1 min read

(Over)Confidence Interval

Why isn’t financial analysis more probabilistic? We make all these assumptions and predictions in making a financial analysis or forecast, but for all the thought that goes into that, those assumptions get buried in the model, hidden behind single value variables. We don’t build ranges right into the model (except in a sensitivity analysis, which is practically an afterthought).

Why don’t we build models to output an expected range of values with a confidence interval (rather than a single value)? Yes it’s more complex, but not much more than the multivariate models used throughout the business world today. It makes the uncertainty much more transparent than anything widely used today.

Maybe the convention of using single values is a big part in much of the bad risk taking going on.