Backdoor Criterion

The backdoor criterion is a formal rule in causal inference used to identify the set of variables that must be controlled to block all non-causal paths between two variables. By blocking these backdoor paths, analysts can isolate the true causal effect of one variable on another.

In the context of market microstructure, this might involve controlling for order book depth when studying the impact of trade size on price impact. If the backdoor paths are not closed, the estimate of the causal effect will be biased by confounding factors.

The criterion relies on the structure of the DAG to determine which variables are necessary for adjustment. It is a fundamental tool for ensuring that empirical results are not just artifacts of correlation.

Proper application of the backdoor criterion is essential for rigorous quantitative analysis in derivative markets.