The report covers every closed trade: the net P&L after charges, the win rate, the average win and the average loss, the reward-to-risk ratio, the expectancy per trade, the maximum drawdown in rupees and in per cent, how long the deepest drawdown lasted, the longest winning and losing streaks, and the P&L by day and by month, with the full trade list and a CSV download. Nothing in it is estimated: an open trade is not counted, and a measure that cannot be worked out shows a dash rather than a guess. For live trading, each algo's own report builds from its real fills as it trades, its card and activity log show what it did today, and your P&L history shows what your whole account made, day by day, from your broker's own fills.
What QuantPower's strategy reports show
| Part | What it shows |
|---|---|
| Net P&L | The total of every closed trade, after charges |
| Wins and losses | How many trades made or lost money; a trade at exactly zero is neither |
| Win rate | The share of trades that made money |
| Average win and average loss | The typical size of each outcome |
| Reward-to-risk | Average win divided by average loss |
| Expectancy | What one trade made on average: net P&L divided by the number of trades |
| Maximum drawdown | The deepest fall of the equity below an earlier high, in rupees and in per cent |
| Return to drawdown | Net P&L divided by the maximum drawdown |
| Drawdown duration | The longest time the equity stayed below an earlier high, in days and in trades |
| Streaks | The most wins and the most losses in a row |
| Day and month | P&L by day, and a month-by-year grid with totals |
| Trades | Every trade with entry and exit time, side, quantity, prices, exit reason, charges and P&L; CSV download |
| Live report per algo | The same measures for every algo running live, from its real fills, shown beside its backtest |
| Price | Reports are free; a backtest costs 2 tokens a run after 10 free a week |
A short history of measuring strategy performance
For most of the twentieth century, an investment was judged by one number: how much it made. In 1966 William Sharpe proposed judging mutual funds by their return per unit of risk, measured by how much their returns varied, and the Sharpe ratio became the standard way to compare one fund with another.1 The idea that a result means little without the risk taken to get it is the starting point of every performance report since.
Traders of futures needed a different measure of risk. A trend-following system can be very profitable over years and still lose a third of its capital along the way, and what decides whether a trader survives is that loss, not the variation of monthly returns. Drawdown, the fall from a peak to a low, became the risk measure of the commodity trading world. In 1991 Terry Young published the Calmar ratio, the annual return divided by the maximum drawdown, as a way to compare trading advisers.2 QuantPower's "return to drawdown" is the same idea, measured over the period of the report.
Individual traders brought their own measures. Writers on trading systems in the 1990s, Van Tharp among them, taught traders to think in terms of expectancy: what each trade makes on average once wins and losses are both counted.3 A strategy with a low win rate and large wins can have a high expectancy; a strategy that wins nine times in ten can have a negative one.
In India, the need for honest measurement has become hard to ignore. A SEBI study published in September 2024 found that about nine in ten individual traders in equity futures and options made a net loss over the three years it examined.4 A trader who measures every strategy on its full record, after costs, at least knows which group his own strategies belong to.
Where the reports come from
QuantPower produces reports in four places:
| Where | What it covers | Built from |
|---|---|---|
| Backtest | A rule tested on past candles | The trades the rule would have closed |
| Forward test | The same rule running on live candles, with no real orders | The trades it has closed since it started |
| Live report | One algo running on your broker account | The fills your broker reported for that algo's orders |
| P&L history | Your real account, day by day and month by month | Your broker's own fills |
The backtest, the Forward test and the live report use the same measures, worked out by the same code, so they can be compared line for line. That comparison is the heart of reviewing a strategy: did it behave on live prices, and then with real money, the way it behaved on past ones?
The live report for each algo
Every algo you run live has its own report, opened from its card. It counts only that algo's trades, from the fills your broker reported for the orders the algo placed, after charges, and it grows with every trade the algo closes. The report opens beside the algo's latest backtest, with the same rows in the same order, so a gap shows at a glance: a win rate ten points lower live, an average loss larger than tested, a drawdown already deeper than the backtest's worst. Trades you placed by hand on the same instrument are not mixed in, because the report follows the algo's own orders.
Beside the live report, the algo's card shows its state, today's legs with their fills and its M2M for the day, and the activity log records every entry, exit, re-entry, change and refusal in one plain line each, with a "Copy log" button for any day. Your account's P&L history then shows what all your trading, manual and automated, made each day from the fills your broker reported.
