More on forecasting

One final slide from the Mike Mauboussin Credit Suisse seminar I attended. Mike discussed the latest research on how executives make decisions. They usually use a single simple analogy or a few cases for making a decision. 

We can do better at classifying the choices of executives as seen through this simple two by two matrix. We can think about a reference class and the weighting between those references. If an executive uses a single case for making a decision, then he is having single recall with a weighting that is event based. In this case, the simple single event story helped to make the decision. If there are a few cases, the executive is using  similarity based weighting. A set of cases are used to make the decision. 

However, we can also think of decisions that are based on a distribution of events. The upper left hand corner would be using the law of large numbers to make a decision. Play the odds based on the set of all events. This make sense of you have a lot of information. The lower left hand corner may be the best possible outcome because you use distributional information as well as conditioning to help make decisions. 

Conditional views on large data sets will give you the best chance of success and also allows the decision-maker to place some well-defined bounds on any decision.