Corsi WOWY, part 3: John Tavares' roller coaster rookie season
In case you missed it, I posted a little update in the comments of part 2. After finally having figured out all tools of Vic Ferrari's sites, I was able to deliver the exact numbers for the discussed line combinations. The conclusions remain the same, though, Bergenheim put up the most impressive numbers together with Nielsen and Hunter.
In this part 3 I'll try to get back from line combinations to individual players. I first wanted to include all the centers, but after having put something together about John Tavares, it's enough for now, I'd say. I'll be back with a summary about the rest of the centers - Rob Schremp Hockey, Bailey and Park that is, as I already looked at the brilliant numbers of Frans Nielsen in part 1, where you can also find what this is all about and how to read the tables, etc.
The centers are a very interesting topic for the Islanders right now. It's still not quite clear who of the young guys will turn out to fill what kind of role in the future. We saw a lot of them of course and more or less know what their roles were last season, but it might help to look at some numbers to get a better idea of how well they really did. Or in this part for example how bad JT exactly was during that big slump.
On the one hand, centers are easy to look at, as we pretty quickly know what some numbers mean. If you look at how certain wingers (e.g. Moulson) did without a certain center (JT), we can pretty easily tell that these numbers correspond to what Moulson did with Nielsen, as Moulson played almost exclusively with either of these two. But then again, some numbers are tough to interpret, as the centers naturally all fill different roles.
However, as we all know, JT had a bit of a roller coaster rookie season. Looking at the productive part at the beginning, the big slump in the middle and the good finish separately might be most telling in his case therefore. You can find the respective tables below, but let's first still look at his season totals. The table includes all wingers who had at least 100 Corsi events in total when on the ice together with JT. The entry at the bottom shows Tavares' overall numbers.
|
|
with Tavares |
without Tavares |
Corsi % |
||
|
|
Total |
Corsi % |
Total |
Corsi % |
WOWY |
|
Sim |
116 |
0.474 |
1444 |
0.443 |
7.14% |
|
Tambellini |
103 |
0.427 |
585 |
0.412 |
3.69% |
|
Bergenheim |
239 |
0.498 |
1265 |
0.492 |
1.10% |
|
Park |
210 |
0.433 |
1741 |
0.430 |
0.73% |
|
Bailey |
268 |
0.481 |
1450 |
0.478 |
0.71% |
|
Okposo |
909 |
0.491 |
1180 |
0.511 |
-3.99% |
|
Moulson |
1222 |
0.494 |
867 |
0.516 |
-4.13% |
|
Comeau |
642 |
0.474 |
818 |
0.496 |
-4.60% |
|
Hunter |
211 |
0.493 |
1247 |
0.527 |
-6.45% |
|
Tavares |
2120 |
0.479 |
|
|
|
As mentioned, when looking at these numbers we always have to consider who the given players were on the ice against and in what kind of situations they were used. Tavares' role was about the same all season long. He centered the most offensive line, shown also by the fact that he had by far the most favourable zone start of the team. He had tough assignments, too, but also played a lot with the best wingers of the team. Overall, a team would normally certainly expect such a line to outshoot and outscore the opposition.
But as we can see in the left column, Tavares was unable to outshoot the opposition no matter who he was on the ice with. He was very close at least when on with the good wingers, but still, these are not very impressive numbers at all. And the right column shows that the best wingers of the team did better when on with other centers. Okposo, Moulson and Hunter actually outshot the opposition when on without Tavares despite in that time mostly playing against tougher competition (with Nielsen particularly) or starting more often in the own zone (with everyone, particularly Park and Nielsen) or being on a line with a worse 3rd forward (particularly combinations centered by Schremp, including perhaps Sim, Tambellini). The only two forwards who benefited from playing on a line with Tavares were Sim and Tambellini. But they didn't play a lot with him, anyway still struggled and mainly were better with Tavares because they were horrible in less offensive roles.
The promising start
I split up Tavares' season in three parts. The first one includes all games from opening night until the game against Boston on Dec 12th. His slump started after that with the game against Florida and despite some improvement in January lasted until the Olympic break - that's the 2nd part. All games after the Olympics are included in the 3rd part. Here are the numbers of the first part:
|
|
with Tavares |
without Tavares |
Corsi % |
||
|
|
Total |
Corsi % |
Total |
Corsi % |
WOWY |
|
Moulson |
687 |
0.483 |
121 |
0.364 |
32.90% |
|
Hunter |
87 |
0.563 |
307 |
0.440 |
28.08% |
|
Okposo |
440 |
0.511 |
373 |
0.464 |
10.25% |
|
Park |
138 |
0.449 |
663 |
0.410 |
9.51% |
|
Tavares |
814 |
0.482 |
|
|
|
As it's a smaller time frame, I included Hunter, too, although he was on with Tavares for fewer than 100 Corsi events in total. JT was on most with Moulson and Okposo of course and they did quite well indeed. As said, we'd expect this line to outshoot the opposition given the situations they were used in. The numbers are nothing too particular therefore, but quite good anyway. Moulson-Tavares-Park seemed to struggle, though. I liked the idea of putting them together for some games back then, but it looks like it wasn't the best idea...
