Summer ENSO gave no clear added dry-winter probability skill in seven California divisions in 1981-2025

We tested whether July-September Nino 3.4 adds dry-winter probability forecast skill beyond prior winter precipitation. We used fixed NOAA nClimDiv and ERSSTv5 records in seven California divisions. Each forecast used earlier winters for its dry threshold, predictor scaling, and logistic fit. There were 45 test winters, 1981-2025. Dry means precipitation at or below the training one-third quantile for November-March. Adding the ocean index gave -1.24% pooled Brier skill, with a conditional 95% five-year block interval of -5.34% to 3.34%. Both predictors also gave -3.40% skill against an intercept-only reference. Fixed checks for penalty size, recent training, and a lower-quarter event had negative point estimates and intervals that included zero. Removing any one year from scoring left pooled skill negative. The result gives no clear added skill under these models. It does not rule out useful ocean information or establish a physical cause. Short records, changing relative thresholds, retrospective products, simple models, and conditional uncertainty limit inference. The package includes source subsets, all forecasts, code, bootstrap indices, a figure, an independent numerical audit, and offline repeat instructions.

ENSO · California · dry winter · probability forecast · Brier score · chronological validation · null result