Energy-aware programming languages and frameworks seek to improve the energy efficiency of computer systems by taking advantage of application-specific information to perform energy optimizations. Effectively predicting an application’s energy behavior enables more powerful energy optimizations and additional energy management techniques. However, application energy consumption is fundamentally dynamic in nature, which limits the amount of effective energy prediction that can be done ahead of time. To address this challenge, we propose FJP, a predication-aware semantics that takes the estimation process online, and serves to partially predict a program’s energy consumption. FJP represents a first step towards improving the expressiveness of energy-aware language and framework techniques in its ability for online adaptive energy prediction.