When the popular LaxPower website shut down last offseason, MaxLax undertook a project to create our own mathematical formula to compare teams in order to give fans — and coaches and players — an additional method of comparing teams.
In the initial calculation, run before Monday’s games, the top of the MaxLax LA Girls Lacrosse Power Rankings looks remarkably similar to the coaches poll Top 20 released Monday, with Palos Verdes, Redondo Union, Westlake and Glendale ranked Nos. 1 through 4.
The algorithm did produce a couple of interesting rankings that deviate from the coaches poll significantly. Newbury Park, for example, is No. 9 in the coaches poll but No. 5 according to the power ranking, although it should be noted the Panthers (2-0) showed only two game results through Saturday, when the calculation was made. Friday’s home game against No. 3 Westlake should provide more insight into the Panthers.
Marlborough, too, is rated much higher by the algorithm at No. 8 than by the coaches poll, which has the Mustangs (2-4) ranked No. 17. Losses to Westlake, Glendale, Crescenta Valley and Chaminade — all Top 10 in the power ranking — prevented Marlborough from dropping far in the calculation.
In contrast, three teams ranked in the Top 15 of the coaches poll were ranked lower by the algorithm — Oak Park, Downey and Palisades.
The complete MaxLax LA Girls Lacrosse Power Rankings are available here, with a MaxLax subscription.
Developed by Michael Traub (Aliso Niguel ’13, MIT ’17), the algorithm processes the final scores of completed games to calculate a rating based on each team’s goal differential in competitive games.
The objective is to determine what each game outcome teaches us about a given team, then reflect that in a numerical rating. An expected blowout that actually ends as a blowout is non-competitive and teaches us nothing. An expected blowout that winds up being close, or competitive, teaches us something.
For these reasons, the algorithm relies heavily on the concept of competitive games, defined as those in which the score was close or was expected to be close based on each team’s relative power rating. These games reveal a team’s true quality and have much more influence over the ratings than non-competitive games.
To hedge against teams receiving credit for running up the score in non-competitive games, the algorithm also evaluates how much a particular game should influence the ratings by calculating a “competitive factor” (CF). Games with a high CF influence the power ratings more than a game with a low CF.
Have questions or comments about the new feature? Contact us at firstname.lastname@example.org.
Editor’s Note: MaxLax is a community- and subscription-supported website whose mission is to spotlight lacrosse, its teams and its players in Southern California. This post is available to the public, with no subscription required. If you follow lacrosse in SoCal and enjoy our coverage, please consider supporting MaxLax with a subscription.