Register
Tutorial

Review Your QuantPower Backtest Results

This tutorial reads a QuantPower backtest report line by line: the summary figures, the equity curve, the drawdown, the trade list, the monthly and day-wise tables, the plain warnings the report adds, and the unseen-days check on option strategy backtests. It ends with the checks to make before you trust a result, and how to decide what to do next. It assumes you have already run a backtest; the tutorial on how to backtest a strategy covers that.

QuantPower Research · Updated 12 Nov 2026 · 17 min read

A backtest report is a record, not a verdict. It shows what these exact rules would have done on those past days, after the costs you set. Reading it well means looking at the shape of the results, not only the total: how the money was made, how bad the worst stretch was, and whether the result depends on a few days that may not come again.

Before you start

  • A finished backtest. Either an option strategy backtest from an algo's card, or a rule backtest from the Backtest screen.
  • The question you asked. "Is a 30% stop better than 25% after costs?" or "Does this rule hold up on unseen days?" A report is easier to read against a question.
  • Your costs set honestly. Brokerage at your broker's rate and a slippage you would expect. A report with no slippage flatters every strategy.

The two reports

ReportFromWhat it covers
Option strategy backtestAn algo's cardDay by day, minute by minute, from entry time to exit time; results by day, month and weekday; exit reasons; the unseen-days check
Rule backtestThe Backtest screenTrades on one instrument over up to 2,000 candles; results by trade, month and day; streaks; expectancy and reward to risk

The two reports share most of their figures. Where they differ, this tutorial says so.

A short history of the backtest report

Early trading-system software in the 1980s printed a page of statistics for every test: net profit, number of trades, percentage of winners, largest loss, maximum drawdown. Those figures are still at the top of most reports today, because they answer the first questions any trader asks.1 What changed later was the attention to how results can mislead. Researchers showed that testing many versions of a rule and keeping the best one finds patterns in noise, and that a strategy's best backtest is usually better than its future.2 Modern reports add checks for that: results split by period, results on data the rules were not built on, and warnings written from the numbers.

Step 1: Read the headline figures

Start with the top of the report:

FigureWhat it tells youWhat to watch for
Net P&LThe total after brokerage, taxes and slippageCompare with gross P&L: if charges take most of it, the edge is thin
Trades or days testedHow much the strategy actually tradedFewer than 30 trades, or 60 days, is a small sample
Win rateThe share of trades or days that made moneyA high win rate with large losses is a warning, not a comfort
Average win and average lossHow the result is madeMany small wins and a few large losses, or the opposite
Best and worst trade or dayThe range of single outcomes, with datesOne best day larger than the whole net P&L means the result rests on it
Maximum drawdownThe largest fall from a peak, in rupees and in per centThe loss you must be ready to sit through

Read net P&L last, not first. The other figures tell you whether to believe it, and a report read in that order rarely surprises you later when the strategy trades live.

Step 2: Check expectancy and reward to risk

The rule backtest report adds two figures that combine the others:

  • Expectancy is the average result per trade: win rate × average win − loss rate × average loss. A strategy with a 40% win rate, an average win of ₹3,000 and an average loss of ₹1,500 has an expectancy of 0.4 × 3,000 − 0.6 × 1,500 = ₹300 a trade.
  • Reward to risk is the average win divided by the average loss: ₹3,000 ÷ ₹1,500 = 2.0 in the same example.
Win rateReward to riskExpectancy per ₹1,000 risked
30%3.0+₹200
40%2.0+₹200
50%1.0₹0
60%0.5−₹100
70%0.5+₹50

The table shows why neither figure works alone. A 70% win rate can be barely positive; a 30% win rate can be good. What matters is the two together, after costs.

Step 3: Read the equity curve

The equity curve is the running total, trade by trade or day by day. Look at its shape, not its end point:

  • A steady climb with small dips suggests a result spread across many trades.
  • A flat line with a few large jumps means a handful of days made the result. Find those days in the trade list and ask whether they could happen again.
  • A climb that flattens or falls in the later part may mean the market changed and the rules stopped fitting. Compare with the unseen-days check.
  • Long flat stretches are periods you would have run the strategy for nothing. Could you have kept running it through them?

A quick test of the curve's shape: cover the last third of it with your hand and ask whether the first two thirds would have led you to expect what the last third shows. If not, the strategy's behaviour changed during the test, and the unseen-days check in Step 9 will usually say the same.

Step 4: Read the drawdown

The drawdown is the fall from the highest point of the equity curve to the lowest point after it. The report shows:

  • Maximum drawdown in rupees, with its start and end dates, and in per cent of capital when you entered a capital figure.
  • Drawdown duration: how long it took to recover, in trades or days.
  • Return to maximum drawdown: net return divided by the worst drawdown. A figure above 2 means the strategy earned more than twice its worst fall over the period.

