Win Rate vs. Expectancy in Algorithmic Trading: What Decides Your Profit?
Blog by Aliceblue News Desk
11 November 2026
9 minutes
Your algo wins six trades out of ten, yet your account is lower than it was a month ago. If you’re new to algorithmic trading, this is the first thing to understand. Win rate counts how often you win. Expectancy shows what an average trade earns or loses in rupees, and that decides your profit. Below, you’ll learn how to calculate both with an Indian example, and how to read the results before you add more money.
Table of Contents
Quick Answer
Win rate is the percentage of trades that make money. Expectancy is the average profit or loss per trade: (win rate × average win) minus (loss rate × average loss). An algo can win most of its trades and still lose money if its losses are bigger than its wins. Always work it out after all costs.
What Is Win Rate vs Expectancy in Algorithmic Trading?
Win rate is the share of your trades that end in profit. If 60 out of 100 trades make money, your win rate is 60%.
Expectancy is the average amount your strategy earns or loses per trade, across many trades. It depends on four things: how often you win, how often you lose, and the size of a typical win and a typical loss.
In algo trading, a program places buy and sell orders using rules you set in advance. A program can place hundreds of trades, so you need one figure to judge the whole run. Win rate is the easy one, but it only counts trades. It doesn’t care whether a win was ₹200 or ₹5,000.
Picture a tea stall that earns ₹10 on most cups but loses ₹100 each time a customer sends back a full flask. Most sales make money. The month can still end in a loss.
What Is the Expectancy Formula?
Expectancy = (win rate × average win) minus (loss rate × average loss)
Win rate is the share of trades that made money. Loss rate is 100% minus the win rate. Average win is total profit from winners divided by the number of winners. Average loss is total loss from losers divided by the number of losers.
A positive result means the strategy made money on average. A negative result means it lost money on average.
How Does Expectancy Work, Step by Step?
- List your trades from a backtest or your live trade report. A backtest runs your rules on past price data to see what would have happened.
- Divide winning trades by total trades to get the win rate.
- Work out the average win and the average loss.
- Put the numbers in the formula.
- Take off brokerage, STT, exchange charges, GST, stamp duty, and SEBI fees. Also allow for slippage, which is the gap between the price your algo expected and the price it actually got.
- Check your work. Multiply expectancy by the number of trades. The answer should land close to your actual net profit. If it doesn’t, a charge or a count is wrong.
Many traders also measure results in “R”. One R is the amount you risk on a trade, fixed by your stop loss. A stop loss is an order that closes a trade at a loss you decided in advance. A trade that earns twice your risk is +2R, so trades of different sizes can be compared.
Example With Real Numbers
Say two algos trade Nifty 50 stocks such as Reliance and HDFC Bank. Each was tested on 100 trades, with total costs of ₹150 per trade.
Item | Algo A | Algo B |
Win rate | 65% | 40% |
Average win | ₹1,200 | ₹3,000 |
Average loss | ₹2,500 | ₹1,200 |
Win side | 0.65 × 1,200 = ₹780 | 0.40 × 3,000 = ₹1,200 |
Loss side | 0.35 × 2,500 = ₹875 | 0.60 × 1,200 = ₹720 |
Expectancy before costs | Loss of ₹95 | Gain of ₹480 |
Expectancy after ₹150 costs | Loss of ₹245 | Gain of ₹330 |
Result over 100 trades | Loss of ₹24,500 | Gain of ₹33,000 |
These are example numbers for learning, not a forecast or a recommendation.
Algo A wins almost two trades in three and still loses money. Algo B loses more than half its trades and comes out ahead, because its wins are much bigger than its losses.
The break-even win rate explains this. You can find it with 1 ÷ (1 + reward-to-risk ratio). Algo A risks ₹2,500 to make ₹1,200, so it needs about 68% wins just to break even, and it manages 65%. Algo B risks ₹1,200 to make ₹3,000, so it needs only about 29%. Costs push both figures higher.
