The name comes from the days when traders wrote imaginary trades on paper and tracked them against the next day's prices. Today the same idea runs on software: a simulated account, live prices, and a record of every trade the rules would have made. In QuantPower the feature is called Forward test. It runs the same algo engine as live trading on a separate account with test money, so the step from simulation to real money changes where the orders go, not the rules. This guide uses "paper trading" for the general practice and "Forward test" for QuantPower's feature, and covers both with worked figures, a staged plan for going live and the mistakes that make paper results misleading.
The two side by side
| Paper trading | Live trading | |
|---|---|---|
| Prices | Live, from a market feed | Live |
| Orders | Simulated by the platform | Sent to the exchange through your broker |
| Fills | Assumed at the price seen, or with a modelled spread | At the price the market actually offers |
| Queue position | Ignored: a limit fills when touched | Real: orders ahead of yours fill first |
| Margin | Checked notionally, or not at all | Blocked in your account; orders refused if short |
| Rejections | Rare or none | Real: margin, price bands, freeze limits, bans, sessions |
| Costs | Usually modelled | Charged |
| Capital at risk | None | Real |
| Emotion | Low | High |
| What it tests | Rules, signals, platform setup | Everything, including you |
A short history of paper trading
Practice trading is as old as markets. Brokers' trainees and students have long kept notebooks of hypothetical trades, and trading competitions, in which players start with play money and the best simulated portfolio wins, have been run by newspapers, universities and brokers for decades. Software turned the notebook into a simulator: by the 2000s, many trading platforms offered a "paper" or "demo" account that took live prices and filled simulated orders automatically.1
Systematic traders gave it a more precise job. A backtest tells you how rules would have behaved on past data; it cannot tell you whether the rules were fitted to that data. Running the same rules forward, on prices that arrive after the test starts, can. Robert Pardo's work on walk-forward analysis in the 1990s set out why testing on unseen data matters, and running a strategy forward on live prices is the purest version of that idea.2
In India, the practice took a regulatory turn in 2024. On 24 May 2024, SEBI issued norms on sharing real-time price data with third parties, which stopped real-time prices from being used in virtual trading games and fantasy apps.3 A simulation on real-time prices now needs a proper data source, usually your own broker connection. And since 1 April 2026, retail algos in India run through brokers under the exchange's rules, with each strategy registered and tagged.4 Paper trading is where a strategy should prove itself before it goes through that path with real orders.
What paper trading proves
- The rules are configured correctly. The strategy fires when it should, at the strike and quantity you intended, with the stop and target attached.
- The signals work on live data. Backtests use clean historical candles; a live feed has delays, gaps and noise.
- The platform and connection work. Logins, sessions and data subscriptions behave as expected.
- The strategy's recent behaviour matches the backtest. A few weeks of simulated results should sit within the range the backtest's good and bad weeks suggest.
- You understand what the strategy does. Watching it run, entry by entry, often reveals a rule that does something you did not intend.
What paper trading cannot prove
- Fills. A simulated limit order fills when the price touches it; a real one may sit in a queue. A simulated stop fills at the trigger; a real one fills at the next available price. In thin option strikes, the difference can be several rupees per unit.
- Market impact. A simulated order does not move the market. A real order for many lots in a less liquid contract does.
- Margin reality. A simulation may not block margin the way a broker does, so a position that runs on paper may be refused live for insufficient margin, especially on expiry day.
- Rejections. Freeze quantities, banned securities, price bands and the broker's own restrictions appear only live.
- Your behaviour. Watching a simulated loss is nothing like watching a real one. Most of the damage traders do to a working strategy happens live, when they override it.
Why results differ
Expect live results to be somewhat worse than simulated results for the same strategy over the same period. The difference comes from:
- Slippage on entries and exits, the largest factor for intraday and options strategies.
- Costs if the simulation did not model them fully.
- Partial fills and orders that never filled.
- Latency between the signal and the order reaching the exchange.
- Manual interference, when the trader pauses, resizes or exits by hand.
If live results are much worse than simulated ones, check the fills first: compare each live fill with the price the simulation assumed. If the fills match and results still differ, the trader is usually the variable.
