Why a 90% Win Rate Can Still Lose Money

The most quoted number in premium selling

Every options seller eventually learns their win rate, and most of them learn it early, because it is the easiest statistic to produce and the most pleasant to look at. Sell puts far enough from the money for long enough and 85, 90, even 95 percent are all reachable.

The number is real. It is also, on its own, almost uninformative about whether the strategy makes money.

This is not a subtle point about statistics. It is arithmetic, and it takes about two minutes to work through. Premium selling produces a payoff shape where the wins are small and frequent and the losses are large and rare, so the frequency of winning and the profitability of winning are close to independent. A strategy can win nine times out of ten and lose money. A different one can win six times out of ten and compound steadily.

Expectancy, in one line

What decides the outcome is expectancy — the average result per trade across many trades:

Expectancy = (win rate × average win) − (loss rate × average loss)

Both halves matter, and the second half is the one nobody quotes in a screenshot.

Take a 90% win rate. Nine trades in ten return an average of $120. One in ten loses an average of $1,080.

(0.90 × 120) − (0.10 × 1080) = 108 − 108 = 0

Zero. Ninety percent of the trades worked, the trader felt competent for months, and the strategy returned nothing before costs. After commissions and the spread it is a loss.

Now move one variable. Same 90% win rate, same $120 average win, but the average loss is $600 rather than $1,080:

(0.90 × 120) − (0.10 × 600) = 108 − 60 = +48

Forty-eight dollars a trade, from an identical win rate. Nothing about how often the trader was right changed. What changed was what happened on the days they were wrong.

Why premium selling has this shape by construction

The asymmetry is not bad luck or poor execution. It is what selling an option is.

When you sell a cash-secured put you receive a premium that is known, fixed and small relative to the capital committed. That is the entire upside, and no outcome, however good, can exceed it. If the underlying rises fifty percent you still collect exactly the premium. Your maximum outcome is capped on the day you open the trade.

The downside is not capped in the same way. It is bounded — the underlying can only fall to zero — but the bound sits far below anything the premium contemplates. A contract paying $120 against $5,000 of committed capital is offering 2.4% for accepting a decline that could plausibly run to 20% in a bad month.

So the ratio between the typical win and the typical loss is baked in before any judgement is applied. The seller is not choosing whether to have lopsided payoffs. They are choosing how lopsided, and how often the bad side arrives.

This is the same shape an insurer runs, and insurers do not evaluate themselves on the proportion of policies that go unclaimed. That figure is near 100% and tells them nothing. They price the claim.

What a high win rate actually tells you

It tells you one thing precisely: how far from the money you have been selling.

Move the strike further out and the win rate rises, because the underlying has to travel further to reach it. It rises because the premium falls, and it falls roughly in proportion to the risk given up. The market is not offering a free improvement — it is offering a different point on the same curve.

Which means a rising win rate, on its own, is as consistent with a deteriorating strategy as an improving one. A seller who reacts to a bad month by moving strikes further out will see their win rate climb and their expectancy fall, because the average loss barely changes while the average win shrinks. The number that looks like progress is measuring the retreat.

The two numbers that make a win rate meaningful

A win rate is interpretable when it comes with two companions.

The average loss, against the average win. This is the ratio that decides everything. If the average loss is nine times the average win, a 90% win rate is exactly breakeven, and every fraction of a percent below 90 is a loss. If it is four times, a 90% win rate is comfortably profitable and 80% is exactly breakeven. Without this ratio the win rate is a number without a denominator.

Whether the record is held to expiry. A strategy that closes winners early at a profit target and lets losers run to expiration will report a spectacular win rate that describes its exit policy rather than its entry quality. Two records showing 90% can mean entirely different things: one where every contract ran to settlement and 90% finished out of the money, and one where profitable positions were closed at 50% of maximum and losing positions were held hoping for recovery. Only the first tells you anything about strike selection.

The failure mode this produces

The practical damage is not that people misjudge a strategy. It is that a high win rate is psychologically expensive in a specific way.

Months of small wins build confidence, and confidence gets expressed as size. The seller who spent six months collecting $120 a week without a single loss does not stay at one contract. By the time the losing month arrives they are at four, and the loss lands against a larger book than the wins were collected on. The sequence — many small wins, then one large loss, at a size set by the wins — converts a strategy with positive expectancy into an actual loss for that particular trader.

Nothing in the win rate warns about this, because the win rate is highest immediately before it happens.

How StratosIQ reports it

The record holds every contract to expiration and scores it on where it finished. Above the strike is a win, below it is a loss, and there is no early-close policy that could flatter the figure. That decision costs the published win rate several points against what a profit-target methodology would show, and it is the only version of the number that says anything about entry quality.

Losses are recorded as losses at full weight, including contracts that were assigned on underlyings that later recovered. There is no category for a loss that was really a planned entry — assignment covers why that distinction matters more than it sounds.

The full published record, contract by contract and including every loss, is on the performance page.

The reason safety carries the largest weight in the score rather than premium is this arithmetic. Increasing the premium raises the average win by a few dollars. Reducing the depth of the bad outcome moves the term that the expectancy calculation is actually sensitive to.

Worth reading next

Probability of profit covers the entry-day estimate of a strike holding, which is the forward-looking version of the same question this article asks backwards. Sector concentration explains why the losses do not arrive independently, which is what makes the tail heavier than the per-contract arithmetic here suggests.