← AlgoRishi Blog · algorishi.com

You win 2 out of 3 trades and still lose money. Here is the arithmetic

9 September 2026 · AlgoRishi

Here is a message we get in some form almost every week: "I win more trades than I lose. My account is still smaller than last year. What am I doing wrong?"

Usually, nothing mysterious. The arithmetic is simply not on your side, and win rate is hiding it from you.

66%win rate in the example
−220points lost over 100 trades
₹20value of 1 point, 1 SENSEX lot

Win rate answers the wrong question

Win rate tells you how often you are right. It says absolutely nothing about how much you make when you are right, or how much you give back when you are wrong.

Risk-reward ratio (R:R) is the other half: your average win divided by your average loss. On its own it is equally useless. A trader with 5:1 R:R who wins once in twenty tries is broke too.

Only the two together mean anything. The number that combines them is called expectancy:

Expectancy = (Win rate × Average win) − (Loss rate × Average loss)

The SENSEX example that ruins the illusion

SENSEX weekly options: BSE, Thursday expiry, lot size 20, strikes in steps of 100. One premium point is ₹20 per lot. Keep that conversion in your head — it is the only maths you need here.

Say you buy weekly options and you are genuinely good at picking direction. Over 100 trades you win 66 and lose 34. Very respectable.

But you take profit fast, because a green number on the screen is uncomfortable to hold. Average win: +8 points. And when a trade goes against you, you wait for it to "come back". Average loss: −22 points.

At ₹20 a point, that is −₹4,400 per lot. Trade five lots and you are down ₹22,000 — while being right two times out of three. And that is before brokerage, STT, exchange charges, and the bid-ask spread you cross on the way in and again on the way out.

9 trades: 6 small wins, 3 big losses — and the running total goes down 0 +8 pt wins −22 pt −22 pt −22 pt running total ends at −18 points
Six green bars, three red ones, and the brown line — your account — still finishes below zero.
The auto-rickshaw version: you can complete forty ₹60 fares and feel like a busy, successful day. One ₹3,000 engine repair erases thirty of them. Nobody counts the repair when they count the fares.

How much R:R does your win rate actually need?

There is a clean break-even formula. To merely stay flat, your average win divided by your average loss must be greater than:

(1 − win rate) ÷ win rate

Your win rateMinimum avg win : avg loss just to break even
30%2.33 : 1
40%1.50 : 1
50%1.00 : 1
60%0.67 : 1
66%0.52 : 1
75%0.33 : 1

Read the 66% row again. At a two-thirds win rate you only need your average winner to be about half the size of your average loser to break even. Our example trader had 8 against 22 — a ratio of 0.36. Below the line. That single gap is the entire loss.

Two levers, and only two

Once you accept the formula, the fix stops being about finding better tips.

Lever one — make the average loss smaller. Keep everything else identical but cut the average loss from 22 points to 10, and the same 66 wins now produce 528 points against 340 lost: net +188 points. Nothing about your entries changed. A stop-loss you actually honour did all of it.

Lever two — make the average win bigger. Keep the 22-point losses but let winners reach 15 points instead of 8, and you get 990 against 748: net +242 points. This is what a scale-out plus a trailing exit is for — book a piece to make holding bearable, let the rest run.

Both levers live entirely in the exit. Neither requires you to predict the market better than you already do. This is why exits, not entries, are where most retail F&O accounts are actually lost.

The weekly-expiry wrinkle

On SENSEX weeklies the loss side gets an extra push that index traders underestimate: theta. As Thursday approaches, an option that simply sits still is losing value every hour. "Waiting for it to come back" is not a neutral act on Wednesday afternoon — the clock is charging you rent.

That is also why a fixed, points-based stop behaves more predictably than a mental one. Twenty-two points is twenty-two points whether you hold one lot or ten; ₹20 per point per lot does not negotiate.

What to actually go and measure

Open your last fifty trades. Write down four numbers: how many you won, how many you lost, your average winning points, your average losing points. Put them into the expectancy formula. Most traders discover their win rate was never the problem, and that one or two outsized losses ate an entire quarter of small, honest wins.

This is general education about trading arithmetic, not investment advice and not a recommendation to buy or sell anything. Options trading carries a real risk of losing your entire capital, and SEBI's own studies show the large majority of individual F&O traders lose money.

Win rate is the number you post in the group. Expectancy is the number that pays your EMI. They are not the same thing, and only one of them is on your side.

Disclosure: AlgoRishi publishes signal tracking on real market prices using virtual capital, for information and education only. Nothing here is investment advice or a recommendation to trade. Futures & options trading involves substantial risk of loss. Past performance is not indicative of future results. AlgoRishi is not a SEBI-registered investment adviser or research analyst.