Quick answer: A clean paper run proves four things. Your entry logic fires on time. Your leg construction is correct. Your broker API round-trip works. Your alerting reaches you. It cannot prove profitability. It hides your real fill price, your real stop-loss slippage, and your real peak margin. Treat a green paper P&L as a plumbing check. Do not treat it as permission to go live at full size.

Every option-selling desk that has gone live has seen the same pattern. Three weeks of paper trading look clean. The trader switches on real capital. The live curve diverges from the paper curve inside the first ten sessions. Our team at EliteAlgo has run this comparison many times, across NIFTY and SENSEX expiry-day short strategies. We stopped treating the gap as bad luck. We now treat it as a structural property of paper trading itself. This piece is the honest version nobody selling a paper-trading feature wants to write, because it tells you to distrust the very demo they are showing you.

What Paper Trading Actually Proves

Before the criticism, credit where it is due. A paper run validates four separate things. Skipping this stage is a mistake in the other direction — going live untested is worse than over-trusting a paper result.

1. Signal and entry logic fires at the intended time

Say your strategy sells a strangle at 9:16 IST on expiry day. A paper run tells you whether your code wakes up on time. It checks whether it reads the option chain correctly. It checks whether it places that order at 9:16, not at 9:19 because of a slow API call. It also checks that strike selection does not silently fail. This is a timing and reliability test. Run it for at least one full expiry cycle before you trust it.

2. Leg construction and strike selection are correct

Paper trading catches the boring bugs that are also the most expensive live. Selling the wrong strike because your ATM calculation rounded the wrong way is one example. Selling calls when you meant puts is another. Sizing two legs unevenly because a lot-size variable did not update after a contract specification change is a third. We have caught leg-construction bugs in paper runs that would have cost real money on day one live.

3. The order pipeline and broker API round-trip work

This covers authentication tokens that expire mid-session. It covers rate limits that silently drop an order. It covers the retry logic meant to handle a rejected order. A paper run against a live-adjacent sandbox, or against your own logging layer in front of the broker API, proves the pipe carries data end to end.

4. Monitoring and alerting actually reach you

Say your kill switch texts you when the day's loss crosses a threshold. A paper run is where you find out the SMS gateway you configured eighteen months ago quietly stopped working. Test the alert path. Do not just test the strategy path.

Those four checks are real. They are worth running properly. What they are not is a profitability test. This is where most option sellers get the tool's purpose backwards.

The Three Things Paper Trading Hides

Here is our test setup for comparing the two paths. We run the identical strategy, same strikes, same session, once through a paper simulator and once through a live account with real fills. Then we line up the day's P&L side by side. Across a sample of expiry sessions, the gap never closed to zero. It was never in the paper run's favour.

1. Fills — paper fills at LTP or mid, live fills cross the spread

Most paper simulators, including the free paper-trading modes on Tradetron and AlgoTest, fill your order at the last traded price or the mid of the bid-ask spread. A real market sell order crosses the spread and fills at the bid. On a NIFTY weekly option with a ₹2–4 wide spread near the money, that is a real cost on every leg, on every trade, every day. It is not random noise that averages out. It is a one-directional tax a paper account never pays.

2. Slippage on stop-loss exits

A paper stop-loss triggers at your trigger price and fills there. A live stop-loss triggers on a fast-moving underlying. By the time your order reaches the exchange, the option has already re-priced. That is often 2 to 5 points against you on an index option, and more on an illiquid strike during an expiry-day spike. Slippage on entries is a cost. Slippage on stop-loss exits is a cost exactly when you are already losing. It compounds the damage on your worst days, not your best ones.

3. Margin and peak-margin reality

A paper account never asks you for margin. A live account does. Under SEBI's peak margin framework, your broker checks margin utilisation at multiple intraday snapshots, not just at order placement. Picture a strategy that looks comfortably capitalised at 9:20. It can breach peak margin at 9:47 if the underlying gaps and your short strikes move against you before you adjust. Paper trading gives zero visibility into this, because there is no real margin ledger behind it.

Why the Optimism Runs in One Direction

This is the part most explanations skip. The paper-versus-live gap for an option seller is not symmetric noise. It is a specific, measurable bias. On every leg of every trade, a paper fill is at least as good as the real market order would have achieved, usually better. You are never paper-filled worse than you would be filled live. Stack that across four legs of a short strangle, across every session in a month, and a strategy showing a modest paper profit can turn flat or negative once real fills replace the simulated ones.

The arithmetic is simple, and it is worth doing before you trust any paper number. Take your per-leg slippage estimate in points. Multiply by the number of legs. Multiply by lots. Multiply by trading days in the month. A conservative 2–3 point per-leg haircut on a four-leg strangle, traded at 3 lots for 20 sessions, is not a rounding error. On NIFTY's 75-unit lot size, that is 2 points × 4 legs × 3 lots × 75 units × 20 days. That works out to well over ₹36,000 of pure slippage cost a paper account never charged you.

