The rainfed paddy farming system faces profound vulnerabilities due to daily climate non-stationarity. This mixed-methods study in Central Java analyses daily climate signals, total rice production, and household adaptation over 25 years (2001–2025). Moving beyond simple correlation, a Principal Component Regression model integrating five climate variables and three agronomic confounders reveals a profound climate–production decoupling. The composite climate index explains only 7.9% of total production variation, while non-climate factors account for 92.1%. Physical stability is maintained through asymmetric temporal scheduling and a distinct hierarchy of responses, employing active, planned adaptations alongside passive, reactive coping. However, quantitative household evaluation reveals this tonnage stability incurs severe hidden costs; the titip gabah post-harvest system maintains a high Yield Stability Index (0.93) but yields a negative Return on Storage (−7.15%), functioning as a risk-mitigation buffer rather than a profit-maximising tool. Furthermore, climate anomalies drive the progressive alienation of traditional ethnoclimatological knowledge, forcing a cognitive shift toward hybridised decision-making. To prevent passive coping from evolving into systemic maladaptation, we propose a stratified policy framework ranging from village-level knowledge integration and Subdistrict daily risk warnings to regency-level subsidies targeted at smallholders (<0.5 ha).