Ask one question of the drawdown: would you have kept running the strategy through it? A maximum drawdown of ₹40,000 on ₹2,00,000 of capital is a 20% fall. Many traders who say they could sit through that switch the strategy off halfway. If you would have, the backtest's total is not your total.

Step 5: Read the streaks

The report shows the longest winning and losing streaks. The losing streak matters more. A strategy with a 50% win rate will, over a few hundred trades, have losing streaks of six or seven in a row by chance alone. If your stops and your patience assume three, the backtest is telling you otherwise.

Use the longest losing streak to size the strategy: if each loss is ₹2,000 and the longest streak is seven, the strategy can lose ₹14,000 in a row before a win. Check that your account limit and your nerve both allow it.

Step 6: Read the trade list

Open the trade list and read actual trades, not only figures:

  1. Open three trades from the best month and three from the worst. Check that each entry and exit happened where your rules say it should.
  2. Look at the exit reasons on an option strategy backtest: how many legs closed on their stop, how many on their target, how many at the exit time, how many on the overall rupee limit. A strategy that mostly exits at the stop is a different strategy from one that mostly reaches the exit time.
  3. Check the time of the entries. If the rule is "enter at 9:20" and some trades entered at 9:25, the candle size may be coarser than you thought.
  4. Check trades where a stop and a target fell in the same candle. The candle cannot show which came first. Read those trades as stops.

Most first backtests find at least one rule that is not what the trader meant. The trade list is where you find it, usually within the first ten trades you open.

Step 7: Read the monthly and day-wise tables

The Monthly table shows the result for each month; the Day-wise table, or "by weekday" on an option strategy backtest, shows the result for each day of the week.

  • Monthly: count the losing months. If most of the profit comes from two or three months, the strategy depends on the conditions of those months returning.
  • By weekday: if the result comes almost entirely from one weekday, such as expiry day, the strategy is really an expiry-day strategy. Since September 2025, NSE expiries fall on Tuesday rather than Thursday, so an older backtest's "expiry day" moved.3

Step 8: Read the warnings the report adds

Every option strategy backtest comes with plain warnings, written from the numbers:

  • The test covers fewer than 60 days, so the period is short.
  • The win rate is 70% or more, but losing days are two and a half times winning days or more: one bad day can take back many good ones.
  • The worst single day lost more than half of the total profit.
  • The unseen days are weaker than the days the strategy was built on.
  • Charges take more than 40% of the gross profit.

Do not skip them. Each one describes a way a strategy that looks good in a report disappoints live, and each is written from your own strategy's numbers, not from a general rule.

Step 9: Read the unseen-days check

On an option strategy backtest, the report splits the period. The first 70% of the days are treated as the days the strategy was built on; the last 30% are unseen. The verdict compares the average result per day in each part:

VerdictWhat it meansWhat to do
Held up on unseen daysUnseen days averaged at least half of the built-on daysWorth a Forward test
Weaker on unseen daysMade money, but less than half as much a dayTest nearby settings; be cautious
Fell apart on unseen daysUnseen days lost money on averageThe rules probably fit the past only
Lost money even in the period it was built onDid not work even where it was fittedGo back to the idea
Too few days to judgeFewer than 20 built-on or 10 unseen daysTest a longer period

This is a quick first filter, not a full walk-forward test. A strategy that fails it rarely needs a longer test.

Step 10: Test the slippage

On a rule backtest, change the slippage and tap Recalculate with this slippage. Rerunning with only the slippage changed is free. Try double your first estimate. If the result turns from profit to loss, the strategy's edge is smaller than the cost of trading it in a fast market. Strategies that trade often, with small targets, are the most sensitive to slippage; strategies that hold for hours are the least.

A worked example: a straddle that looks good

A Nifty expiry-day short straddle is backtested over 18 months:

FigureValue
Days tested76
Win rate74%
Net P&L₹62,000
Average winning day₹2,900
Average losing day−₹5,050
Worst day−₹31,500
Maximum drawdown₹38,000
Charges18% of gross

Read in order:

  1. 76 days is enough to read, but not many.
  2. 74% win rate with losing days 1.7 times winning days: positive expectancy, but the losses are large.
  3. The worst day, −₹31,500, is half the net P&L. The report's warning says so. One day like it in the first month live would wipe out months of gains.
  4. The equity curve climbs steadily, with two sharp drops on two expiry days with large moves.
  5. The unseen-days check says "weaker on unseen days": the last 30% averaged less than half the first 70%.

The trader concludes the rules are not ready. She tests a version with an overall rupee stop of ₹12,000 for the day. The worst day falls to −₹13,800, the maximum drawdown to ₹21,000, the net to ₹49,000, and the unseen-days verdict becomes "held up". She runs the new version in Forward test.

