I'VE HAD JUST ABOUT ENOUGH OF YOU SONNY
If you’re looking for the weekly Corsi charts, you’ve come to the right place. Incidentally, they aren’t here, so you’re also in the wrong place. It’s not you, it’s me.
I’ve been busy finishing up work on my thesis, an additional research paper, and other assorted family things (hence the gap between posts), but I haven’t been completely ignoring hockey. I try to find as much to read about advanced hockey statistics as possible, and I ran across an article on the Maple Leafs’ SBN blog that caused something of a crisis of conscience.
The last two columns are the r-value for even strength Corsi and the R2 value as it relates to winning percentage. The first column to the right of 5v5CF% is the R^2 value over six years of statistics that the folks at Pension Plan Puppets collected. Their analysis indicated that Corsi is both highly replicable and relatively highly correlated with winning percentage.
Visualizing r^2, via BMG Lab Tech
Having said that, I’ve decided to shelve the Corsi project for now. It may be true that Corsi is related to winning percentage, but B1G hockey is starting and I wrote back in my first Corsi post that I’d start doing goal-by-goal analysis posts again at this point. I decided to stop tracking Corsi data because I only have a limited amount of free time to pour into this each week, and if I’m going to spend 5-8 hours a week analyzing Michigan hockey then I think most people would rather see GBGA than something that may be highly correlated with winning but has very little aesthetic appeal. If you’ve been using my Corsi posts in place of Ambien then I’m glad I had the opportunity to both inform you and put you to sleep. I’ll try and get back to Corsi at some point, but I can’t make any promises.
If you’re now wondering what GBGA is then I think I’d describe it as a Picture Pages/UFR hybrid. I break down every goal for and against, hoping to explain what happened in a sport where important things happen in tenths of seconds and add something entertaining/informative to the world of Michigan hockey.
2:51- OSU 0 Michigan 1: Lynch from Kile & Allen
Michigan gets the puck in deep and Lynch goes to carry it behind the net. The two OSU defensemen should follow the routes drawn out on the screen shot; as one goes to cover Lynch on the wraparound, the other should go to the front of the net.
Instead, the right defenseman follows Lynch behind the net. This vacates the front of the net, which comes into play later. The left D has no choice but to leave Lynch and follow the pass to Kile.
A simple pass from Kile to Lynch puts the puck in a dangerous spot for OSU. Since the right D followed Lynch behind the net there’s no one to protect the net-front area. Lynch is going to get an easy wrap around shot. Look at how deep the OSU goaltender is in his net. He’ got the post locked down, but he’s still standing when Lynch gets the puck on his stick and you can see how open to five hole area is. Lynch puts this one away for Michigan’s first B1G conference goal.
15:46- OSU 1 Michigan 1: PPG Dzingel from McCormick & Szczechura
Michigan’s box is shifted far left. You can see that OSU has three players in the frame, which means that two are off screen and essentially undefended.
Motte makes a mistake and tries to block the shot in the slot. This really could fall to the defender in the front of the net, as Motte’s assignment is to cover the far right (where the arrow’s pointing, naturally).
Pretty obvious that this mistake leads to a really, really wide open shot. The goal itself was soft, as it just sort of rolls over Nagelvoort’s glove but the defensive breakdown is still key to this goal.
6:41- OSU 1 Michigan 2: PPG Compher from Moffatt & Guptill
The key here is that the top of OSU’s box has sagged down into the slot instead of staying high to cover the point. This allows Michigan to pass across the zone from the boards.
Moffatt takes the shot that’s there for him. It’s not a bad shot, but it’s not exactly a high percentage shot. The best case scenario is what happens, which is a big, uncleared rebound in front. The beauty of the power play is that Compher (circled above) is going to be undefended if the low defensemen doesn’t get there in time.
While it doesn’t quite work out the way I noted above, it still works out. The Michigan player essentially sets a pick, leaving Compher to backhand the puck into the really, really open half of the net. Also, OSU’s goalie Logan Davis is like whoa slow laterally.
HIGH FIVE METAL BARS I FEEL YOU
14:26- OSU 2 Michigan 2: PPG McCormick from Szczechura & Fritz
DeBlois takes away the passing lane to the blueline, so OSU works the puck down low.
Bennett tries to take away the pass to the slot but is about a half second too late, and the puck gets tipped through Nagelvoort’s legs.
