Paper trading vs backtesting: what each tells you

Paper trading vs backtesting: what each one tells you, how each can fool you, and how many trades you need before a win rate actually means something.

28 September 2026 7 min read #basics #tools

Paper trading vs backtesting comes down to past versus future. A backtest runs your rules over old prices and tells you, fast, whether they would have worked. Paper trading runs them forward in the live market with virtual money, slowly, with you making the calls. You need both, and in that order.

Both let you test a trading idea without risking money. A backtest replays historical prices and reports what would have happened. Paper trading uses real, live prices and virtual money.

People usually pick one and dismiss the other. Backtesters say paper trading is slow and too small to prove anything. Paper traders say backtests are just curve fitting. Each side is right about the other's weakness. That's exactly why they work well together, as long as you know which question each one answers.

What is backtesting good for?

A backtest's big strength is volume. Five years of one-minute candles is hundreds of thousands of bars. A rule that fires twice a day gives you a couple of thousand trades, and a computer can check them in seconds. Nobody could paper trade that many in ten years.

That makes a backtest the right tool for rejecting ideas. If a rule loses money over five years, through bull markets, bear markets and sideways chop, it's very unlikely to start working the week you start trading it. Most ideas fail here, and it's far cheaper to find that out in a spreadsheet than in your account.

How can a backtest fool you?

The trouble starts when a backtest passes. There are four common ways it misleads you.

Overfitting. Try enough combinations and one will look superb. An 8 and 21 EMA, then 9 and 21, then 9 and 20, then add a volume filter, then only on Tuesdays. The winner found the noise in that stretch of history, not an edge. The more settings you tweaked, the less the result means.

Look-ahead bias. The test uses information you couldn't have had when you made the decision. The classic mistake is using a candle's close to decide a trade at that candle's open. It's easy to do by accident, and it makes almost anything look profitable.

Fills that never happen. Most backtests assume you traded at the candle's close, or at your exact limit price. Charges are left out, or guessed at. On index options, a round trip on ten Bank Nifty lots costs around ₹160 (the working is in what a round trip actually costs). So a strategy that makes a rupee of premium per trade on average can look profitable in a backtest and lose steadily in real life.

Data that no longer exists. This one hits options. To backtest an option strategy you need the price history of the actual contracts you'd have traded, like last month's weekly 24,500 call, not the index. Those contracts expire. Most free sources that show charts for live contracts stop serving them once they do.

Expired-contract data is harder to get than most people expect. So many "option backtests" are really backtests of the index, plus a guess about what the option would have done. That guess is where the volatility effects from the index went nowhere and your option lost money quietly disappear.

What is paper trading good for?

Paper trading gives up volume and gets something a backtest can't have: the future. Every trade happens in a market that didn't exist when you wrote your rules. You can't overfit to prices that haven't happened yet, and you can't look ahead.

It also puts you in the loop. A backtest executes perfectly. It never hesitates, never skips a signal because the last three lost, never moves a stop. You will do all of those things.

Paper trading shows whether you can actually follow your rules at 09:20 on an expiry day. That tells you something about the strategy too. A set of rules that nobody can follow in real time isn't much of a strategy.

A good simulator also keeps the fills honest. Your order goes in at the price the market is at when you press the button. Charges match the real contract-note lines. The lot size is the exchange's. Margin on a short is what a broker would block.

How can paper trading fool you?

It has its own weak spots.

Small samples. At two trades a day you get about 40 in a month. That feels like plenty. It isn't.

Say you win 22 of your first 40 trades, a 55% win rate. The standard error tells you how far that estimate could be from your true win rate. Here it is:

√(0.55 × 0.45 ÷ 40) ≈ 0.079

A rough 95% range for your true win rate is 55% ± 16%. That's anywhere from about 39% to 71%. Forty trades can't tell a coin-flip strategy from a good one.

At 200 trades the range shrinks to about ±7%, which starts to mean something. Be wary of any conclusion drawn from a single month.

One kind of market. A month of paper trading might all fall in a strong trend, or all in a range. A strategy that only works in one will look brilliant or useless depending on the month. It tells you nothing about the other.

No real fear. Virtual money doesn't hurt. Even with real prices and real charges, a simulator can't make you feel a loss the way your own money does.

Fills that are still a bit too good. A simulator fills at the last traded price. A real order also pays the bid-ask spread, the gap between the best buy and sell prices. In a thin strike that can be several rupees. We covered this and four other gaps in paper trading vs real trading.

Should you backtest first or paper trade first?

They answer different questions, so use them one after the other.

Backtest to reject. Test the idea on as much history as you can get, with realistic charges. If it fails, drop it. If it passes, be suspicious. The more versions you tried before one passed, the more suspicious you should be.
Paper trade what survives. Follow the exact rules, forward, with real prices and real charges. Keep going until you have a decent sample; 100 trades is a sensible minimum. Then compare your paper results with what the backtest predicted for the same period. If they disagree badly, one of them is wrong, and it's usually the backtest.
Go live small. Trade one lot, the smallest size you can. Keep at it long enough to see if real fills and your real nerves match the paper record. Only then size up.

Does thepapertrade do backtesting?

No, and that's on purpose. This site is a paper trading simulator, not a backtester. There's no way to replay history and no strategy builder. Everything happens forward, in the live market, at the price trading when you place the order. That makes it step two above, not step one.

What you get for that step is a record you can trust. Every trade is stored with its real charges. Your stats (win rate, expectancy, profit factor, average holding time and maximum drawdown) are all net of charges. Expectancy is your average profit or loss per trade, and drawdown is the fall from a peak to a low.

The trade history page works these out for any date range. Set it to match your backtest's test period and you have the comparison for step two.

Quick answers

What is the difference between paper trading and backtesting?

A backtest tests your rules on past prices. Paper trading tests them on live prices, going forward, with virtual money and you placing the trades.

Which is better, paper trading or backtesting?

Neither on its own. Backtest first to throw out bad ideas, then paper trade the ones that survive.

How many paper trades do I need before my win rate means anything?

Forty trades leaves a range of about ±16%, which is too wide to tell. At 200 trades it's about ±7%. Aim for at least 100.

Can I backtest on thepapertrade?

No. The site only does forward paper trading with live prices. It doesn't replay history.

None of this is investment advice, and neither a backtest nor a paper record can promise what a real account will do.

Try it on a simulator first.

₹1,00,000, real NSE prices, and every charge in this post applied to every trade.

Start with ₹1,00,000