A Worked Example: One Expiry Day, Two Fills

Take a single short strangle on a NIFTY expiry session: sell one call and one put, 3 lots each, at a combined premium the simulator prices at ₹180 per lot. In the paper engine, both legs fill at the quoted mid-price the instant the order is placed, so the logged credit is ₹180 × 75 × 3 × 2 legs, before costs. In the live account on the same day, the order crosses a ₹3 spread on each leg because the book is thin two minutes after the open. The realised credit is closer to ₹174 per lot. That four-rupee gap, multiplied across 450 units per leg and two legs, is roughly ₹3,600 of lost credit on one trade — money the paper log never subtracted. Run that same session through a stop-loss exit and the gap widens further, because the exit price is the one that moves fastest when the market does.

The 9:15 Trap

One entry-timing detail deserves its own section. It quietly poisons more paper backtests than any other single factor: the opening-tick entry. A strategy coded to enter "at 9:15" fills cleanly in almost every simulator. That is because 9:15:00 is simply the first candle timestamp in the data feed. In the real market, 9:15 is the most illiquid, most volatile 60 seconds of the session. Spreads are widest. The order book is still forming. By the time your order round-trips to the exchange, the price you saw is stale. Our desk does not place entries at 9:15 for this exact reason. Every strategy we run enters at 9:16 or later, once the opening auction settles and a real two-sided book exists to trade against. A paper run that "proves" a 9:15 entry works is proving something the live market cannot actually deliver.

Paper Trading vs Backtesting vs Forward Testing

These three terms get used interchangeably. They answer three different questions.

  • Backtesting answers: how would this strategy have performed on years of historical data, at realistic cost? It is the only one of the three that gives a statistically meaningful sample size fast. According to our own methodology, every backtest number we publish is re-run under realistic slippage assumptions and full India cost first. That means brokerage, exchange transaction charges, STT on the sell leg, GST, stamp duty and SEBI's turnover fee are all modelled in, because an unadjusted backtest suffers the same optimism bias as paper trading.
  • Paper trading answers: does my code, my connectivity and my alerting actually work in real time? It is an operational test, not a performance test.
  • Forward testing — trading the strategy live at minimum size with real fills — answers the question that actually matters: does this survive contact with a real order book? Nothing else can substitute for this stage, which is exactly why skipping straight from paper to full size is the mistake.

A Paper-Run Protocol Worth Running

If paper trading is a plumbing check, run it like one. Set a defined pass condition, not an open-ended "trade it until it feels ready."

  1. Minimum session count: at least 15–20 trading sessions. That is enough to cover a range of intraday conditions, not just one quiet week.
  2. At least one expiry day: expiry-day option selling has its own volatility and gamma profile. A paper run that skips expiry day has not tested the sessions that matter most.
  3. At least one gap-open session: a day where the market opens well away from the previous close tests whether your strike-selection and margin logic hold up under stress.
  4. A deliberate manual kill-switch drill: on one session, manually trigger the loss-limit condition. Confirm the system actually flattens positions and alerts you. Do not wait for a real bad day to find out the kill switch was never wired correctly.

How Margin Snapshots Actually Work Intraday

This is the piece most traders skip until it costs them. SEBI's peak margin rules mean your broker is not checking your margin once, at 9:15. It is sampling your positions at multiple random points across the session and comparing the highest requirement seen against what you actually held. If your short strangle needed ₹85,000 in margin at the open and the underlying moved sharply by 11:30, the peak-margin snapshot can price the same position at ₹1,10,000 or more for a few minutes. A live account that is capitalised for the open-of-day number, not the intraday peak, can get flagged for a shortfall penalty even on a day that closes profitably. A paper account has no concept of this, because there is no real margin ledger sitting behind it. This is one more reason a clean paper month tells you nothing about whether your capital allocation is actually sufficient.

Building Your Own Paper-vs-Live Slippage Log

Most traders never measure their own slippage because there is no default report that shows it. Building one is simple and worth the half hour it takes. Log four numbers for every trade: the price the simulator or your signal expected, the price you were actually filled at, the timestamp of the order, and the timestamp of the fill. Over 20 sessions, this gives you a real distribution of your own slippage, in your own strikes, on your own broker connection, rather than a generic assumption borrowed from an article.