19:43- OSU 2 Michigan 3: Guptill from Compher
Compher wins the faceoff, which is huge. Even more important, however, is that DeBlois is able to tie his man up. This allows a clean tap across from Compher to Guptill…
...and a very, very clean shot from Guptill. He lifts the puck perfectly, hitting the top corner before Davis knows what (didn’t) hit him.
16:47- OSU 3 Michigan 3: Greco from Fritz
Credit where credit’s due, OSU’s forechecking creates this goal. Michigan makes a bad decision to play the puck back, and there’s no Michigan skater to collect the weak pass. OSU gets there first and gains possession.
No one notices the skater in the slot until it’s too late; Downing was behind the net and doesn’t cover the front fast enough, Motte can’t catch him either, and Nagelvoort (who was locking down the post) can’t stop the wide-open slot.
Michigan’s offense-turned-defense on this play.
3:38- OSU 3 Michigan 4: Copp from Bennett & De Jong
Mac Bennett, man. He sees a huge passing lane and puts a perfect pass….
…on the stick of Andrew Copp.
He gathers, shoots, scores, and then this.
Look, I'm not naive. This is a Michigan blog in the middle of football season, and I understand that the vast majority of you are here to read about football. That's ok, I come here for that too. Having said that, I'm guessing most of us feel a lot like this .gif (which Ace posted on Twitter Sunday night). Most of us probably have visions of the interior offensive line standing by, stationary, like cops by the side of the freeway (Hey, that guy's going by us way too fast. He's gonna cause an accid...see? See, that's what I was saying).
Forget about all that for a minute and get lost in the glorious charts of a Michigan team ranked #2 in the latest USCHO poll. That's right, there's a Michigan team that's climbing up the polls! If they continue playing the way that they have (read: willing to battle for puck possession, realtively intelligent passing, good forechecking) then there are going to be a lot of people who find a more enjoyable way to spend their Saturday nights once football season is over.
Friday, October 25 vs. Boston University (W, 2-1)
Saturday, October 26 vs. UMass Lowell (L, 1-2)
Friday, November 1 vs. Michigan Tech (W, 3-2 OT)
I am Mario and MGoBlueTV is the castle. Someday I shall breach those daunting castle walls, but last Friday was not that day. In other words, no charts.
Saturday, November 2 vs. Michigan Tech (W, 2-1)
- Winning the Corsi battle correlates with winning in the three games charted. It does not, however, correlate with scoring margin. The caveat is the small sample size, which is why I'm charting everything I possible can; we'll see how this plays out over the season.
- Michigan has only won once (against Boston College) when they have had a worse possession percentage than their opponent in the first period. Maybe there's something to be said for starting strong.
- Michigan has looked flat in the third period too often this year. It cost them against UMass Lowell and it could have cost them against Michigan Tech on Saturday. This "flatness" is reflected in the possession numbers above.
- I'm collecting power play data too, but my plan is to publish that around the midseason point in a longer post.
So I'm finishing up the chart for the first game of the series on Wednesday night. I had Fox Sports Detroit on before I started watching the game on my DVR, which I thought nothing of because it's pretty typical for me to have Fox Sports Detroit on. After I finish charting, however, I hit stop on the recording and the Red Wings game comes on and PRETTY MUCH BURNS MY RETINAS WHY IS THIS ICE SO BRIGHT. Hockey in high def is most definitely not what Fox Sports College Atlantic or whatever it was was broadcasting for these Michigan games. Like Brian said in his post, it was basically like watching a legal stream. That wandering anecdote was my way of saying that the Corsi charts have some values that may not be perfect. I'm not really concerned though, as there are things that I may have marked a shot that missed the net or vice versa that even each other out in the overall numbers. This is why I don't change numbers in the shots column to match the official score sheet; it should work itself out when considering all of the categories.
Friday, 10/18/13 at UNH
- Things that are good: spending time not in the penalty box. Things that are bad: spending time in the penalty box. Of UNH's 37 shots 11 came on the power play. Of their misses, 7 of 21 came on the power play. Of UNH's blocked shots, 7 of 14 came on the power play. That's 29.7% of their shots, 33.3% of their misses, and 50% of their blocks. Worried about being outshot? Try to keep five guys on the ice.