The big slump
Includes all games from Dec 14th against Florida until the Olympic break.
|
|
with Tavares |
without Tavares |
Corsi % |
||
|
|
Total |
Corsi % |
Total |
Corsi % |
WOWY |
|
Bailey |
209 |
0.498 |
418 |
0.519 |
-4.15% |
|
Hunter |
77 |
0.519 |
644 |
0.567 |
-8.34% |
|
Comeau |
348 |
0.443 |
381 |
0.520 |
-14.85% |
|
Okposo |
349 |
0.447 |
454 |
0.553 |
-19.15% |
|
Moulson |
325 |
0.452 |
454 |
0.581 |
-22.22% |
|
Tavares |
750 |
0.459 |
|
|
|
That looks ugly. Again the best Corsi ratio with Hunter, but again a very small sample there. And everything else doesn't look good at all. I can't tell for sure, but in order to get JT going again, he probably was used in slightly easier situations even than early in the season and that makes the numbers look a bit worse even. They're particular poor, as his line mates all seemed to be in very good shape. They all put up some impressive numbers when playing with other centers. Even Comeau probably played some good hockey back then already, but was kind of prevented to break out just yet by JT.
The recovery
Includes all games after the Olympics.
|
|
with Tavares |
without Tavares |
Corsi % |
||
|
|
Total |
Corsi % |
Total |
Corsi % |
WOWY |
|
Moulson |
210 |
0.595 |
292 |
0.476 |
25.04% |
|
Comeau |
251 |
0.530 |
149 |
0.436 |
21.46% |
|
Okposo |
120 |
0.542 |
353 |
0.507 |
6.82% |
|
Tambellini |
82 |
0.427 |
92 |
0.413 |
3.34% |
|
Bergenheim |
138 |
0.493 |
290 |
0.479 |
2.80% |
|
Tavares |
556 |
0.502 |
|
|
|
So, after hurting everyone's numbers during the big slump, JT recovered well down the stretch and every winger he spent considerable time with indeed benefited from playing together with JT. He spent some little time with Jackman, Sim, Hunter and Park, too, and didn't do very well with those. That's why his average suffered a bit, but the fact that Moulson-JT-Okposo were back working is surely a good sign. And although production might suggest otherwise, it actually looks like that line was better than Bergenheim-JT-Comeau.
Anyway, it was a good finish, but it also was what such a team would about expect from the top line. So, JT probably will need to bring that throughout the whole season next year. Because just from looking at these numbers you can tell how much that big slump of JT hurt this team. Actually, these numbers look pretty telling indeed. Obviously, we all knew JT had that big slump in the middle of the season, but these numbers just show what happens when someone isn't on his game. Or, even more so, we now know what it means when someone's Corsi ratio struggles. It means he's incredibly ineffective indeed.
Submitted FanPosts do not necessarily reflect the views of this blog or SB Nation. If you're reading this statement, you pass the fine print legalese test. Four stars for you.
16 comments
|
1 recs |
Do you like this story?
Comments
Great stuff again, Ben
Thanks for breaking it down in the three different periods, too. This one really drills home how this whole season served as a sort of crash orientation period for JT: I think Gordon limited his minutes at times during that big slump, but he also stuck with him quite a bit, when a lot of people thought a Bailey-style health scratch or two would be in order.
I guess when you’re dealing with a player you expect to be “The Franchise” a few years from now, you take the lumps as he figures things out. Hoping, of course, that he figured things out.
Lighthouse Hockey: Playing the NHL Lotto
Yeah, your last point surely is a good one and it would indeed have been kind of strange to scratch your number-one pick. Obviously, in comparison to the average fan, or at least to me, the coaches and the managment perhaps weight the longer term a little higher when making their decisions. And rightly so. But still, what essentially got JT back on track was the Olympic break, wasn’t it?
There’s no doubt he could get out of the hole only because he matured and learned from mistakes. But it was the break that helped him most. Now, no idea if a break of only 5 days or a week in early January would have been enough to get him back on track. But it might have and JT as well as the whole team would perhaps have done better towards the Olympic break then.
But well, perhaps, they only didn’t scratch him because they knew a break was coming anyway… And after all, that tough stretch should surely turn out to be a positive thing for JT’s career.