In R terms, Algo B earns about +0.28R per trade (₹330 ÷ ₹1,200). That means roughly 28 paise for every rupee it risks.
Backtest vs Live Results
Many beginners calculate expectancy from a backtest and stop there. That’s where the surprises start.
A backtest runs in a tidy world. Every order fills at the price you wanted and nothing breaks. Live trading is messier. Orders fill a little late, prices move between signal and execution, and some stocks don’t have enough buyers and sellers at your price.
Run the algo with little money first and compare both sets of numbers. If backtest expectancy is ₹330 and live expectancy is ₹40, something is eating your edge. Slippage, costs, or overfitting are the usual causes. Overfitting means tuning your rules so closely to old data that they only fit that data. Find the cause before you add capital.
Advantages
- Expectancy lets you compare two very different algos on equal terms. One may trade often for small gains while another trades rarely for large gains. Their expectancy shows which one is worth running.
- It also shows what to fix. A negative number makes you ask whether the problem is win rate, the size of wins or the size of losses. And it warns you early about costs, because a fast algo with a thin edge can look fine until charges come off.
Risks and Limitations
- Expectancy comes from past trades, so a positive number doesn’t promise future profit. Markets change, and a pattern that worked last year may stop working.
- Losing streaks test your nerve. A profitable algo with a 40% win rate can lose five or six trades in a row without being broken. Some beginners switch it off just before it recovers. Others keep an algo with a high win rate and negative expectancy running, because the frequent small wins feel safe.
- Technical problems can also push live results below the backtest, such as a delayed order, an API error, or a thin market.
- Retail algo trading in India follows SEBI’s framework. SEBI issued its circular on safer participation of retail investors in algorithmic trading on 4 February 2025, and the rollout was extended in phases. Rules and broker requirements change, so check SEBI’s latest circulars and your broker’s current policy before you automate anything.
- Decide in advance when you’ll switch an algo off. A good trigger is a maximum drawdown, which is the fall from your account’s peak, or live expectancy staying negative over your last 50 trades.
Tips for Beginners
- Use at least 30 closed trades, and aim for 100 or more before you trust the number.
- Always use net figures, after all costs.
- Start with little money and risk a fixed amount per trade, such as 1% of capital.
- Compare live results with the backtest every month.
- Track five numbers: win rate, average win, average loss, expectancy, and your biggest single loss.
Win Rate vs Expectancy Compared
Parameter | Win Rate | Expectancy |
What it shows | How often you win | Average rupees per trade |
Includes trade size | No | Yes |
Includes costs | Not by default | Yes, if you use net figures |
Ease of use | Very easy | Needs a few more numbers |
Can mislead on its own | Yes | Less, but needs enough trades |
Best use | Planning for losing streaks | Deciding if a strategy is worth running |
Common Mistakes to Avoid
- Chasing a high win rate. Small wins and big losses can still add up to a loss, so check expectancy first.
- Ignoring costs. A small edge can disappear after charges, so use net figures.
- Judging an algo on 10 trades. That’s mostly luck, so wait for more data.
- Removing the stop loss to lift win rate. One bad trade can then wipe out weeks of gains, so keep it in your rules.
- Tuning the backtest too hard. Test your rules on data they haven’t seen before.
- Changing rules after two losses. Review a proper sample first.
Who Should Use Expectancy?
Anyone who trades by fixed rules and keeps a record of trades. That includes algo users, intraday traders, swing traders, and F&O traders. It’s a good starting point for beginners too, because it’s one formula that stops you from judging a strategy by win rate alone.
It matters less for long-term investors who make a few purchases a year. With so few trades, an average says little, and business quality and valuation count for more.
Conclusion
Win rate shows how often you win. Expectancy shows whether those wins pay. A 40% win rate algo can make money while a 65% one loses it, depending on the size of wins and losses. Before you trust any algo, work out expectancy after costs on at least 30 trades, ideally 100, and compare live results with the backtest. Then pull your last 50 trades and run the formula yourself.
FAQs
Blog by Aliceblue News Desk
