Paper, Backtest, and Live Differences
Published Jun 22, 2026 · Updated Sep 6, 2026
Different results can come from execution assumptions or from a fault. Compare data, configuration, orders, and positions before deciding; matching strategy names alone does not make two runs comparable.
Backtest vs paper
| Difference | Why |
|---|---|
| Market window | Backtests replay the past; paper trades the present |
| Fill model | Backtest and paper use simulation rules; neither records your actual exchange fills |
| Duration | Months of backtest vs days of paper isn’t comparable |
| Version | Confirm paper runs the exact backtested parameters |
Record the data source, date range, initial equity, template version, and parameters. The current Web Engine Backtest uses its default execution model; do not assume it exposes the same controls as the CLI or live venue. Observe paper behavior across different conditions before drawing conclusions.
Paper vs live
| Difference | Why |
|---|---|
| Slippage and depth | Live eats the real book; grows with size |
| Fees | Live pays your actual tier rates |
| Funding | Live positions actually pay/receive funding |
Compare the paper balance and position limits with the intended live configuration, and separate simulated costs from actual venue charges. A similar paper result does not guarantee live performance.
When it IS a real problem
Report with both strategy names, the timestamp, and event-history entries if:
- Paper and live take opposite directions on the same signal at the same time;
- An order shows filled in events but the exchange shows no position change;
- Same parameters but different config revisions on the two detail pages.
Related
When you are ready to scale from paper to live, follow From Paper to Live.