A worked comparison
A trader runs a Nifty short straddle, one lot of 65, in simulation for four weeks and then live for four weeks with the same rules. Over each period the strategy trades 16 times, entering two legs and exiting two legs each time.
| Item | Simulated | Live |
|---|---|---|
| Gross result before costs | ₹18,200 | ₹18,200 if fills matched |
| Slippage: ₹0.15 per unit on each of 64 orders | ₹0 | 0.15 × 65 × 64 = ₹624 |
| Brokerage: 64 orders at ₹20 | Modelled: ₹1,280 | ₹1,280 |
| Statutory charges (STT, exchange, SEBI, stamp duty, GST) | Modelled | Charged |
| One stop that filled ₹3 worse in a fast move | ₹0 | 3 × 65 = ₹195 |
| One manual exit, 20 minutes early, on a nervous afternoon | ₹0 | Gave up ₹1,100 of the day's decay |
The live month ends about ₹1,900 behind the simulated month before statutory charges, on identical rules. Almost half of the gap is the trader's own hand. This pattern, a modest slippage cost and a larger behavioural cost, is the usual one.
How long to paper trade
Two to four weeks for a strategy that trades daily. Enough to see the strategy handle an expiry day, a trending day and a gap, and to confirm that the platform ran it without intervention. For a strategy that trades a few times a month, count trades rather than weeks: twenty or more entries give a fair picture of its mechanics. For a positional strategy held for days, include at least one weekend and one expiry roll in the simulation.
Longer paper trading does not prove the edge; that is the backtest's job, over years of data. Paper trading checks the machinery and the fit between the backtest and the live market. Our guide to backtesting vs forward testing explains the division of work.
A four-week simulation plan
A simulation is most useful when it has a plan. This is one for a strategy that trades on most days.
| Week | Focus | What to record |
|---|---|---|
| 1 | Mechanics | Did every entry happen at the time and strike the rules say? Did every stop and target attach? Did anything need your hand? |
| 2 | Fit with the backtest | Are the week's trades the ones the backtest would have taken on the same days? If you re-run the backtest over this week, do the entries match? |
| 3 | Bad days | Did the strategy handle a gap, a fast trend or an expiry day as designed? How large was the worst day against the backtest's worst days? |
| 4 | Costs and fills | Compare each simulated fill with the bid and ask at that moment. Add a realistic slippage figure to the totals and see what is left |
At the end of week 4, write one paragraph: what worked, what surprised you, what you changed and why. If you changed a rule during the four weeks, start the count again from the change. A simulation of a rule that kept changing is not a simulation of any one rule.
Reading a simulated record
A month of simulated trades is a small sample, so the totals mean less than the details. Look at these, in this order:
- Each entry against the rules. Open five trades at random and check them line by line: the time, the strike, the quantity, the stop.
- The exits. Count how many trades closed on a stop, a target, the time exit or a manual action. A strategy designed to exit mostly at the target that exits mostly on stops is not behaving as designed, whatever the total says.
- The worst trade. Was it within the size the rules allow? If a single trade lost far more than the stop implies, find out why before any real money is involved.
- The days with no trade. Were they days the rules should have skipped? A strategy that quietly failed to trade on good days can still show a profit on the days it did trade.
- The totals, last. Compare them with the backtest's range for a month, not with its average. A month near the bottom of the range is not a failure; a month far outside it is a warning.
Paper trading an indicator strategy
Indicator strategies need one extra check in simulation: that the signal fires on the candle you expect. Take a rule that buys Nifty futures when the 15-minute close crosses above the 20-period EMA. In a backtest, the signal appears on the candle that closes above the average. Live, the signal can only be known when that candle closes, at 10:15, say, and the order goes in after that. A simulation on live prices shows whether your platform reads the condition at the close of the candle, as it should, or reacts to the candle while it is still forming, which repaints and produces signals that disappear.
Check three entries by hand against the chart: the time the candle closed, the time the signal fired and the price at which the simulated order filled. If they line up, the rule is being read correctly. If the signal fired before the candle closed, the rule is using information it would not have had, and the backtest is overstating it.
When to stop and go back to the backtest
Paper trading should end in one of two ways: the strategy goes live small, or it goes back to the drawing board. The signs that it should go back:
- The simulated month falls well outside the range of the backtest's months.
- Trades happen that the backtest would not have taken on the same days.
- The rules needed changing during the simulation to make them work.
- Costs and realistic slippage turn a small simulated profit into a loss.
- You found yourself wanting to override the strategy on most days.
None of these is a disaster. Each is cheap information, bought with test money rather than real money. That is what the simulation is for.
Questions to answer before the first live order
- What is the most this strategy can lose in a day at the size I will start with, and can I accept it without changing my plan?
- Where will I see the live positions, and how do I close them by hand if I must?
- What account limits sit above the strategy, and have I set them?
- Is my broker signed in for today, and does the session last until the strategy's exit time?
- What will I do after the first losing week? (The answer should be: nothing until the monthly review.)
