Hidden Execution, Data Continuity, and Timing Assumptions in Backtests
Summary
The article explains how a backtest can look convincing while relying on conditions that may not hold in live markets. It examines three assumptions: orders fill immediately at the signal price, market data forms a continuous sequence, and prices from a completed bar are available when a signal is formed.
These shortcuts can overstate strategy performance. A quoted price may lack enough liquidity or counterparties, while gaps, halts, contract rolls, and thin trading can disrupt the path implied by orderly bars. Using a bar’s eventual high or close too early also creates look-ahead bias. The article offers conceptual cautions and examples, but no measurements or empirical tests. Its central guidance is to make backtest assumptions explicit and examine whether results remain credible when execution, data gaps, and information timing are modeled realistically. The severity of each issue depends on the strategy’s time horizon and its reliance on precise entries.
Key ideas
- Assuming an order fills immediately at the signal price can exaggerate returns when liquidity or counterparties are limited.
- Market interruptions, gaps, and contract changes can make orderly bar sequences misrepresent tradable price paths.
- Prices such as a bar’s final high or close may not be known when a strategy first evaluates its signal.
- Backtest results become more informative when their execution and information assumptions are explicit and scrutinized.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.