Data-Driven Business Decisions

CHAPTER 3

Dealing with Uncertainty and Chance

3.0. Introduction: What Is the Issue?

Probability is a set of rules and ideas for dealing with uncertainty. When properly applied, it allows us to understand, predict, manipulate, and account for the unpredictable nature of most business outcomes. Here are four instances where you would need some probability theory to arrive at a rational decision.

• Finance Some share analysts claim to be able to pick when the market will rise and when it will fall—not 100% of the time but consistently better than guessing. What really constitutes outstanding performance for the task of predicting stocks? To pin the question down, suppose we asked the analyst to predict the percentage change in tomorrow's market index and we then allocate this into five categories: little change, moderate fall, moderate rise, large fall, and large rise. The definitions of these categories could be adjusted so that each happens with equal probability (20%). Over 1 year (250 trading days), the analyst makes a prediction of what kind of market it will be the next day. So the analyst should get around 20% of 250 or 50 predictions correct (if she is essentially guessing). How much better than 50 would she need to do to distinguish herself as a genuine market seer?
• Health Care Suppose you test positive for certain lymphocytes that are indicative of leukemia. The test is not 100% accurate but there are data on the track record of the test applied to a wider population. What ...