The Equitable Threat Score (ETS), also known as the Gilbert Skill Score, is a skill score for binary (Yes/No) precipitation forecasts. It specifically accounts for hits that might have happened by chance.
The “Equitable” part refers to the fact that the score penalizes “hits by chance.” If it rains 90% of the time in a location (like a rainforest), a forecaster could get a lot of “hits” just by saying it will rain every day. The ETS subtracts those random hits to show the true skill of the forecaster. Industry pros search for “ETS vs. Threat Score” to see a more realistic view of model performance in rainy climates.
While ETS is great for common rain events, it can be unstable for very rare events (like snow in Miami). In those cases, even a single “Hit” or “Miss” can cause the score to swing wildly. Professionals search for the “Symmetric Extremal Dependency Index (SEDI)” as a modern alternative when verifying rare, high-impact weather events.
While the Brier Score verifies the probability (0-100%), the ETS verifies the binary outcome (Did it rain or not?). Forecasters use the Brier Score to tune their confidence and the ETS to evaluate their ability to correctly “call” a storm.