A worked example: a rule backtest on futures

A rule backtest of an EMA crossover on Nifty futures, 15-minute candles, 2,000 candles (about 80 trading days):

FigureValue
Trades41
Win · loss15 · 26
Average win₹4,800
Average loss−₹1,900
Expectancy₹550 a trade
Reward to risk2.5
Loss streak7
Maximum drawdown₹17,600

A 37% win rate with a reward to risk of 2.5 gives a positive expectancy. The losing streak of seven at ₹1,900 each is ₹13,300 in a row, which the trader's account limit allows. 41 trades over 80 days is a small sample, so the trader reruns the same rule on hourly candles, where 2,000 candles cover more than a year, to see if it holds over a longer period before going further.

Reading results against your capital

Rupee figures alone can mislead, because the same result means different things at different sizes. Enter the capital you would trade the strategy with, and the option strategy report adds the return and the maximum drawdown in per cent.

CapitalNet P&LReturnMax drawdownDrawdown in %
₹2,00,000₹62,00031%₹38,00019%
₹5,00,000₹62,00012.4%₹38,0007.6%

The strategy is the same; only the capital differs. At ₹2,00,000 the drawdown is close to a fifth of the account, which many traders could not sit through. At ₹5,00,000 it is under a tenth. Use the per-cent figures to decide how much capital the strategy needs, not only whether it made money. And remember that capital here means what you set aside for this strategy, not the margin the broker blocks: a short straddle may block less margin than the drawdown it can produce.

Rule changes inside your test period

Indian derivatives rules changed several times between 2024 and 2026. A backtest that spans these dates is testing across several different markets:

ChangeFromWhat it does to a report
One weekly expiry per exchange; larger contract size20 November 2024Weekly strategies on indices that lost weekly expiry stop trading in the report after this date; rupee results per lot differ before and after
NSE expiry moves to Tuesday1 September 2025The "by weekday" table shifts: expiry-day results move from Thursday to Tuesday
Higher STT on futures and options1 April 2026Charges per trade rise in the later part of the period
F&O trading closes at 3:40 PM3 August 2026Exit times near the close have ten more minutes after this date

When a result is strong in the early part of the period and weak in the later part, check whether a rule change sits between them before you blame the strategy.

Comparing two versions

Most backtests are run to compare: a 25% stop against a 30% stop, an entry at 9:20 against 9:45, with or without an overall rupee limit. Comparing well takes a little discipline.

  1. Change one thing at a time. If you change the stop and the entry time together, you cannot tell which made the difference.
  2. Keep the period the same. A different period is a different test, whatever the rules.
  3. Compare the same figures in the same order: worst day, maximum drawdown, unseen-days verdict, then net P&L.
  4. Prefer the version that is steadier, not the one with the larger total. A version with ₹10,000 less net P&L and half the drawdown is usually the better one to run.
  5. Write down what you tested. After ten versions, memory is not reliable, and the best-looking result of ten is likely to be partly luck.

A useful way to lay out a comparison is a small table of your own:

VersionNet P&LWorst dayMax drawdownUnseen days
Stop 25% per leg₹48,000−₹9,200₹19,500Held up
Stop 30% per leg₹55,000−₹14,100₹26,000Weaker
Stop 35% per leg₹51,000−₹18,700₹31,000Weaker

In this example, the 30% version has the largest total, but the 25% version has the smallest worst day, the smallest drawdown and the only "held up" verdict. Many traders would choose the 25% version. Notice also that the result does not change smoothly from 25% to 35%: a strategy whose result swings with small changes to one setting is fragile.

Reading the exit reasons

On an option strategy backtest, every leg's exit has a reason: its own stop-loss, its own target, the strategy's overall rupee stop or target, or the exit time. The count of each tells you what kind of strategy you actually have:

Most legs exit onWhat it says
Their stop-lossThe stop is tight for the strategy's normal movement, or the entry is often wrong
Their targetThe strategy usually works quickly, or the target is close
The overall rupee limitThe legs move together; the account-level line does most of the work
The exit timeThe strategy holds through the day; its result depends on where the market closes

A short straddle where most legs reach the exit time is collecting time decay. One where most legs hit their stop is mostly paying for being wrong about range. Both can show the same net P&L.

Reading a backtest that Quanty or your AI app ran

Quanty, and AI apps such as Claude or ChatGPT connected through Connect AI, can run a backtest of your rules and show the result. They use the same engine and the same report, with the same warnings. Two habits help:

  • Ask for the report's figures, not a summary. The worst day, the drawdown, the unseen-days verdict and the warnings are what you need, and a summary can leave them out.
  • Check the rules it tested. An AI that built a strategy from your words may have chosen a strike rule, a stop or an exit time you did not intend. Read the strategy before you read the result.