- Even though Michigan was handily defeated in terms of possession, the game was a little more open than the numbers indicate. A lot of this game was played in the neutral zone, with each team trying and failing to create an offensive zone presence. An argument could be made, however, for UNH carrying the play because whoa that third period was rough. Michigan was circling and circling and circling in the defensive zone and I had to keep pausing the recording so I could mark more stuff down for UNH.
- UNH's numbers look pretty dominant in overtime but we're looking at a five minute sample. UNH had possession but I can't remember any part of OT where I thought Michigan dodged a bullet.
Saturday, 10/19/13 at UNH
- This was a game of extended periods of back-and-forth punctuated by furious bursts of offensive zone activity. Michigan looked like they were handily outplayed in the second period (especially the second half of the second period, where they couldn't clear the puck for anything) and the numbers bear this out. At one point UNH went on a tear of four consecutive shots on goal in what had to be under a minute. It looked like the tide of the game may have been turning, but the clock mercifully ran out.
- UNH picked up where they left off and generated a number of chances early in the third period. Their offensive zone time came in smaller bursts, however, and the game transitioned back to an up-and-down affair. Michigan eventually got their own extended offensive zone time and ended up with a slight advantage overall in Corsi for the period.
- Special teams were the theme of Friday night but weren't a factor after the first two periods on Saturday. In the first period 12 of Michigan's 21 shots (57.14%) came on the power play. In the second period 10 of UNH's 25 shots (40.00%) came on the power play. After that? Not even a power play opportunity for either team.
- I like Nagelvoort and his giant leg pads. They're like belly putters in golf; comedically oversized but effective in their own right.
“Bring me your tired, your poor,
Your huddled masses yearning to breathe free…”
Look, we need to talk. I’m worried about you. I mean, look at you. There are bags under your eyes, you’re pale, I mean…look, something just isn’t right. It’s the loss to Penn State, isn’t it? I haven’t seen you this upset after a loss since…well, since that other guy (you know, the one who wore the red wrist band) was still in town.
Guess what? I’ve got something to soothe your chapped and chafed sports-soul. It’s Michigan hockey. Weren’t expecting that after last year, were ya? Frankly, neither was I. And yet here we are, just two games into the new season and last year seems like a memory that we might just be able to get rid of and finally move on from. Last year was a memory burr; it hung around longer than it should have and felt like it might be impossible to get rid of, but once it was gone there was such sweet relief. See, a win over a Boston College squad that’s one of the top five in the country isn’t enough to do it alone. It’s the combination of not only winning but also showing defensive acumen and energy and situational awareness that’s so reassuring.
The official scoring sheet says that Michigan outshot BC over the course of the game, but let’s dig a little deeper. Below you’ll find Corsi tables, and if you aren’t sure what Corsi is read this post from last week.
Things didn't look good early on. Michigan was keeping up with Boston College, but that's about it. Granted, they did get a nice power play snipe from Luke Moffatt (whose performance was the focal point of Brian's excellent post), but that's about it. It seemed as though Michigan was doing all they could just to hang on for 20 minutes; hanging around for 60 is an entirely different thing, and winning that next 40 minutes is a different animal entirely.
Well, looks like the tables are starting to turn. Michigan and BC went back and forth this period, but Michigan looked better getting the puck out of their zone and moving it through the neutral zone. The shot totals above don't match the official scoring sheet, so it's worth noting the caveat that comes with this analysis; when the game is only available in standard def that's what I have to watch it in, and it can be hard to tell whether the puck hit the goalie or went just wide. Having said that, I do believe in what I saw and wrote down. We'll see how this takes shape over the course of the season.
The numbers end up looking like Michigan dominated the period, but that isn't how it began. BC carried the play in the first few minutes, and Michigan was allowing this to happen. They moved a forward high to defend and were dumping the puck in. Midway through the period this changed, and Michigan dropped the conservative schtick and started to move the puck again. One thing that went well: passing toa teammate on a zone entry and letting them carry the puck in instead of dumping it into the corner.
Michigan outplayed a higher ranked opponent by forechecking hard, backchecking hard, and making intelligent passes to keep the puck moving and away from the opponent. Michigan won. Read that again. Now do it again. Michigan (yes, that Michigan) beat an incredibly talented team and by game’s end made them look overmatched. Whether Michigan hockey is really back remains to be seen but this is certainly the only way to start.
