What paper trading shows that a backtest cannot
- Whether the strategy as deployed matches the one you tested, with the same legs, times and exits.
- How signals behave on the live feed rather than on clean historical bars.
- How often orders would have been retried or rejected, and the slippage you can realistically expect.
- Whether the daily limits and square-off behave the way you intended.
A checklist before going live
- The same strategy definition runs on paper as in the backtest. A separate live rewrite is where drift starts.
- Run it for long enough to see more than one kind of day, including an expiry day.
- Compare paper fills with the backtest trade by trade, and explain every difference.
- Set per-deployment risk limits and a daily maximum loss, and confirm the kill switch works.
- Confirm orders carry your algo identifier.
- Check the end-of-day reconciliation: your platform's positions should match the broker's book.
- Start live with the smallest size you can, and treat the first sessions as a test of the process.
Monitor the things that fail quietly
The dangerous failures are the silent ones: a broker session that expired, a feed that stopped updating, an order that never filled. Latency, slippage and fill telemetry per deployment, a session watchdog with automatic re-authentication, and an activity log of every state change make these visible. A zero that appears because something stopped working should be treated as a defect, not as a quiet day.
Be honest about what is and is not ready
Backtesting and paper trading are available to everyone on the desk. Live execution runs through your own broker account and is currently open to invited accounts while broker testing widens. Whatever platform you use, do not skip the paper stage because the backtest looked good.
Common questions
How long should I paper trade before going live?
Long enough to see different market conditions, including at least one expiry day, and to compare every paper trade against what the backtest would have done. There is no fixed number of days, but you should be able to explain any difference.
Is paper trading the same as a backtest?
No. A backtest replays history; paper trading runs your strategy against the live market feed with simulated fills, so it exposes data, timing and execution behaviour that history cannot.
What risk controls should be set before live trading?
At minimum per-deployment limits, a daily maximum loss with a working kill switch, a time-based square-off, and a small starting size.
Try it on the desk
Build the strategy without code, run it over history, and read the result before any money is involved.