open | £346.84 |
| BP −20% to 424p, IV up 5 points | 7.74p | Fires at the open | £415.07 |
The same 10% fall either fires the stop or leaves it untouched, depending on what implied volatility does overnight, and when it fires the loss is set by the opening price, not by the stop. On a bought option the premium is still the most that can be lost, so the stop only changes where in that range the loss lands. On a written option there is no such floor, and a gap through a stop is how a small planned loss becomes a large one. The alternatives each move risk rather than remove it: a defined-risk spread caps the loss by construction and costs part of the premium; an alert on the share price, rather than a stop on the option, avoids selling into a wide quote but needs someone to act; a time-based exit such as the 21-day convention limits how long a position is exposed; and a smaller size limits what any of them can cost.
A dollar position in a pound account
Model inputs. A hypothetical US share at $100 (not a real company), a 45-day 95 put, IV 30%, US rate 3.625% (the midpoint of the Federal Reserve's target range on 17 August 2026), American, priced on the binomial tree: model value $1.95, filled at $1.95 on the $0.05 tick. Pounds at an illustrative $1.356 per £1 on 17 August 2026 (ECB reference-rate cross, 1.3559); a broker's own rate will differ. The example uses the standard 100-share US equity contract.
Written cash-secured, the put brings in $195.00, £143.82, and ties up $9,500 of collateral: £7,006.42, 14.0% of a £50,000 account. The 5% collateral line would need an account of £140,128.33. If the shares went to zero the put would cost $9,305.00, £6,862.60. None of those pound figures is fixed. At $1.25 per £1 the same $9,500 of collateral is £7,600.00, 15.2% of the account: £593.58 more, with nothing about the trade changed. A dollar position therefore carries two exposures, one to the share and one to the pound, and a sizing line measured in pounds moves with the exchange rate. The planner's US example sizes a put spread on the same hypothetical share in pounds.
For tax, each leg is converted to pounds on its own date, never as a net dollar figure: the tax page's FX rule works an example. Trading US options from a UK account, including the W-8BEN form and US trading hours in UK time, is covered on the US options page.
UK rules that change the arithmetic
Spread bets and CFDs have margin floors; listed options do not
For spread bets, CFDs and other restricted speculative investments sold to retail clients, the FCA sets minimum margins: 5% of the exposure on a major stock index and 20% on a single share (COBS 22.5.11R), a close-out when equity falls below half the required margin (COBS 22.5.13R), and a limit on a retail client's liability to the funds in the account (COBS 22.5.17R, negative-balance protection). Listed options bought and written on an exchange are outside those rules. No FCA minimum margin, mandatory close-out or negative-balance protection applies to them: margins are set by the clearing house and the broker, and a written option can lose more than the account holds, leaving a debt to the broker. The Level 3 page covers what that means in practice, and the comparison page sets the three routes side by side.
Compensation covers a broker's failure, not a trading loss
The FSCS protects investments up to £85,000 per person, per firm, when an authorised firm fails and cannot return clients' money or assets. It does not pay for losses on the investments themselves; its investments page says it "can't accept any claims that are for poor investment performance" (FSCS, checked 28 September 2026). Protection against a trading loss has to come from the positions themselves: their size and their structure.
Cash or margin account
Writing uncovered options and most spreads needs a margin account. Covered calls and fully cash-secured puts can be written in a cash account where the broker permits it: US Regulation T allows a broker to sell an option for a customer in a cash account "as part of a covered option transaction" (12 CFR 220.8(a)(3)), and IBKR's cash account allows covered calls with the shares restricted and naked puts with the cash restricted. The account types page maps broker permission levels to this course's levels.
What a loss is worth after tax
Options cannot be held in an ISA, so gains and losses fall in a general investment account (the wrappers page explains why). A bought option that lapses is a disposal at a loss in the tax year it lapses (TCGA 1992 s144(4); see bought options on the tax page). If the BP December 500 call lapsed on 18 December 2026, the £493.90 loss would reduce 2026/27 gains, worth £88.90 at 18% or £118.54 at 24%, assuming the £3,000 annual exempt amount is used by other gains. The loss after tax is then £405.00 or £375.36. With no gains to set it against, the loss carries forward; it has to be claimed within four years of the end of the tax year. The framework sets its lines on the loss before tax, because the relief depends on other gains that may not exist.
Strategy pages that apply this framework
Each strategy page states its own maximum loss, collateral or stress figure and links here for the method:
How these numbers are calculated
The pricing model, betas and stress inputs behind the sizing figures
- The BP, HSBC, Tesco and FTSE 100 option values and Greeks in the sizing tables are the pricing model's, at the inputs of the methods page's assumptions table for Monday 17 August 2026: Black-Scholes-Merton for American calls with no ex-date in their life; a Cox-Ross-Rubinstein binomial tree (200 and 201 steps, averaged) for American puts and for calls with an ex-date, with dividends as discrete amounts (escrowed-dividend approximation). Greeks of tree-priced options are bump-and-reprice on the same tree. Time is exact calendar days ÷ 365. Fills are on the tick: 0.25p on BP, HSBC, Tesco and the minis; $0.05 on US options below $3.
- Delta-£ = share-equivalents × share price in pounds. Vega is £ per volatility point; theta is £ per calendar day. Beta-weighted £ = delta-£ × beta. Betas and correlations are estimated from 250 daily log returns of closing prices to 17 August 2026 (price data: Yahoo Finance, checked 28 September 2026); beta = covariance with the FTSE 100 ÷ variance of the FTSE 100.
- Stress rows move every share by the stated percentage at entry and add the stated parallel shift to each option's implied volatility. Marks are model values, before the bid-ask spread.
- Runs of losses: with q the chance of a loss, the chance of no run of r losses in n independent trades, Qn, is 1 for n < r, 1 − qr for n = r, and Qn = Qn−1 − (1 − q) qr Qn−r−1 above that. The page shows 1 − Qn, computed exactly (it can be rebuilt in a spreadsheet column).
- Drawdown tables: each trade risks a fixed share f of the current account and either gains b × f or loses f. The chance of a fall of D from the running peak within 100 trades is computed exactly, by tracking the probability of every possible distance below the peak after each trade; a seeded simulation of 200,000 paths per cell agrees to within 0.2 percentage points in all 44 cells.
- Kelly fraction f* = p − (1 − p) ÷ b. Long-run growth per trade = p ln(1 + b f) + (1 − p) ln(1 − f).
- Uncovered margin uses the Cboe strategy-based formula as a yardstick; brokers set their own requirements on ICE contracts. Tax lines use 2026/27 rates of 18% and 24%, with the £3,000 annual exempt amount assumed used by other gains.
- Probabilities are model probabilities under stated inputs, not forecasts. BP, HSBC and Tesco are model underlyings; the dates are fixed in the past, and nothing on this page is a view on any company or a recommendation to trade.
Continue in the options library
- Options hub: all 26 strategies
- Options basics: start here
- Options glossary
- Greeks, pricing and put-call parity
- Implied volatility, IV rank and skew
- Assignment and expiry
- UK options tax: worked examples
- Reporting options on SA108
More strategies and guides
How UK Tax Drag holds itself to account
Pages are checked against the primary sources listed on each page and reviewed by the Editor; corrections are logged in the changelog. Pages are drafted with the help of AI tools (how AI is used). No affiliate revenue. No sponsored content. No paid placements.
UK Tax Drag is an independent publication by Finsolve Consulting Limited, not affiliated with or endorsed by HMRC, GOV.UK or any government body.