On Faceoffs, Possession Margin, Turnovers, and Winning Percentage

None of what follows implicitly or explicitly asserts that faceoff play doesn't matter. Faceoff production does matter -- it's an aspect of the game that impacts results. Teams want to be strong at the dot, and they should work to dominate draws. The approach here, instead, is to look at faceoff play and understand a little more about how faceoff play is affecting overall winning percentage. This work obviously builds off of Michael Mauboussin's study a few years ago, and while it is not as complete as Mauboussin's work, it does reaffirm some relatively unique thoughts about how delineated elements of the game impact wins production.

How many times have you looked at a boxscore and thought, "Team A won 17-20 faceoffs; there's the game right there"? How many times were you bashed in the skull for thinking that?

A faceoff win does not change the count on the scoreboard. Rather, something needs to happen after a faceoff happens in order for the scoreboard to change. A faceoff victory or loss is merely the marking of an offensive or defensive opportunity, and what happens on that opportunity can -- in important ways -- dictate the volition of a result. This intuitively makes sense: The mere existence of possession is useless without something accruing from that possession. This is exactly why the correlation between faceoff percentage and winning percentage resembles a pretty untidy fit:

In a moment -- or a series of moments -- faceoff play carries heavy weight: A team may need to win a faceoff to stop the bleeding or to earn possession to run out the clock or any other reason that one can surmise. But a faceoff result is merely the impetus for what a team needs to do: Score a goal, get a defensive stop, etc. It's the run of play that impacts winning percentage much more than mere faceoff percentage. As a potential key to identifying the reasons for a team's success or failure, there are better places to look than only faceoff percentage.

If the preceding thoughts about faceoff percentage and its correlation to winning percentage rings hollow in your head melon, you're probably of the mindset of "But possessions have to matter to a team's ability to generate wins!"

First -- you're right. Second -- are you thinking about possession-generation as occurring only through faceoff victories?

There are lots of ways to generate an offensive opportunity: Faceoff victory, defensive stop that leads to a clearing attempt, an opponent failed clear, opponent turnovers in non-clearing or non-riding postures, etc. In fact, a huge chunk of a team's offensive opportunities come from non-faceoff scenarios: In 2014, Division I teams averaged 109.78 more total clearing attempts -- both for and against -- over the course of the season than total faceoff postures. Most of the game happens outside of the faceoff dot, and those aspects of play contribute in important ways to a team's overall possession profile. And that's important because possession margin holds a tighter correlation to a team's winning percentage than faceoff percentage:

This is important for a number of reasons, but mostly because of what's implied in this information. For example, a clear doesn't happen unless a team makes a defensive stop, and possession margin won't widen unless a team converts that clear into a bucket. The dependent and independent variables here do not exist in a vacuum, and that's especially important when considering the "FACEOFFS MATTER MORE THAN LIFE ITSELF!" faction of the college lacrosse universe.

There is all kinds of valuable information in a boxscore, but it baffles me that people hinge their focus so heavily on faceoff percentage when there are other pieces of information in the boxscore that are potentially more useful in understanding how a game unfolded. One of those pieces of information is turnover margin, although I come to it a different way than how it appears in a boxscore. Instead of using per game turnover margin, I look at turnover margin on a per estimated opportunity basis, thereby erasing some of the noise possession margin creates. Of the three correlations in this piece, turnover margin on an estimated per possession basis has the tightest fit:

This dovetails nicely with the preceding words on faceoffs: An opportunity, on its own, is only that -- an opportunity. How teams maximize opportunities has more importance than simple opportunity generation. A team wants to create as many opportunities as possible to both create circumstances conducive to victory and to deny an opponent an opportunity to create inverse circumstances, but those opportunities hold greater worth when they are maximized in an efficient manner. Turnovers are the antithesis to efficiency, a signpost for misery. It, like all other considered items herein, isn't a silver bullet, but it is indicative -- to a degree that is greater than other annotated items -- of a team's ability to generate victories. It's not just having the bean that's important; it's what you do with the bean when you have it. The question here is straightforward: Is a team better than its opponent at limiting mistakes and taking advantage of what they've created? That just makes sense, and the math is supportive of that principle.

And turnover margin doesn't exist in a vacuum, either. It contributes to possession margin, and possession margin has a relationship to a team's winning percentage. There are so many pieces that build into a team's ability to win that holding strong to one metric dictating correlative results is insane, which is why the overabundance of words dedicated to faceoff outcomes is somewhat odd. (As an aside: This Elo-style method of ranking faceoff men is really cool. Again: I'm not trying to diminish the important of faceoff play. Rather, I'm trying to offer some words on how faceoff play isn't an unimpeded Rosetta Stone for win-generation. )

Here are some brief takeaways:

• Faceoff play isn't the reason a team wins or loses, but it a reason -- in connection with other reasons -- that team may win or lose. Broaden your horizons, even if in isolated scenarios a faceoff can help dictate the result of a game.
• The metric with the strongest correlation to winning percentage is expected winning percentage, and that accounts for a team's efficiency in scoring and preventing goals and how it creates possession margin. As a baseline consideration, expected winning percentage illustrates that no metric exists on its own as the reason a team wins or loses (e.g., what builds into offensive efficiency? what builds into defensive efficiency? what builds into possession margin?).
• If your preferred team is either exceptional or disastrous at faceoff play, ask yourself this: What else is going on with my team that is helping to drive my team's winning percentage? Nothing exists in a vacuum. Bad teams are bad because they're bad in a lot of different ways, and good teams are good because they're good in a lot of different ways. Don't silo your ability to understand what's happening on the field.