Paper trading runs your strategy against live market data, but every order is simulated. Nothing reaches a broker and no money is at risk. It is the step between a backtest and real capital, and it answers a question neither of the others can.
What it actually proves
A backtest tells you how a strategy would have behaved on history. Paper trading tells you how it behaves on today — on prices that are moving now, with the same engine that would place real orders.
| Backtest | Paper trading | Live trading | |
|---|---|---|---|
| Data | Historical | Live | Live |
| Orders | Simulated | Simulated | Real |
| Money at risk | None | None | Yes |
| Needs a broker | No | No | Yes |
| Needs a dedicated IP | No | No | Yes |
| Shows slippage honestly | No | Partly | Yes |
| Time to a result | Seconds | Days to weeks | Ongoing |
| Answers | Was the idea sound historically? | Does it behave as expected now? | Does it survive real fills? |
The row that matters most is the second-to-last. A backtest gives you an answer in seconds, which is exactly why it is so easy to over-trust. Paper trading is slow, and that slowness is the point — it makes you watch the strategy through conditions you cannot fast-forward.
Where paper trading sits in the sequence
Each stage of testing answers one question that the stage before it could not. Skipping a stage does not remove the question, it just defers it to somewhere more expensive.
| Stage | The question it answers | Cost of finding out here |
|---|---|---|
| Backtest | Has this idea ever worked? | One credit |
| Fragility check | Is the result robust or fitted? | One or two credits |
| Paper trading | Does the logic behave correctly on live prices? | Time only |
| Live, minimum size | What does execution actually cost me? | Small, and worth paying |
| Live, full size | Does the edge survive at size? | Real money |
The pattern is that cost rises at every step while the question gets more specific. The reason to be disciplined about the order is simply arithmetic: finding a logic error at stage three costs you nothing, and finding the same error at stage five costs you whatever it traded.
What it cannot prove
This is where paper trading is routinely oversold, so it is worth being blunt about the limits.
- Fills are assumed, not earned. A simulated order does not queue behind real ones. Real fills on an illiquid strike will be worse.
- No market impact. Your simulated size never moves the book. Real size sometimes does.
- No margin reality. A paper strategy never gets rejected for insufficient margin, so it will not warn you that the position is unaffordable.
- No emotional load. Watching a simulated drawdown is not the same experience as watching a real one, and that difference changes behaviour.
- No broker failures. Rejections, session expiry and IP problems do not exist in paper mode, so none of them get tested.
What to use it for
Used properly, paper trading is very good at catching a specific class of problem:
- Logic errors. Entry fires at the wrong time, a leg is the wrong side, the exit never triggers. These show up on day one and are embarrassing to discover with money on.
- Parameter sanity. A stop loss that is too tight gets hit constantly, and you see that within a week.
- Behaviour in conditions your backtest under-sampled. A gap open, a volatility spike, an expiry day.
- Your own readiness. Whether you can actually run the morning routine every day before you have money depending on it.
The one nobody expects to matter
That fourth one is underrated. Live algo trading has a daily operational routine — broker login, start the bots, check the logs. Paper trading is where you find out whether that routine fits your actual mornings, at zero cost for discovering it does not.
How paper fills are decided
A fair question, since the realism of the whole exercise depends on it: where does a simulated fill price come from?
Paper orders are priced from the live market at the moment the strategy acts. That is what makes paper trading more informative than a backtest — the prices are genuinely current, moving, and reflect whatever is happening in the market right then.
What is real and what is assumed
What is simulated is the counterparty. Your order does not join the book, so it does not wait, does not partially fill, and does not move the price. On a liquid at-the-money strike that is a small assumption. On a thin strike it is a large one.
How long to paper trade
Long enough to see the conditions the strategy is supposed to handle — which for an index options strategy usually means covering at least a couple of expiry cycles, not three quiet days.
A better rule than a day count
A fixed number of days is the wrong target. The right target is a number of situations:
- At least one trending day
- At least one range-bound day
- At least one expiry day
- At least one gap open
- At least one day where the strategy correctly did nothing
That last item is deliberate. A strategy that trades every single day is either very active by design or has an entry condition that is not actually filtering anything. Seeing it decline to trade, and agreeing with the decision when you read the logs, is a real checkpoint.
