Back in the prehistoric days of coaching basketball, most player evaluations were done by the eye test. One would have been hard pressed to find a coach who couldn’t rank his players 1-15 overall, or for that matter, in any specific phase of the game (defense, for instance) and they did so without the use of anything remotely similar to Player Efficiency Ratings(PER).
But, in the modern-day analytics arms race in all sports, the terms “eye test”, “best guess”, “hunch” and even “strategy” don’t have the impact they once had because of the recent trend of relying on mathematical data for coaches to make decisions of all kinds.
Take evaluating one’s own players, as an example. All NBA players and many NCAA players’ “net worth” is determined by John Hollinger’s PER (Player Efficiency Ratings). Before PER came along, most coaches could, and still do, infer a wealth of knowledge about their players with a five-minute examination of a season box score. While I wouldn’t begin to argue PER’s validity or usefulness, there are at least two good reasons for all coaches to consider modifications to PER as the only mathematical tool for evaluating their own players:
Player Efficiency Ratings only takes official box score stats into account. One overlooked “stat” that many coaches might find important is Turnover +/-. (Please read previous Hoop Coach article, “Turnover +/-: The Defensive Stat You Should Be Tracking”). Other important unofficial stats to consider are the NBA Hustle Stats: Contested Shots, Charges Drawn, Deflections, Loose Ball Recoveries and Screen Assists. (Take a look at Hoop Coach article, Examining the NBA’s Player Hustle Stats: A Follow-Up” for a little more clarity). Other important concepts to consider are a player’s unsuccessful attempts at blocks, steals and taken charges. To give credit for these successful actions with no penalty for misses sends the wrong signals to players and warps their true net worth.
Since every system is different, it makes sense that the players in each system be evaluated by the expectations of that system only. As an extreme example, John Beilein’s Michigan’s and Bob Huggins’ West Virginia’s systems are at opposite ends of the spectrum. While Michigan’s program is, in good part, founded on low turnovers and fouls committed, West Virginia’s system would more easily excuse fouls and their own turnovers, in exchange for opponents’ turnovers. In short, WVU scores a significant percentage of their offense off their pressure defense, while Michigan scores at a high rate off their offensive sets. As a result, each coach would likely place greater Player Efficiency Ratings weight on the official and unofficial stats that best feed into each of their systems.
Obviously, charting unofficial stats requires either live statisticians charting said stats or video work after the fact. The key, however, is for each coach to determine for himself which stats are important to him and then to take control. Analytics for analytics sake is a waste of time. Any coach can and should track whatever he wants for whatever reasons he wants. Fitting statistical analysis into the uniqueness of a program should probably be the real focus, not trying to force the concept of analytics into a program. It just takes a little independence and a little creativity to get started. A team’s players’ unique net worth is a good starting point.