When Code Is Correct, Tests Pass, but the Feature Never Actually Runs

A software team discovered that eight of nine real bugs in their codebase shared a common trait: the code was written correctly and tests passed, yet the relevant code path never executed in production. One case involved a forecasting tool where a volatility-conditioning feature was permanently disabled by a default boolean set to false, causing weeks of incorrect outputs with no errors or alerts triggered. Another bug stemmed from stateful functions in Pine Script being placed inside conditional branches, meaning only the active branch updated its internal state, silently corrupting moving average calculations. A third issue involved a text sanitiser designed for a specific character substitution that inadvertently mangled indentation across unrelated files when run globally. The core lesson identified is that unit tests validate functions in isolation, but fail to catch misconfigured defaults, broken wiring, or unintended side effects on untested code paths.