Tempo-Free Lacrosse FAQ

With Tempo-Free Lacrosse fully up and running (hopefully), I’ve fielded a few questions about the what, the why, etc. behind the numbers. Here’s a rundown of some of the commonly asked questions. If you have any you’d like answered, feel free to comment (or hit me on the Twitter).

What is this thing?

The concept of tempo-free statistics – to the best of my knowledge, at least – started in the world of basketball. The basic concept is that teams play at a different tempo, or pace, and so comparing the raw numbers (particularly scores) can be misleading.

Let’s put it this way: Lacrosse Team A gives up 10 goals a game, but the opponent has 30 possessions. Lacrosse Team B gives up 7 goals a game, but the opponents has only 10 possessions. Team A’s defense is much better than Team B’s, even though it gives up more goals per game. It’s giving up far less (.333 to .700) per possession.

The concept extends to basically any stat. The goal of TFL is to be a Kenpom-like site for lacrosse. Of course, the world of advanced statistics in lacrosse is far less developed, so there will be growing and learning together – that’s why comments, questions, etc. are greatly appreciated: we’re all in this together.

So what doesn’t come over from basketball?

First, there’s the tricky issue of an uneven number in possessions. Unlike basketball, where each possession by one team immediately results in a possession by the other, a lacrosse team could theoretically control every possession in a lacrosse game thanks to the concept of the faceoff.

There are also statistical… inefficiencies… when it comes to the way the record-keeping works in lacrosse. That means the whole thing is still a little developmental.

What are the big numbers?

OK, this is where is gets real. The calculations and explanations:

OFFENSIVE POSSESSIONS:
Faceoff wins + offensive clearing attempts + opponent offensive failed clears

DEFENSIVE POSSESSIONS
Faceoff losses + defensive clearing attempts (opponent offensive clearing attempts) + offensive failed clears

POSSESSION PERCENTAGE
Offensive possessions / (offensive possessions + defensive possessions)

DEFENSIVE/OPPONENT POSSESSION PERCENTAGE
Defensive possessions / (offensive possession + defensive possessions)
AKA
1 – possession percentage

OFFENSIVE EFFICIENCY
Goals scored / offensive possessions

DEFENSIVE EFFICIENCY
Goals allowed / defensive possessions (AKA Opponent offensive efficiency)

ADJUSTED OFFENSIVE EFFICIENCY
Offensive efficiency * (nat. avg. defensive efficiency / avg. opponent defensive efficiency)

If team A scores a ton of goals against poor defensive teams, the adjusted efficiency should be lower than raw efficiency. National average defensive efficiency will be a lower number than average opponent defensive efficiency (high numbers are bad for defensive efficiency), so the fraction will be less than 1. That will make the adjusted number lower. You can’t be the best team in the country by beating Wagner a bunch of times.

ADJUSTED DEFENSIVE EFFICIENCY
Defensive efficiency * (nat. avg. offensive efficiency / avg. opponent offensive efficiency)
Same way, if they let in very few goals because they’re playing shitty teams, the fraction will be greater than than 1, the adjusted defensive efficiency will be a higher number (worse efficiency) than the raw.

PYTHAGOREAN EXPECTATION (AKA TFL RATING)
Win = 1 / (1 + (runs allowed / runs scored) ^ 2)

This is something that comes from baseball, but applies to various other sports with adjustments to the exponent. Our “runs allowed” is adjusted defensive efficiency TIMES defensive possession percentage, to equate to goals allowed per tempo-free game. Our runs scored is adjusted offensive efficiency TIMES possession percentage, to equate to goals scored per tempo-free game.

Thus
Win% = 1/ (1 + ((adjDEff * defPOS%) / (adjOEff * POS%)) ^ x).

What else is there?

As noted above, we’re still discovering as we go along. A lot of the general stats in the game – turnovers committed, caused, etc., can be adjusted for the tempo-free numbers (turnovers committed per offensive possession or caused per defensive possession, for example). Poke around the team and game pages a bit, and if there’s something you don’t understand, or something that you think would make sense to add (or at least explore more), drop me a line. I’m all ears.

The 2013 season is the first in which we’re expanding to player stats, so there may be some hiccups along the way. Bear with us and point out anything that doesn’t seem quite right.

12 Responses to Tempo-Free Lacrosse FAQ

  1. Jason says:

    Is there a way to get the data from TFL into excel, including the individual game outcomes (scores or just w/l)? I’ve got some ideas I’d like to test out.

    • Tim says:

      As far as I know, there is not. You might want to ping Matt over at College Crosse, because he keeps his data all Excel-ified.

      Anything in particular you were trying to accomplish with the data? We’re always looking for new features to be added.

      • Jason says:

        I want to test the “expected win probabilities”. I have a hunch, that lacrosse is a game where a team that has a physical/scheme/talent advantage can repeatedly take advantage of that, and as a result, wins are less random than hockey, baseball, soccer, etc.

        I want to look at the games where a less than 20% or so win probability was expected, and see if the underdog is actually winning the expected share.

  2. CU77 says:

    What value of the exponent are you using for the Pythagorean expectation?

    • Tim says:

      I believe 3.1. We ran a regression to find the exponent that had the least error between expectation and win percentage for the 2011 and 2012 seasons, and that was best for us.

  3. BLL513 says:

    High Point beat Jacksonville 8-7 and HPU’s record is 7-3 now.

  4. Phil says:

    How did you go about calculating SOS and Pace? Also, out of curiosity, how did you Ride%? Love the site!

    • Tim says:

      SOS is average Pyth of opponents played. Pace is total possessions (offense and defense) per game. Ride% is opponents’ failed clear percentage.

  5. Brandon Cyrus says:

    I am a big fan of advanced statistics, Dean Oliver, and Kenpom. I want to know what are and what are the formulas for eShooting % and EMO Reliance?

  6. Brandon Cyrus says:

    How do you determine eShooting %?

    • Tim says:

      I honestly have no idea how this was derived (you’d have to ask Hoya Suxa at College Crosse), but the formula I find says…

      ((.167*EMO Goals)+(Total Goals)+(.83*Man-down goals))/Total Shots.

  7. phillip pfeifer says:

    DEFINING POSSESSIONS. If the team on offensive loses the ball but gets it back all in its offensive end, that counts as 2 possessions for the offensive team and one for the defensive team. Right? Another approach would be to just ignore any back and fourth in one end of the field and say possession changes only if the ball gets advanced to the other end. So this proposed definition of possessions would be FO wins + successful clears. Comment?

Leave a Reply

Your email address will not be published. Required fields are marked *