Going live: a staged approach
- Confirm the simulated results match the backtest's expectations for the same weeks.
- Check live margin for the intended size with an F&O margin calculator, including expiry-day additions.
- Start with one lot. The purpose is to measure live fills and your own reactions, not to earn.
- Set account limits before the first live order: a daily loss limit, and a stop for the whole account at an amount you can lose without changing the plan.
- Compare fills for the first twenty trades with the simulation's assumptions, and update the strategy's slippage estimate.
- Scale in steps, doubling size only after a full month that matches expectations.
- Keep a [trade journal](/trade-journal) of every override, with the reason. Overrides are the most common cause of live underperformance.
Size in simulation should match size live
A common trap is paper trading at a size you would never use with real money. Fifty lots in simulation teaches nothing about one lot live, because the fills, the margin and the emotions are all different. Simulate at the size you will actually start with. If you plan to scale later, scale the simulation in the same steps.
The behaviour gap
The same rule feels different with real money, and the difference shows up in a few predictable ways:
- Cutting winners early. A short straddle that has earned half its premium by noon is easy to hold in simulation. With real money, the urge to "lock it in" at 1:30 pm is strong, and the afternoon's time decay is lost.
- Moving stops. A stop that is about to be hit gets widened "just this once". In simulation nobody bothers; live, it is the most common override.
- Skipping entries after a loss. After two losing days, the third entry is skipped. That third day is often the one that would have recovered the first two.
- Adding size after a win. A good week live invites a larger next week, before the strategy has earned it.
None of these shows up in a simulation, which is why the first live weeks are a test of the trader as much as of the strategy. Two habits help: an automated strategy that you do not touch during the day, and a written rule that changes are made only at a scheduled review.
Running simulation and live side by side
Once a strategy is live, keep a copy running in simulation with identical rules. The two records then measure two different things:
| Comparison | What a gap tells you |
|---|---|
| Live fills against simulated fills on the same entries | The real cost of the spread and slippage for this strategy |
| Live exits against simulated exits | Whether you or the platform intervened |
| Live days traded against simulated days traded | Whether rejections, margin or sessions stopped live entries |
| Live total against simulated total | The sum of all of the above, over the same weeks |
Review the comparison monthly. A small, steady gap is the price of execution and can be built into the strategy's expectations. A growing gap means something has changed, such as a less liquid strike, a broker restriction or a habit of overriding, and is worth finding.
A checklist for a fair simulation
- The lot size and contract rules are current.
- Costs are included, and slippage is set to a realistic figure for the contracts traded.
- The size is the size you will start with live.
- The rules are the final rules, not a draft that keeps changing.
- The simulation runs on live prices from a proper source.
- It covers at least one expiry day and one volatile session.
- You recorded every time you would have overridden the strategy, even though you could not.
Paper trading options: what to watch
Options are where simulation and reality part most often:
- Wide spreads in far strikes. A simulation that fills at the last traded price can be ₹1 or more away from a real fill in a far strike.
- Expiry-day speed. Near-expiry options can move several rupees in seconds. A stop that fills at its trigger in simulation may fill well past it live.
- Margin on expiry day. Short index options attract an extra margin on expiry day; a position that fitted the day before may not fit live.
- Physical settlement. Stock options that expire in the money settle by delivery of shares. A simulation may simply cash-settle them; your broker will not.
- Lot size changes. Lot sizes were revised several times in 2024 and 2025. A simulation set up with an old lot size gives wrong rupee figures.
Paper trading for manual traders
Simulation is not only for algorithms. A trader who places orders by hand can use a simulated account to learn a new instrument, rehearse an order type or try a structure such as a spread before paying for the lesson. The same limits apply: fills are kinder than real ones and the emotions are muted. The useful habit is to write down, before each simulated trade, the entry, the stop and the reason, and to review them at the end of the week. A manual trader who cannot follow a written plan with test money will not follow it with real money either.
Manual simulation is also a good way to learn a platform's screens: where the order sheet asks for the product type, how a stop-loss order differs from a stop-limit, what the margin preview shows and how a position appears afterwards. Mistakes made learning the screens are free in simulation and expensive live.
Common mistakes in both modes
- Paper trading for months. After a few weeks, further simulated results add little. Either the strategy goes live small or it goes back to the backtest.
- Treating simulated profits as a forecast. They are an upper bound.
- Going live at full size. The first weeks live are a measurement, and measurements should be cheap.
- Changing the rules after a bad live week. Decide the review schedule in advance, such as monthly, and change rules only then.
- Comparing different periods. Compare simulated and live results for the same weeks, or the comparison measures the market rather than the strategy.
