Between a promising backtest and real capital lies a step many traders skip: running the strategy against live data with simulated orders. Paper trading will not prove a strategy works, but it is remarkably good at revealing why it might not.

What paper trading catches

  • Data differences. Live feeds arrive late, out of order or with gaps that historical datasets have quietly cleaned.
  • Code paths never exercised. Session boundaries, holidays, corporate actions and reconnects rarely appear in backtests.
  • Timing assumptions. Signals computed on a closing price cannot be executed at that same price in reality.
  • Operational load. Monitoring, logging and alerting either work under live conditions or they do not.

Designing a meaningful paper period

Decide in advance how long the period will run and what results would cause you to stop. A common approach is to require enough trades for basic statistical comparison with the backtest — dozens at minimum, more for high-frequency strategies. Write the criteria down before you start; it is surprisingly easy to move goalposts once results arrive.

Decide what would make you abandon the strategy before you see the results.

Compare like with like

Re-run the backtest over exactly the paper-trading window and compare trade by trade. Differences in entry times, prices and position sizes point directly to modelling errors. Frameworks such as Freqtrade (dry-run mode) and QuantConnect (paper brokerage) make this comparison straightforward.

Limits of simulation

Paper fills are still simulated. They will not capture queue position, partial fills or the impact of your own orders. For strategies where these matter, a final stage of small live allocation is the only realistic test.

Checklist before going live

  • Pre-registered duration and stop criteria
  • Trade-by-trade reconciliation with the backtest
  • Tested kill switch and maximum-loss limits
  • Alerts for data gaps, rejects and unexpected positions
  • A plan for the first live allocation size

Common surprises

Traders who run disciplined paper periods report a consistent set of discoveries. Strategies that rebalance at the close find that their orders cannot be placed in time to execute at the closing price they assumed. Signals depending on data released at a specific time find the release arrives seconds or minutes late through their feed. Strategies trading many instruments discover that some symbols have changed, been delisted or halted.

None of these issues necessarily invalidates a strategy, but each must be resolved before real capital is at risk. Finding them in a simulated environment costs nothing; finding them live can be expensive.

Monitoring as a deliverable

Treat the monitoring set-up as something the paper period must prove. By the end, you should know how you will learn — within minutes, not days — that a strategy has stopped trading, is trading unexpectedly, or has lost its data connection. Many traders discover their alerts were silently failing only after something goes wrong.

When to stop paper trading

Paper trading indefinitely can become a way to avoid decisions. When the pre-agreed duration ends, evaluate against the written criteria and decide: proceed with a small allocation, return to research, or abandon the idea.

Quorient Editorial Desk

Articles credited to the Editorial Desk are researched and written for Quorient and reviewed by the editor, Callum Reid, before publication, in line with our Editorial Policy. Spotted an error? Request a correction.