How to read the report
No single figure tells you whether a strategy works. Read them together.
| Measure | What it tells you | Watch for |
|---|---|---|
| P&L by day and month | Whether results build steadily or come from a few days | A total carried by one or two large days |
| Maximum drawdown | The deepest fall from a peak | A Forward test drawdown beyond the backtest's worst |
| Drawdown duration | How long the strategy took to recover | Months below a peak that you would not sit through |
| Win rate | The share of trades that made money | A high rate with average losses far larger than average wins |
| Average win and loss | The size of a typical outcome | Average losses growing over time |
| Expectancy | Whether each trade earns anything at all | A figure near zero once charges are counted |
| Streaks | The worst run of losses | A losing streak longer than you could stand live |
| Trade list | What actually happened on each trade | Exits that do not match the rules |
Our guide to backtest performance metrics explains each measure and its formula in more depth. The tutorial on how to read backtest results walks through the report screen itself.
How the figures are worked out


The definitions matter when you compare two reports, so here they are exactly as the report uses them:
- A win is a trade with P&L above zero after charges; a loss is below zero. A trade at exactly zero is neither, and it ends both a winning and a losing streak.
- Average win and average loss are the mean P&L of the winning and the losing trades. The average loss is shown as a negative number.
- Reward-to-risk is the average win divided by the size of the average loss.
- Expectancy is the net P&L divided by the number of trades.
- Maximum drawdown is the deepest fall of the running total of closed-trade P&L below its earlier high, in rupees. In per cent, it is measured against the largest position value in the run, the entry price times the quantity.
- Return to drawdown is the net P&L divided by the maximum drawdown.
- Drawdown duration runs from the trade that set the high to the trade that got back to it, or to the last trade if it never did.
- Day and month P&L use the date of each trade's exit, in Indian time.
Only closed trades count. A trade still open when you read the report is not in any figure until it closes.
How many trades before you judge?
A report of ten trades says very little. Chance alone can turn a weak strategy into a good-looking one over ten trades, or a sound one into a poor-looking one. The fewer the trades, the wider the range of results the same rules could have produced.
Some rough guides traders use:
| Trades in the report | What you can say |
|---|---|
| Under 20 | Almost nothing about the edge; check only that the trades follow the rules |
| 20 to 50 | Whether the strategy behaves roughly as tested; large gaps are worth investigating |
| 50 to 100 | A first view of win rate, expectancy and drawdown |
| Over 100 | A reasonable basis for keeping, resizing or retiring |
A strategy that trades once a week takes two years to reach a hundred trades. That is one reason to backtest over a long period first, and to treat the first months of a Forward test or live run as a check on behaviour rather than a verdict on profit.
Drawdown and the gain needed to recover
Drawdown hurts more than its size suggests, because a loss is measured from a higher base than the gain that repairs it. A 20% fall needs a 25% gain to get back to the peak; a 50% fall needs 100%.
| Drawdown | Gain needed to recover |
|---|---|
| 10% | 11.1% |
| 20% | 25% |
| 30% | 42.9% |
| 50% | 100% |
That is why the report shows the drawdown's duration as well as its depth. A strategy whose drawdowns are shallow but last for months can be as hard to live with as one with a sharp fall that recovers in a week. The max drawdown calculator works out the recovery needed for any drawdown.
Charges in the report
Every figure in a backtest and a Forward test report is after charges, and the report says how the charges were worked out. A strategy that looks profitable before charges can lose money after them, especially an options strategy that trades several legs a day.
For your real account, the P&L history uses the fills your broker reports. Many brokers' trade books do not include charges. Where they do not, the history shows charges as zero, with a plain note that they were not given and not taken off, and you should take the true charges from your contract notes. The note is there so that a gross figure is never mistaken for a net one.
Comparing the Forward test with the backtest
A Forward test should look like its backtest over the same kind of period, after costs. When it does not, the gap tells you where to look. The figures below are invented to show the method.
| Sample strategy | Backtest | Forward test, first 40 trades |
|---|---|---|
| Win rate | 58% | 55% |
| Average trade after costs | ₹420 | ₹150 |
| Slippage per trade | ₹60 assumed | ₹290 |
| Worst drawdown | ₹18,000 | ₹9,500 so far |
In this sample the win rate is close to the backtest, but the average trade has fallen by ₹270, and ₹230 of that is extra slippage. The rules are working as tested; the fills are not. The fix is the order type or the entry timing, not the strategy's logic.
The report states the slippage it used, for example "0.5% per side" or "no slippage (fills at the signal price)", so you always know what a backtest assumed. A backtest run with no slippage will almost always look better than the same strategy live.
Comparing live results with the Forward test
Once a strategy runs live, compare four things for the same weeks: the Forward test report, the algo's live report, its activity log and your P&L history.
- Trades. Did the live algo enter and exit on the same days as the Forward test? The activity log lists every entry and exit. A missing entry usually has a reason in the log: a refusal by a risk check, a broker that was signed out, or margin that ran short.
