Last year I wrote about projection fragility as it applies to DFS, and this year I’m taking it a step further to explore it in the best ball format. The core principle is the same, in those projections (which, under the hood, power season-long rankings) are for median values, but in tournaments (whether best ball or DFS), we don’t care about median outcomes. We need ceiling outcomes to win. For the non-mathematically inclined, a median value within a range of outcomes means that half the time, the actual outcome should be below the median, and half the time, it should be above the median. Here are examples of two types of distributions of values that can both have exactly the same median:
Which of those would you want in a tournament? Well, the answer isn’t that simple. You don’t ALWAYS want one or the other type of distribution because it depends on the rest of your roster. The important thing to know is that the barbell distribution has a broader range of actual outcomes. In a normal distribution, most outcomes are clustered near the median, whereas in a barbell, most of the outcomes are fairly far from the median. So if you’re seeking to add more variance to a roster, you’d want a player with a barbell distribution, whereas if you feel like your roster already has a lot of variance and you want to add some stability, you’d want a player with a normal distribution.