Before analytics became in fashion, I bet that clutch and popular were basically one and the same. Players, across all sports, didn’t have the same sort of access to fans that they do today through video and social media. So perceptions about a player revolved around two things: what was written about them, and whatever the most notable things they did on the field were.
I would argue that what drove the former was the latter. In other words, you got mentioned in the media if you did something notable, and the media coverage was probably driven by whether what you did was good or bad, but equally, when you did it. Make a mistake in the first inning, who cares. Go 1 for 5 with four Ks, but have the 1 be a walk-off home run: hero.
There was no WAR, there was much less reliance on rate stats, and we didn’t have people tearing apart video to find the nuggets of value like we have today. Today we have arguments about players like Derek Jeter, who check every box on the clutch side of the list. Are they really great, or did their greatest moments drown out other, more mediocre attributes? The opposite side of that spectrum is a player like Tim Raines, who had all the WAR you could ever want but never had the spotlight to shine under. (Ok, enough baseball. This is a lacrosse site, right?)
As a fan of sports, I still think that clutch matters. Fandom is about belonging and nothing fosters those bonds more than shared memories. And memories are created in the moments that would be considered clutch. Unless your job is to build a roster in some capacity, then the cumulative value of a player’s entire contribution is simply less important. Clutch is cool.
That said, I think every fan who has an interest in analytics also moonlights as an armchair GM, so we care about that stuff too. We like to able to point out that: “of course player x had more (enter counting stat here), they had more opportunities to amass them!” Naturally, I couldn’t resist the urge to do some clutch vs non-clutch analysis on last year’s game data.
The process was fairly straightforward. We looked up every goal scored in 2017, and we compared it against the team’s odds of winning the game before and after the goal. A clutch goal is, by definition, one that raises the team’s odds of winning the game by a certain threshold. This naturally favors goals scored toward the end of close games and goals scored by players whose teams are in dire straits. In other words, as you should expect, “clutch goals” are goals that come just when a team could really use one. (See the end of the post for our specific numerical criteria).
We limited our population to any player who scored at least 15 goals last season. And when the table below says that Nick Mariano added 10.8 percentage points of win probability per goal, that means that his average goal increased the Cuse’s odds of winning a game by about 11%.
|Player||Team||Goals||Clutch Goals||Avg Win Prob|
|Julian Garritano||Sacred Heart||20||10||11.6%|
|Sergio Perkovic||Notre Dame||23||8||10.9%|
|Jon Vogel||Holy Cross||22||9||9.7%|
|Jack Mangan||Mount St Marys||15||6||9.2%|
How about Syracuse? They’ve got 3 guys that appear in these tables (Mariano, Sergio Salcido, and Jordan Evans down below). To an extent, this speaks to the ability of the Syracuse playmakers to come up big in big situations. If they weren’t finding the net in these high-leverage situations, they wouldn’t be showing up here.
But of course, it also speaks to the sheer volume of high-leverage situations that Syracuse found themselves in (remember those 11 one-goal games?) They won 9 of those 11 games of course, and it’s in large part to the fact that these three guys scored all these “clutch” goals. (Note: of course, a goal is not just the effort of one guy, so don’t read this as ignoring the efforts of the players that led to these goals.) If that seems a bit circular, it is, but it’s like the old adage: “Luck Is What Happens When Preparation Meets Opportunity”
You only get on this list if your goals are consistently coming in important spots. Going back to the initial premise of this post, clutch is cool, although maybe not especially relevant from a true analytics perspective. A goal in the last second of the game is worth just as much as a goal off the opening face off. But the reason that this is interesting is that, undoubtedly, there is a skill, however intangible, in being able to perform in the highest intensity situations.
Shout out to some of the players on this list who have tended to play in relative obscurity: Julian Garritano, Brian Prunty, Colin Burke, Jon Vogel, Jack Mangan, Tyler Keen. These names are probably not going to become household because of the stakes that these teams are playing for. But their coaches, fans and teammates know that they scored some big goals last year; and that is always worthy of a paragraph.
The Cupcake Challenge
This analysis seems somewhat unfavorable for a guy like Connor Fields, who appears in our third table. This view shows the top goal scorers along with the number of “clutch” goals that they registered on the year. Playing for Albany, you just weren’t going to have a lot of high-leverage situations. This chart might be more interesting as a scatter plot where the number of goals is plotted against the average win probability increase. But I suspect you’d just see lots of second tier conference guys on one end and top-tier conference guys on the other.
|Player||Team||Goals||Clutch Goals||Avg Win Prob|
|Mac O’Keefe||Penn State||51||13||3.3%|
|Tre Leclaire||Ohio State||50||14||6.8%|
But I think that if there is one take away from this table, it’s that scoring goals is not the same as winning games. Since voters are humans, having a minimum number of goals that shifted games from a loss to a win is the type of thing that people look at when casting their ballots. It’s easy to look at a season where 2 out of 55 goals are classified as clutch and chalk the success up to the lack of competition. Not saying that is fair; but it happens.
One Shining Moment
I think of our fourth table as the statistical equivalent to the “One Shining Moment” montage at the end of the NCAA basketball tournament. These are the players who had the highest ratio of “clutch” goals to total goals. In other words, when they scored, it mattered. Will McCarthy of Lafayette is the best example: of his 15 goals, 8 registered as “clutch”.
|Player||Team||Goals||Clutch Goals||Clutch Pct|
|Julian Garritano||Sacred Heart||20||10||50%|
|Connor Robinson||High Point||16||7||43%|
You could build a narrative around this table. These are the guys who maybe do not see themselves as first options, but when the game is tight, and their team absolutely needs a goal, they step up and take that chance.
That is probably overly romantic, but it’s a nice thought. Since this post is our compromise between breathless fandom and JS-Mill-style roster building, I’ll allow it.
Appendix: Calculating Clutch
This really is a very simple rule of thumb. A goal was considered “clutch” if the scoring player’s team saw at least a 5% increase in win probability because of their goal. To calculate this, we used our normal win odds model to calculate the chance of each team winning at every point in the game. When there was a goal scored, we looked at the last quoted win odds, subtracted the post-goal win odds, and if the difference was greater than 5%, we marked the goal as clutch.
If you think about the math here, there are a few ways that a goal could satisfy these criteria. Also several ways that won’t work. A goal scored in the first quarter is very unlikely to qualify as clutch because the win odds calculation weights the relative strengths of the teams more at the beginning of the game. Game situation and score become more strongly weighted as the game progresses. So a first quarter goal may bump the raw chances of winning by more than 5%, but since game situation is still less than 25% of the calc, it would have to increase the raw win odds by 20% to qualify.
Also, if a player’s team is way ahead or way behind, it’s going to be hard for a single goal to move the win odds much.
In fact, if we check how many “clutch” goals were scored by quarter, this bears out:
- Q1: 10
- Q2: 577
- Q3: 989
- Q4: 973