[Coller/MGoBlog]
Most old posts are embarrassing. My takeaway in reading through old work for research purposes is usually some stupid line that I wish I had framed a different way or a dumb joke that I forgot I made and spend the rest of the day regretting. Occasionally, though, I’ll read something that makes me feel exactly what I was feeling when I wrote it.
When I dug through old Goal-by-Goal Analyses featuring Michigan Tech I came across the mini-column I wrote at the bottom of this year’s Great Lakes Invitational post and felt the still-too-familiar raging bewilderment that marked much of the 2016-17 Michigan hockey season. The piece ended with boiled-over frustration about Michigan’s offense and their inability to get the puck in the zone; there’s a mention of how badly Michigan was out-attempted, but the Corsi hamblasting from Tech wasn’t unusual enough to garner anything more than an unfeeling, fleeting mention.
furthest right column is Tech’s Corsi For %
That’s Warner Bros. DC movie-level destruction; that’s also reflective of how difficult to stomach the 2016-17 Wolverines could be. If Mel Pearson’s Michigan Tech teams are any evidence, however, Michigan fans are in for a rebalancing in their squad’s putrid possession numbers. College Hockey News has some advanced stats available from 2013-14 on, including Corsi. Remember that Corsi is every shot attempted: shots on goal, shots that missed the net, blocked shots. (There’s a more complete primer at the bottomr of the post.) The general idea behind this is that a team has to have the puck to shoot it, so Corsi is a puck-possession proxy.
thanks, Seth
Generally speaking, Pearson’s teams were good against very good teams and great against bad teams, at least in terms of possession. You can see from the trend line that they did exactly what you’d want a team to do against teams outside the PairWise top 16; the trendline drops below 50% on the doorstep of teams that would make the tournament.
[After THE JUMP: looking for surprises in Tech’s possession numbers]
From there, I thought it would help to look at how Pearson’s teams did against different tiers of competition. On the y-axis of the below charts, Green corresponds to teams that finished in the PairWise top 20, yellow corresponds to teams that finished 21-40, and red corresponds to teams that finished 41-60. As for the bars themselves, “Clear Win” indicates MTU possession % above 55%, “Neutral” indicates MTU possession % between 46% and 54%, and “Clear Loss” indicates MTU possession % from 45% on down.
If you’re interested in taking a look at the original data that David Nasternak put together, you can find the spreadsheet here.
The good, the meh, and the ugly
Tech didn’t face many bad teams, but at least they never allowed one to carry the game. Their most notable stretch of the season, Pearson’s third, came in January against Ferris State, which finished fifth in PairWise. Tech travelled there and then played them at home the next weekend; they notched two clear possession wins and two neutral possession games, recording a season-high possession% of 71% in the second game of the road series. For what it’s worth, Tech’s two games that were charted as neutral had possession%s of 53%, which obviously gives them a slight edge in possession in both games and the two series.
Weird fact: In 2013-14, there was not a single series in which Tech didn’t improve their Corsi the second night of a series. It seems Pearson’s staff was adept at making overnight adjustments and the players did well implementing them, though this resulting possession boost wasn’t repeated over the next three seasons.
Tech played better and worse teams more often in 2014-15, though the song remained largely the same. They kept the puck away from bad teams and saw mixed results against the best teams in the country. By virtue of being a fellow WCHA member, Tech played Minnesota State, the team that finished #1 in PWR, five times in 2014-15. They twice finished with possession% of just 43%; they also posted a possession% of 53% and 55%.
The one series against a green-tier team that Tech clearly lost: Michigan. Tech’s possession% in an early-season two-game home stand against Red’s squad was 46% in game one and 37% in game two. That would have stood as their worst possession% of the season if not for an oddly awful 34% in a November road game against middling Bemijdi State (27th in PWR).
There’s not much we can say about how good Tech really was in 2015-16 thanks to a very weak WCHA. The two contests they played against quality opponents that season came in one-off neutral site contests against Yale (7th in PWR) and Michigan (10th in PWR). Tech did an admirable job, though, with a 51% possession% against Yale and a 57% possession% against Michigan.
The two games that stand out for the wrong reasons came against Alabama-Huntsville (a putrid 56th in PWR) and Lake Superior State (45th in PWR). Tech possessed the puck an abysmal 41% of their first game against Alabama-Huntsville, though they rebounded the next night and possessed the puck 60% of the game. Things were inverted against LSSU, as Tech had an excellent 62% possession% in the first game of their series before putting up a weak 42% the next night. That game reads like a fluke.
Tech started the season with a clear possession loss and a neutral game against Minnesota-Duluth, which ended up second in PairWise; the clear loss still featured a 44% possession% and Tech out-attempted by six shots, which is relatively respectable. Over the course of the season Tech posted more neutral possession games against bad teams than they had in the past four seasons; this pattern held across tiers as well.
They saved their best performance for the NCAA Tournament, where they out-attempted eventual national champion Denver by one and finished with a 51% possession%.
What do these newfangled fancystats mean?
Michigan’s puck possession is about to increase, and it’s possible that the increase is going to be dramatic. We don’t have data for Michigan Tech before Pearson got there, but we can look at the rosters with which he obtained the above results for additional context. Those rosters had three (‘13-14), four (‘16-17), or five (‘14-15, ‘15-16) NHL draft picks; Pearson’s about to inherit a roster with nine such picks, and that seems low for a Michigan team.
Pearson parlayed said roster with only four NHL Draft picks into a team with the fourth-highest CorsiFor% in the nation (56.5%). Tech also finished fifth overall in 2015-16 (56.0%), third overall in 2014-15 (55.4%). For reference, Michigan’s roster had eight draft picks in 2016-17 and posted a final CorsiFor% of 42.8%.
Puck possession isn’t perfectly correlated with winning, but good teams tend to hold the puck longer; looking through the list, Denver finished sixth, Penn State first, North Dakota second, and Quinnipiac third. Michigan won 13 games last season while Tech won 23, Denver won 33 and a national title, Penn State won 25, and Quinnipiac won 23. There’s little chance that Pearson is preaching Corsi itself to his team, but his system has churned out the common byproduct of the systems run by some of college hockey’s blue bloods: puck possession (and wins, and tournament appearances).
This is an excellent advanced stats primer from RMNB that you might recognize from Brian’s Monday post on Pearson