As for overall shot percentages, I’ll let Seth handle that. He had a cool table in last week’s Dear Diary post and I don’t want to step on his toes because intellectual property, man. A quick programming note: I read your comments on the last article and haven’t given up on goal-by-goal analysis posts. I’m going to continue with this (because I think collecting the data over a full season will be worthwhile) but also start GBGA’s when the B1G season starts. Also, I can only do Corsi analyses for games that I have in full, so there won’t be one for the RIT game. Should be a Corsi post for this weekend's games, though, so look for that next week.
Think back to when you started reading mgoblog. Remeber that feeling of validation that there were indeed people out there like you, people who wanted to write about football in way that was funny but also disarmingly analytical? A site that values both memes and knowing the average yards per play for every formation Michigan has run so far this season is my kind of place. It's the reason I've been reading this site since 2006.
My objective for the coming hockey season is to add something empirical to the mix. I've always gravitated towards advanced stats in hockey, and for those of you who follow college hockey know all too well these types of statistics aren't readily available outside of the NHL. What I'm going to attempt to do is track Michigan's Corsi rating over the entire season.
Of course, Corsi is just one statistic (even if I'll break it down into a number of different components). The "big idea" behind Corsi is that you have to hold on to the puck to score, and that the team who does a better job of this has a better chance of winning. At the end of the day it provides some interesting insight into puck possession and could be useful for gauging the strength or weakness of special teams play, but it isn't a be-all-end-all stat. There are score effect problems, most notably that even strengh Corsi or Corsi from within one goal in the first and second periods is correlated with winning but it gets dicey in other situations. More on this later.
If you're looking for a nice overview of the statistic you can find that here.
What I need to know from you guys is whether you find it interesting and useful enough to continue tracking. If the fine folks of the MGoCommunity don't like it then I'll go back to writing up goal-by-goal analysis posts like these.
That's cool and all but this post is really boring me. No pictures yet? Come on, do you at least have charts?
Charts? This is mgoblog, fergodsake. Of course I have charts! Let's start by looking at things by period before looking at the bigger picture.
You don't have to know much about Corsi to see that Michigan carried the play in the first period. 72.9% of shot attempts came from the Wolverines, yet Waterloo ended up with the lone first period goal. Michigan made one mistake in defensive coverage in front of their own net and Waterloo took advantage. It's worth noting, however, that Michigan's Corsi total was bolstered by time on the power play (Waterloo didn't have one in the 1st). UM recorded five shots, seven missed shots, and five blocked shots over their two power plays.
Waterloo seemed to carry play in the second period and the Corsi numbers reflect that. Michigan's goal was something of a fluke, coming after Waterloo's goaltender badly misplayed the puck in front of his own net. Waterloo did get on the power play in the 2nd period, but they failed to register a shot. Their power play generated one missed shot and two blocked shots. Michigan's power play registered three shots, two missed shots, and two blocked shots.
Here's where the score effect problem I mentioned earlier comes into play. Waterloo was content to carry the puck into the neutral zone and play dump-and-chase in order to burn clock, and when you're playing that style a byproduct is a reduction in the number of shots you take. As you can see, that's certainly reflected in the numbers above.
Michigan had the edge in every category tracked here, yet they couldn't convert opportunities into results. I think that the reason for this lies in the type of shots Michigan was taking; most were from the perimeter, and perimeter shots are much easier for a goaltender to stop than shots through traffic. It will be interesting to see how the possession game plays out against BC, a team with a notoriously stringent defense.
Don't you usually draw on screencaps or something? Why are there all these charts?
Yeah, I call the screencap thing goal-by-goal analysis. I'm not set on moving away from that completely, but I want to know if people find the info above interesting. Like it and I'll keep tracking it, hate it and I'll go back to GBGA.
This piece was going to start with something about how fall is here and evoke images of pumpkin pie or pumpkin spice lattes but there’s a thread on the board about hockey practice jerseys that has double-digit comments so let’s get to the point; it’s almost hockey season. Also, you’re probably envisioning pumpkin pie or a pumpkin spice latte now anyway so mission: accomplished.
Twenty two days remain before the puck drops on the 2013-14 season, and even though it might seem like a good time to do a season preview* I’m not done writing about last year. There are two things that I could write about ad infinitum; applying advanced stats to college hockey and what went wrong during the 2012-13 season.This might as well be a Sports Illustrated expose because there just isn’t going to be a smoking gun that alerts us to the singular reason behind the 2012-13 collapse, but with advanced stats in the fold there are new ways to look at what happened.