- Fills. Were live fills close to the Forward test's? The Forward test fills at the price you see; a live order meets the spread and, in thin strikes, a price that moves against it.
- Totals. Does the algo's live report for those days match what the Forward test would suggest, after charges, and does your account's P&L agree with it? Your broker's trade book does not always give charges; where it does not, the P&L history says so, and you should take them from your contract notes.
Small gaps are normal. Large ones have a cause, and finding it is worth more than any single month's result.
Reading the month grid
The month grid puts one row for each month and one column for each year, with a total for each year at the bottom. It is the quickest way to see whether a result is steady.
Look for three things. First, how many months are negative, and how large the worst one is. Second, whether the profit is spread across many months or comes from two or three. Third, whether the same calendar months are weak every year, which can point to something seasonal, such as low volatility in a holiday month or a run of event days. A grid where every year is carried by one large month is a strategy that depends on that month repeating.
Several strategies together
Most traders run more than one strategy, and the account goes through their combined result, not each one in turn. Two strategies that each lose ₹8,000 on their worst day can lose ₹16,000 together on a day when both are hit, which is exactly the kind of day that tends to hit both: a gap, an event, a sudden change in volatility.
When you review strategies together:
- Add up their daily P&L from each report's day list, or read the day total from your P&L history for live trading.
- Look at the worst combined day, not the sum of each strategy's worst days. The worst days may or may not fall together.
- Check that the account-wide controls fit. Your daily loss limit and Smart Exit stop should sit above the combined worst day you expect, with room for a gap.
- Watch for strategies that are secretly the same. Two short straddles on Nifty at different times of day carry much the same risk.
The account-wide controls in QuantPower read the combined position of every connected broker, so the limit you set covers all your strategies at once.
When to pause a strategy
Set your pause rules before the strategy goes live, when you can think clearly about them. Typical signs that a strategy needs a pause and a review:
- Drawdown beyond plan: the live drawdown passes the backtest's worst by the margin you set in advance.
- A longer losing streak than tested: more losses in a row than the backtest's longest losing streak.
- Slippage far above the assumption: fills keep coming in worse than the backtest allowed for.
- Trades that break the rules: wrong strikes, missed exits or duplicate orders. Pause until the cause is fixed.
- Changed market rules: a new lot size, expiry day, margin or STT rate since you tested it. Nifty's lot became 65 from January 2026, and STT on futures and options rose from 1 April 2026.5
- Events it was not tested on: budget day, an RBI policy meeting or election results.
A losing week inside the tested range is not, on its own, a reason to pause. Stopping a strategy at every dip means you keep its losses and miss the recoveries that the backtest counted on. Decide separately what to do with any positions a paused strategy still holds; the tutorial on how to pause automation covers the steps.
Pause, stop or stop and square off
An algo gives you three ways to step back, and they leave different things behind:
| Action | New entries | Open positions |
|---|---|---|
| Pause | None | Kept, with their targets, stops and exit time still watched |
| Stop algo, keep positions | None | Kept, but no longer watched: no stop-loss and no exit time on them |
| Stop algo and square off | None | Closed, then the algo stops |
Pause is the usual choice for a review: the open trade still has its protection. "Stop, keep positions" hands the positions to you, so use it only when you will manage them yourself.
After a pause: four choices
- Keep it if the cause was outside the rules, such as a broker outage or an untested event, and the numbers are within range.
- Resize it if the drawdown is more than you can carry but the strategy still earns per trade. Halving the lots roughly halves the drawdown in rupees.
- Change one thing, backtest the change, and Forward test it before going live again. Several changes at once make it impossible to know which one helped.
- Retire it if the expectancy after costs has stayed near zero or below over a long run of trades.
A worked example: reading a report from top to bottom
Suppose a Forward test of a breakout rule on Nifty futures has run for three months and closed 48 trades. Read its report in this order:
- Net P&L and expectancy. The net is positive and the expectancy is ₹310 a trade after charges. Each trade, on average, earns something.
- Win rate with average win and loss. The win rate is 42%, the average win ₹2,100 and the average loss ₹980. A low win rate with wins twice the size of losses is typical of a breakout rule. Reward-to-risk is about 2.1.
- Streaks. The longest losing streak is seven trades. Ask whether you would have kept trading after six losses in a row. If not, the strategy is not one you can run, however good the total.
- Drawdown and its duration. The maximum drawdown is ₹9,800 and it lasted 19 days and 11 trades. Compare with the backtest's worst. If the backtest's worst was ₹12,000, the Forward test is inside it.
- Month grid. Two of the three months are positive. One month carried most of the profit. That is not a fault, but it means the result depends on that kind of month coming round again.
