Let’s say 1% of kids are born with allergies. The statistic is useful for a public policy maker, someone who is setting up food menus in schools or regulating restaurants. However, the statistic is useless for you as a parent. Unless you are planning to have around 10+ kids, you are never going to experience the statistical outcome. If you have one child, either the child would have an allergy or he won’t. Statistical outcomes matter only when someone plays a game sufficient times.

The same goes for many of the other statistics. A man with cancer does not care what the survival rate is. This man has one life and he wants to save that. A doctor treating 100 cancer patients a year can look at it as a statistical problem.

The average returns of the S&P 500 make sense for an investor who purchased SPY. The average returns of the S&P 500 are meaningless to an employee of a company listed in the S&P 500 list. She isn’t getting average career outcomes. Her career and compensation are tied to one company’s performance.

In many ways, this issue is similar to Binary vs Graded outcomes.

However, there are scenarios where one can only play a binary game. E.g. most people can hold only one full-time job at a time. Or, even worse, when someone is forced into a binary game. E.g. when infected by difficult-to-cure diseases.