For a regional U.S. bank Prescio updated previously developed, as well as developed new, models of the Energy segment of the commercial loans portfolio for the CCAR stress testing exercise. This Energy segment includes tightly structured lending facilities for companies involved in Exploration & Production and Midstream and Energy Services businesses through a variety of single bank and syndicated facilities that support general corporate purposes. The model development and update involved fitting a model to the historical segment’s multi-billion dollar balances. Multiple modeling approaches were considered including ARIMA and Cochrane-Orcutt regression techniques. The models were constructed utilizing standard macroeconomic factors such as GDP, unemployment rates, Treasury rates, housing starts, and so forth. Developed models were tested to ensure the models were homoscedastic, the explanatory variables were not collinear, and that the residuals were stationary, Normally distributed, and without autocorrelation. The models were back-tested by removing segments of data, refitting the parameters and examining how the new models fit the omitted data. The forecasts of the models in Baseline, Adverse, and Severe scenarios were considered and compared to business intuition. Candidate models which met all criteria had their sensitivity analyzed by examining their response to having both their input data and their parameters varied.