Once you have that log, three numbers matter more than the rest. The average slippage per leg tells you the baseline cost to fold into every sizing decision. The worst-case slippage, not the average, tells you what your stop-loss exits actually cost on a bad day — this is the number that should size your stop distance, not the average. And the gap between your expected entry time and your actual fill time tells you whether the 9:16 rule, or whatever entry time your strategy uses, is being honoured by your own infrastructure or quietly drifting later as load increases. A slippage log that sits untouched after the first week defeats the purpose; review it after every five sessions and update your cost assumptions, not just once at the end of the paper run.

The Graduation Checklist Before Live Capital

Before moving any strategy from paper to live, our team runs a fixed checklist rather than relying on a feeling of readiness:

  • Full India cost structure modelled into every backtested and paper number — brokerage, exchange transaction charges, STT, GST, stamp duty, SEBI turnover fee.
  • Slippage modelled explicitly, per leg, using the arithmetic above rather than assumed away.
  • Margin headroom verified against a peak-margin scenario, not just the margin required at order placement.
  • Kill switch tested with a manual drill, with confirmed alert delivery.
  • Position size cut for the first live month, typically to a fraction of the paper-tested size, so the first real-fill data you collect does not cost you the same as a full-size mistake.

Paper trade every strategy, strike rule, and entry-time change first — this checklist assumes you have, and none of it is investment advice. EliteAlgo builds and researches analytical trading software; nothing here is a recommendation to trade any specific instrument or size.

What This Means for Your Next Paper Run

Run the paper phase for what it is good at: timing, leg logic, connectivity, alerting. Do not ask it the one question it cannot answer. Price in slippage and real margin before you size the live version, using the arithmetic in this piece rather than the paper account's number. A strategy that survives that repricing is worth forward testing at small size. One that only looked good with mid-price fills was never really tested at all.

Frequently Asked Questions

How long should you paper trade an option-selling algo before going live?

Run it for at least 15–20 trading sessions that include one full expiry cycle and one gap-open day. Do not pick a fixed calendar duration instead. A quiet three-week stretch with no expiry day and no gap-open session tells you far less than two weeks that happen to include both.

Is free paper trading on platforms like Tradetron or AlgoTest enough?

It is enough to validate signal timing, leg construction and connectivity — the operational checks this piece covers. It is not enough to validate profitability. Free paper-trading modes on these platforms fill at LTP or mid-price, which structurally hides fill slippage and stop-loss slippage.

Why do live results differ from paper trading results?

Live fills cross the bid-ask spread instead of filling at mid or last traded price. Stop-loss exits slip against you during fast moves. Live trading also carries real intraday margin requirements that a paper account never enforces. All three push in the same direction: live results run worse than paper, not randomly different.

Can you paper trade an expiry-day strategy realistically?

You can test the mechanics realistically — order timing, strike selection, kill-switch behaviour. You cannot paper-test the fill quality on an expiry day. Expiry-day spreads widen sharply into the close, and a mid-price fill understates the real cost by more than it does on a normal session.

Does paper trading help with trading psychology?

Only partially. It builds familiarity with your own process and interface. Because no real money is at risk, it does not reproduce the decision-making pressure of watching a live loss cross your threshold. Psychology preparation happens properly only once you forward test at small real size.

What is the difference between paper trading and a backtest?

A backtest runs your strategy against years of historical data in seconds. That gives a large statistical sample, at the cost of being entirely retrospective. Paper trading runs your live code against today's real-time market feed, testing operational reliability rather than statistical performance. Use a backtest to judge whether an idea is worth pursuing. Use paper trading to judge whether your implementation of it actually works.

How much slippage should I assume for an index option strangle?

A conservative starting assumption for a NIFTY or SENSEX index option near the money is 2–3 points per leg on entry, with stop-loss exits running higher during fast moves. Multiply that by your leg count, lot size and trading days to see the monthly cost before you trust a paper P&L.

Should you paper trade the same strategy across multiple expiries before changing anything?

Yes. One expiry day is one data point, and expiry-day behaviour varies with volatility regime. Our own data set for expiry-day short strategies only starts to mean something after three to four expiries, not one. Changing a rule after a single session, good or bad, usually reacts to noise rather than a real signal.

Read the buyer-side view of this same problem in our complete backtesting software checklist, which covers what to verify before you trust any tool's numbers, paper trading included. If you are building the execution side yourself, our walkthrough on automating an option-selling strategy on Zerodha's Kite API covers the same order-pipeline and alerting checks from the code side. Before sizing anything for live capital, it is worth reading where automation itself can go wrong — see our honest list of the disadvantages of algo trading. Traders who would rather have this graduation checklist applied for them can look at our managed algo trading service.

Cost figures referenced above are consistent with the transaction charge and turnover fee schedules published by SEBI, and with the securities transaction tax framework documented by the Income Tax Department. Model both into any paper-to-live comparison rather than relying on a simulator's built-in cost assumptions.