Each week, we are going to highlight some of the sneaky-hard matchups that ranked teams face. To us, bad matchups are not just about going up against great offenses or stifling defenses. Everyone knows that Connor Fields makes Albany a bad matchup, no matter who they play. Highlighting those isn’t especially interesting.
Instead, we are going to focus on more subtle challenges. If a top team gets a lot of offensive opportunities from unsettled situations off faceoff wins and they are facing a team that is very good in transition defense, then even if the matchup overall doesn’t seem threatening, that facet of the game could cause problems and make the game tighter than expected.
To be clear, these are not predictions, just statistical curiosities that could mean that the match up is a bit tougher than some might expect. And if any of these games do result in closer than expected results, these clues might help explain what happened.
Notre Dame is a tough match up for anyone, but #3 Duke is especially impacted because Duke has not been as efficient when they play slower than average. And Notre Dame is a team that has forced teams to be more deliberate than perhaps they’d like. And indeed, you can see that in the first matchup between these two, Notre Dame forced Duke into their slowest pace of the season, averaging ~55 seconds before they took their first shot on a given possession.
Duke is favored to win this one (and we aren’t accounting for Danowski May Magic), so it’s still probably unlikely that this will tip the scales the other direction. But given the history between these two teams, you look for things that give the slightest edge, and this is one example where it leans in Notre Dame’s favor.
Towson is not a good match up for #10 Hofstra because Hofstra’s toughest games have been those where they’ve been limited in their transition opportunities off of faceoffs. Bad news is that Towson has been #5 in the nation in making teams less efficient than normal in these situations. So far, Loyola and Hopkins have been the only teams that have been able to play above their average efficiency in these situations. The fact that Denver and especially Ohio State were held in check should be troubling for Hofstra’s staff.
It’s worth noting that if you look at the win odds for this game, based on ELO, it’s pretty close to a toss-up as is. Perhaps this subtle advantage will push Towson over the top. But Hofstra has made a lot of people look silly this year, so who knows.
Cornell is not a good match up for #13 Princeton because Princeton struggles when they take a quick shot after making a defensive stop, and in this one, they are going up Cornell, a team that has forced teams to do just that. In 4 losses, they’ve attacked much more quickly after defensive stops, taking a first shot after an average of only 31 seconds. In their wins, they’ve been more deliberate, averaging nearly 45 seconds before the first shot in these situations.
And it’s not as if the results have been the same. In those 4 losses, their offensive efficiency after a defensive stop is only 24.7%. In the wins, it’s 34.8%. That has worked out to nearly a full goal less per game in more chances.
Princeton is a fairly significant favorite to win this one, so the small advantage that Cornell may have is likely not going to result in a W. But this column is about highlighting games that could be closer than expected, and this is one of those cases. And if Princeton tries to attack quickly in those situations, they could be running themselves out of some good opportunities and into a tighter game than they want.
Re Princeton: I think you’re confusing cause & effect. Princeton went quickly in games they were losing because they needed to catch up! A better vertical axis, IMO, would be time-averaged score differential; then I’d expect a positive correlation for all teams.
This is a great suggestion. You are right that the current approach will include garbage time possessions that may or may not reflect the team’s strategy. I created a second plot for Princeton that only includes possessions where either A) it’s after halftime and the score is tied or B) it’s before halftime and the score is within 1. The idea was to restrict the chart to only possessions where Princeton wouldn’t be impacted by the time/score so that the “strategy” could come through.
In short, the Tiger losses are still the games where their average first shot came with fewer seconds of the possession having elapsed. I’ll continue to refine this analysis, but thanks again for the suggestion.
April 28, 2017 @ 4:09 pm
Re Princeton: I think you’re confusing cause & effect. Princeton went quickly in games they were losing because they needed to catch up! A better vertical axis, IMO, would be time-averaged score differential; then I’d expect a positive correlation for all teams.
April 28, 2017 @ 4:31 pm
Even better would be goal differential at that moment vs time to first shot.
May 1, 2017 @ 8:43 pm
This is a great suggestion. You are right that the current approach will include garbage time possessions that may or may not reflect the team’s strategy. I created a second plot for Princeton that only includes possessions where either A) it’s after halftime and the score is tied or B) it’s before halftime and the score is within 1. The idea was to restrict the chart to only possessions where Princeton wouldn’t be impacted by the time/score so that the “strategy” could come through.
In short, the Tiger losses are still the games where their average first shot came with fewer seconds of the possession having elapsed. I’ll continue to refine this analysis, but thanks again for the suggestion.