Well put Howard. While in B-school, I had an internship with a chemical company that wanted to model margins for their product three years into the future. I gathered decades of data -- macroeconomic, industry production, sales, inventory levels, capacity, etc. -- ran my regressions, and created a model. The initial output of the model was garbage so we decided that we needed to remove data associated with special events (wars, industrial accidents, etc.). The result then showed margins quickly reaching an equilibrium point and stabilizing at that level. This was a highly cyclical industry so this did not fit expectations either. However, as we thought about it, we realized that what the model was telling us was that the industry has a theoretical equilibrium which it never reaches due to unpredictable external events that constantly knock the industry out of equilibrium. From that understanding, we were able to use the model to perform various "what if" analyses so the company could prepare for various scenarios. The model was not a crystal ball able to predict the future but a tool to explain possible outcomes given different scenarios.