Why a 90% win rate can still lose you money

The most persuasive number in bot marketing is the win rate. It is also the least useful one on its own. A strategy can win nine trades out of ten and still drain an account, and the reason is arithmetic rather than bad luck.

Two different questions

A win rate answers: how often does this win?

Expected value answers: what is this worth per trade, on average, once the size of the wins and losses is taken into account?

Only the second question tells you whether a balance grows. They are easy to confuse because a high win rate feels like an answer to both.

The Digit Differs example

Deriv's Matches and Differs contracts settle on the last digit of the price. There are ten possible digits, 0 through 9.

A Differs contract wins when the next digit is anything other than the one you named. Nine of the ten digits qualify, so it wins about nine times in ten. That is a 90% win rate, and it is genuine.

It is also a losing proposition. To offer a contract that wins 90% of the time, the payout on a win has to be small relative to the stake, while a loss costs the full stake. The nine small wins are sized to be worth slightly less than the one large loss. That gap is the margin.

Our Digit Differs bot says exactly this on its card: wins about nine times in ten, but a single loss costs far more than the wins it took to get there.

The mirror image

The Lucky Digit bot is the same trade inverted. It bets that the next digit is exactly the one you picked, which happens roughly one time in ten, and it pays about nine times the stake.

A 10% win rate sounds far worse. In expected value terms it is the same kind of bet, priced with the same margin. What changes is not the return but the shape of the ride: rare large wins instead of frequent small ones.

This is the useful insight. Win rate is a description of variance, not of profitability. Choosing a high hit rate is choosing a smoother experience, not a better one.

Why this matters for staking

The confusion becomes expensive when it meets a martingale multiplier.

A 90% win rate feels safe, so it invites increasing the stake after a loss. But losses on a 90% strategy are rare and large, and rare events cluster more than intuition expects. Two losses in a row on a 90% strategy is roughly a 1 in 100 event, which sounds remote until you notice that a bot running for an hour places a lot of trades.

The higher the win rate, the more a losing run costs, and the more tempting it is to chase. That combination is what empties accounts.

What to look at instead

When you are assessing a strategy, ask three things:

1. What does it win, and what does it lose? Not how often. How much.
2. What is the worst run it can produce? Set the max losses in a row control to something you would actually accept.
3. How many trades does it place? More trades means more exposure to the margin. A selective strategy is not smarter, it is just smaller.

The honest summary

None of the ready-made strategies on BinaryTick have a positive expected value, and the risk note on each one says so. That is not a limitation of these particular bots. It is a property of the instruments, and any bot on any platform trading them faces the same arithmetic.

What you can control is variance, exposure, and when you stop. Those are worth taking seriously. A win rate on its own is not.