This paper evaluates the Generalized Adaptive Capped Estimator (GACE), a deterministic forecaster that blends several growth signals, applies volatility-aware asymmetric caps, damps seasonal factors, and moderates recursive growth. GACE 1.0.0 is distributed on CRAN. The benchmark runs the CRAN R package directly, with R forecast-package baselines (naive, seasonal naive, ETS, Theta, ARIMA(1,1,1)+drift), in 24 rolling-origin tasks across eight public series spanning monthly, quarterly, annual, and nonseasonal data. GACE reached mean sMAPE 27.02, mean MASE 2.113, median runtime 3.1 ms, and zero detected forecast-path instability. Theta had the best mean MASE (1.445) and ETS the best mean sMAPE (23.24); GACE ranked last on mean MASE and fourth on mean sMAPE, while running faster than every fitted statistical baseline. A Friedman test found overall MASE differences (chi-square = 33.55, p = 2.9e-6); GACE's MASE was significantly worse than Theta, ETS, and seasonal naive, and statistically tied with naive and ARIMA. The evidence supports a conditional claim: GACE is a fast, stable forecasting alternative, not one that dominates established statistical methods on average accuracy.

time-series forecasting,GACE,robust forecasting,structural change,adaptive capping,forecast stability,ARIMA,Forecast