Definition
A dry run is the practice of running a trading system in full operational mode — monitoring markets, detecting opportunities, calculating position sizes — without placing any real orders. All trades are logged as if executed, letting you measure what the system would have done with real capital.
What dry runs validate
- Matching quality: Are the identified pairs genuine complements, or are some false matches?
- Opportunity frequency: How many viable spreads (positive net edge after fees) appear per day?
- Edge distribution: What are typical spread sizes? How often does a spread exceed 2%?
- Execution logic: Does the system correctly calculate position sizes given book depth?
- Circuit breaker logic: Does the unwind trigger correctly in simulated failure scenarios?
Arbitrage Agent's dry-run mode
Arbitrage Agent's Starter plan runs in dry-run mode. The agent monitors thousands of matched pairs in real time, logs every opportunity it would trade (with exact size, edge, and reasoning), and populates the dashboard with simulated P&L. This lets you see exactly what live trading would look like — with zero capital at risk.
Limitations of dry runs
Dry runs assume your simulated orders would fill at the quoted price. In reality, placing large orders moves the price. Slippage, partial fills, and API latency affect live results in ways dry runs can't fully capture. Live results are typically 80–95% of dry-run performance, depending on position size and market liquidity.
How long to run dry before going live
Two weeks of dry-run data typically provides enough signal to evaluate opportunity frequency, edge quality, and system reliability. More data is better — but waiting indefinitely for "perfect" confidence is not necessary. Arbitrage Agent's circuit breaker ensures losses on individual trades are bounded even if live behaviour differs from dry run.