- Trade list. Open the three largest losses. Did each exit at its stop, at the expected time? If one exit came much later than its rule, find out why before anything else.
Only when every step passes is the strategy ready to run live at a small size.
A worked example: a high win rate that is not good enough
Suppose a short straddle rule shows a win rate of 78% in its Forward test. It sounds safe. The report says:
| Measure | Value |
|---|---|
| Win rate | 78% |
| Average win | ₹1,150 |
| Average loss | −₹4,600 |
| Reward-to-risk | 0.25 |
| Expectancy | −₹115 a trade |
Each loss takes back four wins. Over many trades the strategy loses money, slowly, with a high win rate the whole time. The single figure that shows this at once is expectancy. Any strategy whose average loss is several times its average win needs a win rate high enough to pay for it, and this one does not have it.
Keeping your own notes beside the report
The report shows what happened. It cannot show why. A short note for each unusual day closes that gap: "broker signed out at 9:10, missed the entry", "budget day, paused by hand", "exit filled 40 points late, thin strike". After a few months the notes explain most of the gaps between the backtest, the Forward test and the live record, and they stop you from changing rules to fix what was really a one-off event. Keep the notes short, dated and next to the report you are reviewing, and copy the algo's activity log for any day you write about.
A review routine
- Daily, briefly: read each live algo's card and activity log. Note anything it refused or waited on.
- Weekly: P&L against plan, any trades that broke the rules, slippage and rejected orders.
- Monthly, in depth: compare the live weeks with the Forward test and the backtest over a similar period, check drawdown against your pause rule, and decide whether to keep, resize, change or retire.
- After any intervention: note what you did and why, so the record explains the numbers later.
Common mistakes
- Judging on the last week. A week is a handful of trades. Judge on the full record.
- Reading the win rate alone. It says nothing without the size of wins and losses.
- Comparing a no-slippage backtest with live fills. Give the backtest a realistic slippage before you compare.
- Changing the rules after every loss. Each change resets the record. You end up with many short tests and no long one.
- Ignoring charges. Expectancy before charges can be positive while expectancy after charges is negative.
- Reviewing strategies one at a time only. Add up their results by day, because the combined figure is what your account goes through on a bad day.
Where strategy measurement is going
Two changes are under way. Reports are getting stricter about what counts: closed trades only, charges included, slippage stated, nothing estimated. And the rules around retail algos, in force in India since 1 April 2026, put more weight on being able to show what an algo does and how it has behaved. A trader who keeps the backtest, the Forward test and the live record side by side has that record ready.
Pricing
Reports are free in QuantPower: reading a backtest report, a Forward test report or your P&L history costs no tokens. A backtest costs 2 tokens a run, and the first 10 each week are free. A Forward test of an algo costs 100 tokens per algo per 28 days, and a live algo 200 tokens per 28 days (1 token = ₹1, GST extra). See pricing.
Start with 500 free tokens
Connect your broker, build a strategy and test it on paper before a single live order goes out.
Frequently asked questions
Are the backtest, Forward test and live reports worked out the same way?
Yes. All three come from the same code and use the same definitions, worked out from the trades each one closed, so you can compare them line for line. The live report of each algo opens beside its backtest for that reason.
Does an open trade count in the report?
No. Only closed trades count. An open trade enters every figure when it closes.
What is expectancy in the QuantPower report?
The net P&L divided by the number of trades: what one trade made on average, after charges. A positive expectancy over a long run of trades is the first thing to look for.
How is maximum drawdown in per cent worked out?
It is the drawdown in rupees divided by the largest position value in the run, the entry price times the quantity.
Can I download the trades?
Yes. The trade list downloads as a CSV file, with entry and exit times in Indian time, side, quantity, prices, the exit reason, charges and P&L.
Should I pause a strategy after a losing week?
Not if the loss is within what the backtest and Forward test showed. Pause on the limits you set in advance, such as a drawdown or losing streak beyond the tested range.
Related guides
- Backtest meaning: what a backtest is, in short
- How to read backtest results: the backtest report, step by step
- Backtest trading strategies: the method, step by step
References
- William F. Sharpe, "Mutual Fund Performance", *The Journal of Business* (1966)
- Terry W. Young, "Calmar Ratio: A Smoother Tool", *Futures* magazine (1991)
- Van K. Tharp, *Trade Your Way to Financial Freedom* (1998), on expectancy
- SEBI, study on profit and loss of individual traders in the equity F&O segment, September 2024 (sebi.gov.in)
- NSE: revision in market lot of index derivative contracts from the January 2026 series; Finance Act 2026, STT on futures and options from 1 April 2026 (nseindia.com)
- QuantPower: report definitions, algo card, activity log and P&L history, from the app's code, checked 12 November 2026