Paper trading on Arthalab
1
Build or pick a strategy
Your own from the Option Strategy Builder, or a Research Analyst strategy.
2
Deploy it in Paper mode
Paper mode is selected at deploy time. No broker connection is required.
3
Start it
Deploying is not starting. A deployed bot sits idle until you start it.
4
Watch the logs, not just the P&L
The logs show every decision, including the ones where it correctly chose not to trade.
Paper trading needs an active plan but no broker, no IP and no capital — which makes it the cheapest way to find out whether a strategy is worth the rest of the setup. When to switch to real money covers the next step.
What a useful paper period looks like
A worked version, because "paper trade for a while" is advice nobody can act on precisely.
Deploy the strategy on paper and start it every trading day for four weeks. That covers roughly four weekly expiries. Each day, after the exit time, open the logs and read what the strategy did — not the P&L, the decisions.
What you should know by the end
By the end you should be able to answer all of these without looking:
- What the strategy does on a quiet day.
- What it does on a day the index gaps open.
- How often the per-leg stop loss fires, and whether that feels too often.
- Whether the exit time ever closes a position you would rather have held.
- What the worst day looked like, in points and in rupees at your intended size.
If you cannot answer those after four weeks, you have been watching the P&L rather than the strategy, and another four weeks of the same will not help.
Paper trading an RA strategy
Research Analyst strategies are included in the plan, and paper trading one is the sensible way to form your own view before committing capital to someone else's rules.
The published backtest tells you what the analyst measured. A month on paper tells you how it behaves now, how often it trades, and whether its drawdown pattern is one you would actually sit through. Those are different questions and only the second is about you.
Cloning and comparing
You can also clone an RA strategy into your own strategies and change the rules, then paper trade both side by side. That costs two of your three running slots, and is a far better use of them than running two unrelated strategies you have not yet understood.
A mistake worth avoiding
The common failure is treating paper trading as a scoreboard rather than a test. If you run five strategies on paper and promote whichever one made the most money, you have selected on noise over a short sample — the same mistake as picking the best of twenty backtests.
Decide in advance what you are testing for, and judge against that. The same discipline that applies to reading a backtest report applies here: look at the worst stretch first, not the total.
The short version
Paper trading answers a narrower question than people assume, and answers it better than anything else.
- Prices are real and current; the counterparty is what is simulated
- It catches logic errors and bad parameters within days
- It cannot show you slippage, margin rejections or your own nerves
- Target conditions, not a day count — trending, quiet, expiry and a gap open
- It needs a plan but no broker, no IP and no capital
Treat the results as a ceiling. If a strategy is only marginally profitable on paper, it is very unlikely to be profitable live.
Frequently asked questions
The prices are real and current — they come from the live market at the moment the strategy acts. What is simulated is the counterparty, so your order never queues, never partially fills and never moves the price.
Yes. On liquid at-the-money strikes the fill assumption is close to reality. On thin strikes it is optimistic, so expect a larger gap when you go live.
You can, but then you have tested several strategies briefly rather than one properly. If you need to change something, treat it as restarting the clock.
It needs an active plan, but no trading capital. There is no brokerage and no loss, because no order reaches an exchange.
No. Paper orders are simulated, so there is no broker connection and no dedicated IP needed.
No. A backtest replays history in seconds. Paper trading runs forward in real time on live prices, which is why it catches a different class of problem.
Usually not exactly, and usually slightly worse. Paper fills are assumed rather than earned, so real slippage and real margin constraints only show up live.
Long enough to cover the conditions the strategy claims to handle — at minimum a couple of expiry cycles, including a trending day, a quiet day and a gap open.
Yes. Paper and live bots share the same limit on simultaneously running strategies.
Yes, and it is a sensible way to watch one before deploying it live.
Start with a free 3-day trial
Build a strategy, backtest it and run it on paper — no broker, no IP and no money needed to try it.
