The most important stats in a lacrosse game

Mathematical analysis of sports performance has become an important tool for player evaluation. In baseball, Bill James analyzed more than a hundred years worth of box score data to determine which statistics had an impact on game outcomes. Over time, Baseball's General Managers have used his system – Sabermetrics –  to evaluate players. GMs seek to trade low-rated players for higher-rated players to improve their teams' onfield performance. The same statistical exercise goes on in basketball, football, and hockey. Now, the math whiz behind LaxPower, Larry Feldman, has produced a set of factors that determine, not only player performance, but tell us which factors of our game have the biggest impact on winning and losing Lacrosse games.

 Once these factors are given relative weights – determined via multivariate regression analysis – it is possible to rank player performance. Feldman goes one step farther, by adjusting player's performance rankings to the quality of opponents a player and his team face. This data, for NCAA Divisions I, II, and III, is now available.

 So, what factors from the box score does he consider and how does he weight them to produce accurate predictions of game outcomes and relative player performance? Here they are, sorted by largest absolute number, regardless of positive or negative value:

 Assists 1.25
Goals 1.00
Caused Turnovers 1.00
Goals saved 0.80
Goals allowed -0.80
Turnovers -0.60
Groundballs 0.25
Faceoffs won 0.22
Faceoffs lost -0.22
Shots on Goal -0.10
Missed shots off goal -0.05

 Without having attempted to quantify these factors myself, I – like many lacrosse observers – have long felt that Assists, Goals, Caused Turnovers, and Goalie Save percentage were the most important determinants of victories. That's not to diminish Groundballs. Though Feldman gives a positive weight of 0.25 to a single Groundball, fact is that there are LOTS of Groundballs in a game, and the team that wins the majority of them will cumulate a substantial statistical advantage.

Still, Feldman has produced a set of "weights" for lacrosse statistical events that help us understand the game in new ways. I believe that Feldman's system, or a variant thereof, will become a standard set of tools used by coaches to analyze their own players, as well as the players on opposing teams.

Finally, Feldman takes into account the quality of competition faced in determining overall rankings of players by position. I'll post on this player ranking scheme in the next day or two.