I recently read about applying baseball’s quality start statistic to goaltenders (you can read more here and here). A lot of blame was heaped on Michigan’s trio of netminders last season, and while some of it may have been understandably levied I still believe that an unwarranted amount of criticism was given to the guys between the pipes and not enough was placed on those standing in front of them.
Quality starts are worth looking at because they are highly correlated with winning. At the NHL level, a team wins 77.5% of the games in which their goaltender recorded a quality start. There are, of course, some flaws to the statistics. First and foremost, there’s still not a good way to control for a poor defensive team. If a team gives up lots of breakaways and odd-man rushes then the goaltender’s save percentage is still likely to suffer. At the same time, quality starts are useful from a consistency standpoint. If a goalie fluctuates between shutting out teams and blowing up and allowing five goals then they typically won’t record as many quality starts as a goalie who consistently goes out and allows two or three goals per game.
A goaltender is awarded a quality start if they 1.) start a game and 2.) have a save percentage that is above the league average (or, in this case, above the DI average). In the context of a portion of a season we aren’t going to escape the problem of small sample size, but there are some nuances to what happened last year that we can glean from the stats available. As you can imagine save percentage and quality starts are closely linked. Steven Racine’s 89.9% save percent wasn’t exactly a revelation, but at the same time his improvement over the season did get Michigan within one game of another NCAA Tournament berth. This is where not just quality saves but the associated statistics become worth investigating and discussing.
A cheap win occurs when a goaltender records a save percent below the DI average but his team wins anyway. Conversely, a tough loss (or wasted quality start) is granted if a goaltender has a save percentage at or above the DI average but his team loses.
[QS= Quality Start, NQS= Non-Quality Start, CW= Cheap Win, WQS= Wasted Quality Start]
Steven Racine’s 2012-13 stats are featured in the table above. We’re going to look at his starts because he’s the only goalie on Michigan’s roster that started more than 15 games, and 15 starts was the cutoff for having one’s save percentage counted towards the DI national average. The DI national average turned out to be 90.2%, with 76 goaltenders being included in the calculation. As a quick aside, three of Adam Janecyk’s nine starts were quality starts and three of Jared Rutledge’s nine starts were quality starts. Having only 33.3% of your starts qualify as quality starts is just bad; the NHL standards at Hockey Prospectus state that a quality start percentage of 40% or below is considered very poor.
Racine fared better than his counterparts, with 12 QS out of his 22 starts. That means that 54.5% of Racine’s starts were quality starts, putting him relatively close to the 60% QS that Hockey Prospectus considers elite. While Racine’s QS numbers weren’t all obtained during the team’s nine game win streak that put them in the CCHA Championship game he definitely had a statistically better latter half of the season. Racine recorded five QS in his first twelve starts and seven QS is his last ten. Again, it’s hard to tell whether the shift from 41.6% of starts being QS to 70% of starts being QS is due to Racine settling in and adapting to the speed and angles of the college game or whether it’s because the team defense buckled down and started, like, defending but there’s no question that the increase in QS% was huge.
Five cheap wins in 22 starts means that 22.7% of the time Racine started in net he didn’t have a SV% that was at or above the DI average, with three of those cheap wins coming after the team’s nine game win streak started on February 22nd. That doesn’t surprise me that much, as you’d expect that a Michigan goalie is going to get some cheap wins when the offense is ranked seventh in the nation in goals per game.
Only two of Racine’s 22 starts were wasted quality starts, which is also indicative of having a good offense; in only 9% of his starts did Racine or the team perform well defensively only to watch the game slip away because they couldn’t put the puck in the net.
Tl;dr. What’s your point?
Michigan’s goaltending wasn’t as bad as it may have seemed, with 54.5% of Steven Racine’s starts being quality starts. The percent may seem low until you find out that having a 60% QS% is considered elite in the NHL. If Michigan can get their forwards to backcheck and forecheck and if the defense corps isn’t a punch of pylons or rovers then the 2013-14 season should go better than 2012-13.
While there are some obvious flaws to the QS stat, it should be interesting to track over multiple seasons. There will always be the huge caveat of shot quality being untrackable, but it’s definitely a better (and both more nuanced and interesting) stat than *shudder* wins.
*If you are looking for a season preview then perhaps ordering Hail to Hoops and Hockey would be a good idea. You get an actual preview from Brian, as well as an article by me about what went wrong last season. Wait, what do you mean there’s a theme to what I write?