Does A Coach Have To Be A Great Player

There is a common belief that a great player will make a great coach, but is it true? I have long thought that it is not true. Going back to statistics and learning about probabiliy distribution, each skill or trait should have a distribution where the elite skill is only present in a small percentage of the population. A specific combination of skills and traits will therefore be even less likely. Simplified, if elite passing is present in 2% of the population, elite passing and elite hitting will be present in 2% of 2% (or 0.04%) of the population. The greater number of skills, the less likely that combination is present in the population.

Then comes coaching. I think it is common for people to think that playing ability equates to coaching ability. Coaching is a completely different animal. Some players are good at a skill, but they might not know how to describe what they do, or how they do it. They might not be able to teach how to do it. What they do might be instinctive. I remember hearing a sound clip of Brett Favre thanking Ty Detmer in an interview or press conference for teaching him what the Cover 2 defense was. Favre was a future Hall of Fame quarterback, and Ty Detmer was his backup who only played in 7 games during their time as teammates.

Coaching is a complex stew of skills. Knowing the fundamental skills of your sport is important, but being an elite player and having the skills to be an elite coach is going to be ridiculously unlikely. While not approached from this theoretical base, this article about Barry Bonds on fivethirtyeight.com demonstrates this idea with numbers. It's a nice treatment on the subject using baseball's massive collection of stats. It takes advanced playing stats of Baseball hitting coaches who played major league baseball and an advanced stats measure of team hitting stats versus their expected stats.

As theory predicts, there is no correlation between playing ability and coaching ability. In fact the plot looks like just another way to visualize the probability distribution. There are a lot of hitting coaches who were poor players relative to the rest of the population evaluated. There are fewer hitting coaches who were great players. Average improvement over expectations for team performance is pretty close to zero. The line on the graph indicating that average actually drops slightly below zero improvement as the coach's playing ability increases. This suggests that hiring a stud hitter as a hitting coach will most likely result in no improvement over the previous hitting coach. It almost suggests that hiring the stud hitter will actually result in a decline. The best hitting coach was a pretty poor hitter compared to the rest of the sample.

The take home message is that coaches shouldn't worry about how good of a player he or she was. It is more important to learn and improve in those skills specific to teaching the game, and improving your knowledge base.