Backtesting an alternative data signal is fundamentally different from backtesting a price-based strategy. Price data is clean, uniform, and retrospectively consistent. Alternative data is noisier, often irregularly sampled, and subject to changes in coverage or methodology over time. Getting a backtest to "work" is easy. Getting a backtest that reflects what you would have actually experienced as an investor is much harder.
This post covers how to approach alternative data backtesting rigorously: what questions to ask before you start, how to structure the test, and how to interpret results w