Most arbitrage backtests flatter themselves. Arbitrage backtesting only means something when every leg could really have been bought at the size, time and terms you assumed. This lesson shows why last-seen quotes and a 100% fill assumption inflate results, and how to build a test that does not. You follow a funnel from a thousand raw signals down to the few that complete both legs after contract checks, freshness, limits and execution delay. You learn the strict information cutoff, so a decision at time t uses only what was known at t. Then you see what to report beyond wins: failed attempts, tied-up capital and rejection rates. A number you can defend beats a big one.
Learning objectives
- Explain this technique and its assumptions in the context of real market rules.
- Reproduce the worked calculation or scenario.
- Identify false positives and when a proposed position should be rejected.
