We recently built an econometric model to forecast house prices for a specialist UK financial services group focusing on attractive segments of the UK retirement income market over the next fifty years. Our model used the latest developments in Bayesian machine learning to select key contributing factors out of a large dataset of identified potential house price drivers. Our approach is designed to uncover more stable relationships in the data than traditional econometric models which typically struggle in the face of large-scale data testing with limited data sample sizes. We also used our expertise in Monte-Carlo simulation to simulate thousands of potential house price scenarios to help the client understand and quantify the uncertainty around the central house price scenario. Our analysis enabled the client to develop a data-driven business strategy that reflects the likelihood of different house price outcomes over the following decades.