Hm, sorry, not quite sure what you’d like to know. Could you please tell me a bit more precisely what you mean? Is it how these numbers would look if I took Fenwick stats instead of Corsi stats? Or how everyone’s individual Corsi (Fenwick) stats are?
Yeah – I checked into the Fenwick stat more and realise the stupidity of my question (because the first instance of the Fenwick stat that I saw was a Fenwick WOWY of sorts).
I do think it would be interesting to use the Fenwick stat instead of Corsi as it factors out blocked shots – which could significantly change things, seeing as the Isles were 2nd in the league for blocked shots (with 1379). I suspect that everyone’s numbers would be comparitively better.
However, this is based on the premise that winning is not correlated with blocked shots and arguably corsi is better if you are looking for a pure territorial measure (if you combine it with zone starts or the like).
Ah, well, I just didn’t know what you mean and wasn’t sure if your question made sense or if I just didn’t understand it because of my limited English…
Anyway, honestly, I don’t know enough about the coherences and about statistics in general to really tell you something educated about it. But I can give you some numbers and thoughts. What I see is that it doesn’t make everyone’s numbers comparatively better, it brings the guys closer together.
Of all players with more than 100 total Fenwick events, Fenwick looks better for the majority of the players, but only marginally (0-3%) for most of the guys. It’s clearly better than Corsi for six players (sorry, don’t know how to paste tables here in the comments):
Corsi % Fenwick % diff
Thompson 0.379 0.433 14.33%
Jackman 0.406 0.441 8.60%
Martinek 0.430 0.463 7.74%
Park 0.431 0.462 7.39%
Sutton 0.458 0.480 4.98%
Tambellini 0.414 0.432 4.39%
…it’s worse for a few:
Corsi % Fenwick % diff
Kohn 0.472 0.451 -4.37%
Witt 0.438 0.426 -2.69%
Flood 0.474 0.467 -1.48%
Hunter 0.522 0.515 -1.32%
Schremp 0.505 0.498 -1.25%
Di Pietro 0.503 0.497 -1.10%
Gervais 0.481 0.477 -0.90%
Moulson 0.503 0.500 -0.55%
Tavares 0.479 0.478 -0.13%
It’s very obvious that those who have the worst Corsi numbers benefit the most from excluding blocked shots. That’s logical. They spend so much time in their own zone and almost certainly will have much more blocked shots against than blocked shots for.
The 2nd group is much more strange. Three (weak) defenders suffered most from excluding blocked shots. Why is that? Have these guys no clue about blocking shots? The rest of the guys there are more obvious again, there’s four (of total eight) players with a Corsi above 0.500 and these are kind of expected to shoot the puck a lot and have shots blocked as opposed to block opposing shots in their own zone.
Just in case you wonder, here are the season totals for the wingers when being on with Tavares (same as column 1 in the first table). I have no clue how to interpret the differences, though.
Corsi % Fenwick % diff
Bergie 0.498 0.511 2.70%
Moulson 0.494 0.488 -1.30%
Hunter 0.493 0.446 -9.56%
Okposo 0.491 0.503 2.52%
Bailey 0.481 0.484 0.56%
Sim 0.474 0.500 5.45%
Comeau 0.474 0.483 2.10%
Park 0.433 0.421 -2.76%
Tambellini 0.427 0.430 0.71%
or if I just didn’t understand it because of my limited English…
Your English is about as limited as Federer’s tennis game! I’m continually impressed with your word choice and can almost forget it’s not your first language.
Lighthouse Hockey: Playing the NHL Lotto
Firstly, thanks for the numbers and the analysis.
Secondly, the odd name out is Brendan Witt – I mean he blocked 74 shots in 42 games compared to Kohn (13 in 22GP) and Flood (5 in 6GP). You would think that based on this that his Fenwick would be better than his Corsi. If you consider the context though – that he was facing weak opposition with good teammates (like Kohn also), the results make more sense (along with the fact that he was brutal this year).
The most interesting thing is the fact that the Fenwick shows a solid pattern of player Corsi numbers regressing to a mean figure (just below .5) and that it seems to also remove some of the bias against players that spend a lot of time in the defensive zone.
As said, I don’t understand enough about all this to tell whether Fenwick would be more accurate than Corsi. But it’s definitely interesting and I agree with your conclusions. I just don’t know if regression to the mean, bias against players in offensive roles, etc. are good things or not… Of course, I’ve worked more with Corsi and find that more meaningful perhaps because of that, but to me it looks like Fenwick could be tougher to interpret. Fenwick seems to kind of adjust for the competition faced and situations used in. That’s cool, but it obviously adjusts only a bit and you still have to take those things into account when taking a closer look at the numbers. And although we never know by how much we have to adjust the numbers, probably we struggle particularly when looking at Fenwick. Not sure if all this makes sense, though…
As for Witt, well, he’s particularly tricky, as he was used in a shutdown role early in the season. It’s unbelievable, but Gordon indeed used him a lot with Martinek against tough competition in the first three, four weeks. In sum, he surely spent more time against weak comeptition, but his overall numbers are still affected by the tough start. Here are Witt’s numbers with the respective centers.