- Ignoring the simulation once live. Keep a copy running in simulation after going live; it is the cleanest benchmark for your live fills and your overrides.
Forward test in QuantPower
QuantPower's paper trading feature is called Forward test, and it comes in three forms:5
| Form | What it does |
|---|---|
| An algo in Forward test | The same algo engine as live: every entry, exit, stop and target, on your Forward-test account. Switch an algo on in Forward test instead of live |
| Manual trades | Your own Forward-test account, which behaves like a broker's book, with orders, positions and M2M, and ₹10,00,000 of test money to start |
| A rule forward test | A rule on one instrument read candle by candle on live prices, with the trades and an equity curve |
How it behaves:
- Prices. With your own broker connected, prices are live. Without one, the account sees the last close, not real-time prices, in line with SEBI's 2024 norms on real-time data.
- Fills. Orders fill at the price the account sees. That makes Forward test a fair check of the rules and an optimistic check of the fills, exactly as described above.
- Separation. Forward-test positions never mix with real money. Your real portfolio, funds and the account's risk rules read only your real brokers. "Reset" clears the Forward-test account and stops the algos running in Forward test; live algos are not touched.
- Same rules, different destination. Moving an algo from Forward test to live changes where its orders go. You switch it off and switch it on live; the legs, entries and exits are the same. A restart never moves an algo from test money to real money on its own.
- Records. Each algo has its own live performance report beside its backtest, and Forward-test and live records are kept apart.
What it costs
An algo in Forward test costs 100 tokens per 28 days; live, 200 tokens per 28 days (1 token = ₹1, GST extra). An algo already paid for live can run in Forward test in the same period at no extra cost, and an algo paid for in Forward test that goes live pays only the difference. Switching on part-way through a cycle is charged for the days left only. Manual trading on your Forward-test account costs no tokens.
Where paper trading is going
Two forces are changing it. Regulation now requires real-time simulations to use a proper data source, which favours platforms that connect to your own broker. And as strategies are increasingly written with AI help, the forward test becomes the main check that a strategy does what its author thinks it does, before real orders follow. The habit worth building is simple: no strategy goes live until it has run forward, on prices that did not exist when it was written. Simulations are also becoming more honest about costs: modelling spreads, charges and margin closer to the real thing narrows the gap between simulated and live results, though it can never remove the part that comes from the trader. The better the simulation, the more clearly what is left over points at behaviour.
Further reading in QuantPower
The forward test page covers the three forms in detail, and the live trading terminal shows real positions, funds and orders once a strategy goes live.
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
Is paper trading realistic?
For signals and rules, yes. For fills, margin and your own behaviour, no. Treat simulated results as the best case.
How long should I paper trade before going live?
Two to four weeks for a daily strategy, including at least one expiry day and one volatile session, or about twenty entries for a strategy that trades less often.
Why is my live P&L worse than my paper P&L?
Slippage, costs, partial fills and latency account for most of the gap. Compare the fills trade by trade to see how much, and check whether any trades were changed by hand.
Does Forward test use real market prices?
With your own broker connected, yes: prices are live and only the orders are simulated. Without a broker connection, Forward test uses the last close.
Can I run a strategy in Forward test and live at the same time?
Yes. Many traders keep a copy of a live strategy in Forward test as a benchmark, and test new strategies there while established ones trade live.
Does a Forward-test loss affect my real account limits?
No. The daily loss limit, Smart Exit and the kill switch read only your real brokers. Forward-test positions are kept on their own account.
What should I do if my simulated results are much better than my backtest?
Be suspicious before you are pleased. A few good weeks can be luck, and a simulation that fills every order at the price seen will flatter any strategy that trades often. Check the fills against the bid and ask, add realistic slippage, and compare the same weeks in a backtest. If the advantage survives all three, it may be real; if not, the backtest was the fairer picture.
Related guides
- Paper trading meaning: the term in plain words
- Paper trading platform for options: practising options strategies in India
- Live trading terminal: your real positions, funds and orders
References
- Investopedia, "Paper Trade: Definition, How It Works, and Benefits" (investopedia.com)
- Robert Pardo, *Design, Testing and Optimization of Trading Systems* (1992), on walk-forward analysis
- SEBI, norms on sharing real-time price data with third parties, 24 May 2024 (sebi.gov.in)
- SEBI, circular on safer participation of retail investors in algorithmic trading, 4 February 2025, in force for every broker from 1 April 2026 (sebi.gov.in)
- QuantPower: Forward test, the Forward-test account and token rules, from the app's code and price list, checked 12 November 2026