Paper trading should test the complete operating system around a strategy, not only whether simulated entries and exits are profitable. It is the best time to discover duplicate orders, stale data and unsafe recovery behavior.
The bot should be idempotent: receiving the same event twice must not create two positions.
Run long enough to encounter different market conditions. A few successful days establish almost nothing about robustness.
Set position limits, loss limits and alerts before enabling live orders. Document how to revoke the key and close risk manually if the automation stops responding.
Test the event flow
Record how market data arrives, how signals are produced and how order state is reconciled. Simulate delayed updates, rejected orders, partial fills and a process restart while an order remains open.The bot should be idempotent: receiving the same event twice must not create two positions.
Measure realistic performance
Include fees, spread, slippage, funding and latency. Compare expected and simulated fills. Track drawdown, exposure, turnover and the difference between the strategy signal and the executed portfolio.Run long enough to encounter different market conditions. A few successful days establish almost nothing about robustness.
Prepare restricted live access
Create a dedicated API key with only the permissions required for trading. Disable withdrawals, restrict IP addresses when supported and store the secret outside source code.Set position limits, loss limits and alerts before enabling live orders. Document how to revoke the key and close risk manually if the automation stops responding.