Visible light energy transfer catalysis has emerged in recent years as an attractive synthetic method for accessing high-energy intermediates, leading to the discovery of novel reactivity modes inaccessible via thermal methods. Computational methods have played a crucial role in understanding and predicting energy transfer catalysis, bypassing the need for complex and laborious photophysical measurements. Specifically, adiabatic triplet energies have been used as a predictive tool in the design of substrates amenable to sensitization, as well as a mechanistic tool. However, this approach fails to accurately predict the likelihood of triplet energy to molecules that undergo large structural changes upon excitation, and provides qualitatively incorrect predictions of E/Z-isomerism under energy transfer catalysis. Here, we introduce a new metric, dynamic vertical triplet energies (DvTE), based on the evaluation of change in vertical energy gaps throughout direct dynamics trajectory simulations. This approach improves the predictive capabilities of density functional theory computations and provides further support for the "hot-band" mechanism of energy transfer. We demonstrate excellent performance, with R2 = 0.96 and a mean absolute error (MAE) of 2.1 kcal/mol, for a collection of 20 small organic molecules, whereas the traditional adiabatic model performs significantly worse (R2=0.58, MAE = 8.3 kcal/mol). We anticipate this approach will be valuable for predicting E/Z isomerization triplet energies, for which currently there is no available computational protocol.