Total Corsi Corsi Total Fenwick Fenwick diff +/-
Park 246 0.362 175 0.360 -0.49% -15
Schremp 132 0.485 92 0.435 -10.33% 0
Nielsen 183 0.514 132 0.530 3.24% +1
Tavares 318 0.434 230 0.404 -6.82% -6
Most of the time with Park must have been early on against tough competition. And they were awful. However, the reason why his Corsi surprisingly is better than his Fenwick comes from his time spent together with Tavares and Schremp and thus against weaker competition. So, Fenwick surely removes some of the bias against players in defensive roles. But well, does that really help to interpret the numbers? And how much are these numbers affected by individual patterns. I don’t know, but I’d say Witt’s Fenwick with JT and Schremp is that bad because those guys and the other forwards on their lines (Hunter, Moulson, etc.) struggle to block shots in their own zone, but have lots of their own shots blocked by the opposition. And not because Witt spent so much time in the own zone when being on with JT and Schremp. But well, I don’t know, Witt is tricky anyway…
…and I’ve just found that Witt & Co. can’t get a shot through. 0.43 of all shots taken by Witt were blocked by the opposition. The samples are small, though, and in terms of Corsi (or Fenwick) the effect is obviously not very big, as it’s much more important how well the forwards on their lines get the shots through. But it seems to have an effect. After all, Kohn’s Fenwick suffers from the fact that he doesn’t get a shot through. But how does that make Fenwick more meaningful than Corsi? I don’t know. I’d still rather have Kohn’s shot getting blocked than him not taking a shot (and thus perhaps not spending as much time in the offensive zone).
shots that were blocked / shots taken / ratio
Kohn 10 20 0.50
Flood 5 11 0.45
Witt 23 54 0.43
Hillen 49 137 0.36
Meyer 39 116 0.34
Reese 10 30 0.33
Gervais 52 157 0.33
Streit 66 210 0.31
MacDonald 24 81 0.30
Sutton 27 103 0.26
Martinek 8 35 0.23
It appears that Fenwick and Corsi stats both have their advantages. The Corsi is a better out and out territorial measure – as it includes all offensive events. However, Fenwick is better as a tool for comparing defensive and offensive players and possibly determining who has a bigger effect on victory by putting shots on and near the net (blocked shots may not correlate with winning percentage – for example see this recent article on the hockey or die blog).
It’s good fun to talk about either way…
That’s true, but I think we have to make sure what we’re talking about. We can’t apply the same reasonings for the team level as well as the individual level. See shooting percentage for example, where individuals are better than others, but teams don’t differ nearly as much. The same goes for blocked shots, on both sides of the rink. Some are better in getting the shots through than others and some are better in blocking shots than others. I’m not saying that is why we should look at Corsi instead of Fenwick on the individual level. But somehow it’s still more intuitive to me.
Yeah, I hear you – but obviously not all shots on net are equal either. However, a blocked shot will never directly result in a goal – regardless of how good the block to stop the shot was or the potential quality of the shot.
I do think I will check into the correlation between winning percentage (excluding shootouts of course) and proportion of blocked shots to overall shots taken sometime over the off-season. If there is still no correlation then I might be a bit more sure about believing in the Fenwick stat (yes, player value and events that result in victory are different), as if there is a decent correlation then you wouldn’t want to use Fenwick.
Hockey's moving parts
This is fun — all part of that discussion of, “What is it I want a hockey player to do?” Well, ideally I want him to play against the toughest opposition in the toughest situations and still outscore them. And since there are but a handful of players who can actually do that consistently, we’re left to figure out how to evaluate all the mere mortals as they are helped and/or held back by other mortals, and so many moving parts.
(e.g. the way you don’t want to “punish” a player’s eval simply for spending more time in the defensive zone, but at the same time you’d prefer he spend less time there if possible.)
I like the way this discussion tries to tackle the gray areas of figuring out who contributes what, and I think I like that there is no simple, set-in-stone answer.
Lighthouse Hockey: Playing the NHL Lotto
I like the way this discussion tries to tackle the gray areas of figuring out who contributes what, and I think I like that there is no simple, set-in-stone answer.
Except for Streit, that guy is a stud – no gray area there.
And gray areas are what keep things fun especially in the offseason – dreaming of free-agents and draft targets, and evaluating the future of the Isles basically (especially when there is only one game of the playoffs playing).

by 










