Neither Quanty nor a connected AI app will tell you which strategy to trade. Both are software, not advisers.

What the report cannot tell you

  • How you will behave. A drawdown on a screen is easier to sit through than the same loss in your account.
  • What the next market will be. The period you tested covers some kinds of day and not others.
  • Fills in thin strikes. The test fills at candle prices with the slippage you set; a large order in a far strike may get a worse price.
  • Events outside the data. Exchange outages, broker outages, sudden margin changes and trading halts do not appear in candles.

These are reasons to run a Forward test after a good backtest, and to go live small after a good Forward test.

On a phone

The report reads the same way on a phone, one section at a time. Start with the headline figures, then scroll to the warnings and the unseen-days verdict before you look at the equity curve. Open the trade list in Report view to read each trade's entry and exit. A phone is a good place to read a report; for comparing several versions side by side, a laptop is easier.

Before you trust a result

CheckWhy
Net P&L after realistic slippageThe most common reason a backtest profit becomes a live loss
The worst day or trade against the totalOne outlier can carry the whole result
Results by month and by weekdayWhether the result is steady or concentrated
The unseen-days verdictWhether the rules fit only the past
Nearby settingsWhether a small change turns profit into loss
The trade listWhether the rules did what you meant
The period against rule changesExpiry days, lot sizes and STT changed in 2024 to 2026

Deciding what happens next

  1. If the result fails a check, change one thing and test again, or drop the idea. Do not change several things at once.
  2. If it passes, run it in Forward test on live prices for several weeks. Forward test costs 100 tokens per algo per 28 days.
  3. Compare the Forward test with the backtest for the same weeks. A large gap usually comes from fills, spreads or a rule that behaves differently live.
  4. Then go live small, and compare again. The guide to strategy performance covers the live report for each algo beside its backtest.

Common mistakes

  • Reading the total first. It is the least informative figure on its own.
  • Trusting a high win rate. It says nothing without the size of the losses.
  • Keeping the best of many tests. Test fewer versions, and read the unseen-days check.
  • Ignoring charges. If charges take 40% of gross, the edge is thin.
  • Reading a small sample as proof. Thirty trades can look good by chance.

Common problems

ProblemWhat to check
"Data is not available"Your broker has no candles for that instrument and period. Try a shorter period or another time frame
Very few tradesThe rule rarely triggers on that time frame. Try a smaller candle or a longer period
The equity curve is a straight lineCheck the trade list: the rule may never exit, or slippage may be zero
Results changed after rerunningSomething besides slippage changed; that is a new backtest
The report shows "too few days to judge"Test a longer period

Where backtest reports are going

Reports are moving from a page of totals towards checks that answer "can I trust this?": results on unseen data, warnings written from the numbers, and comparisons with what the strategy did live. As AI assistants make it easy to produce many strategies quickly, those checks matter more, not less.

Next steps

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

Which figure should I look at first in a backtest report?

The worst day or trade and the maximum drawdown, then the number of trades. They tell you whether the net P&L can be trusted and whether you could have lived through the strategy.

Why is my report's win rate high but the net P&L low?

The losing trades are large compared with the winning ones. Check the average win and average loss, and the expectancy.

Does recalculating with a different slippage cost tokens?

No. Rerunning the same backtest with only the slippage changed is free. Any other change is a new backtest at 2 tokens, with the first 10 each week free.

How many trades does a backtest need before I can trust it?

There is no fixed number, but fewer than 30 trades, or fewer than 60 days for a daily option strategy, is a small sample: a good result can come from luck. More important than the count is whether the result holds across months, across nearby settings and on the unseen days.

Why does my live result differ from the backtest?

Live trading meets the real spread, real fills, broker sessions that end and orders that are refused, none of which a backtest sees. Compare a Forward test with the backtest for the same weeks to find where the gap comes from.

What does "weaker on unseen days" mean?

The last 30% of the days made money but averaged less than half of the first 70%. The rules may fit the earlier period better than the later one.

References

  1. Robert Pardo, *Design, Testing and Optimization of Trading Systems* (1992), on the standard measures of a trading system test
  2. David H. Bailey, Jonathan M. Borwein, Marcos López de Prado and Qiji Jim Zhu, "The Probability of Backtest Overfitting", *Journal of Computational Finance* (2016)
  3. NSE, circular on the revision of expiry days of derivatives contracts, effective 1 September 2025 (nseindia.com)
Backtest results are simulated on past data and do not guarantee future results. The figures on this page are examples to show how a report is read, not results of any strategy. Trading involves substantial risk. This page is for information and is not investment advice.