Adverse Selection in the Draft

Predictive analysis is really easy when there is a market. If I am making a model to project baseball games, and my model says that the Rockies are 90% to win against the Mets when Max Scherzer is pitching, all I have to do is look at any sports book to see that I am wrong, and my model is terrible. This is easy enough.

Making predictions without a market is tough. Take the MLB draft. Sure, there are public rankings, and some of them are quite good. At the end of the day though, the draft has a lot of hidden info not baked into public rankings, and all 30 teams aren’t publicly submitting their rankings, so it’s hard to get a sense of how good your predictions are. More importantly, it’s hard to tell if you have edge or if you are getting adverse selected.

The adverse selection problem is particularly impactful if you work for a team that just started using analytics in the draft. Imagine that you find that backspinning fastballs are good and sinkers are bad, and thus you only draft the former and never the latter. This is generally correct, and many teams did well by doing this over the past half decade. But in 2022, the teams that exploited that edge have moved on and are learning new things, whereas your department hasn’t caught up yet due to a lack of experience. What do you do if there is a pitcher in the seventh round who fits all of these prized traits? He has a backspinning fastball with a ton of carry, and a hammer 12-6 curveball to go along with it. Seems like a slam dunk pick. However, you have to consider why this player has fallen to the seventh round, and why teams who exploited this edge prior to you have passed him over. If it’s a medical reason, but your doctors are fine with his medicals, you are good. If it’s a makeup issue, but your player development met with him and likes his personality, go ahead and draft him. If it’s neither, you should probably be concerned. At this stage of the draft, it probably makes sense to take him because the stakes are fairly low, but I would pay special attention to how he develops and what new things you learn about him after he is in your organization. By understanding that there is not a market to look at and you can be getting adverse selected as a result of this, a team that is just starting with analytics can get up to speed quickly.