There was a bit of a discussion last week between truck, Ben and myself about the different quality of competition values given on behind the net. I thought it would be worthwhile to post the numbers for the Jets. The second part of this article is simply me trying to get my head around some of the points truck and Ben were making. (In other words feel free skip that part).
Here is a pretty good explanation of Corsi QoC. As I understand it (and the previous discussion will prove I'm certainly no expert) Corsi Rel QoC is the same as Corsi QoC except that it uses Rel Corsi instead of Corsi in the calculation. (There's also a +/- QoC which uses opposing players +/- instead of Corsi).
Here are the Jets numbers for both. The player's ranking by position is indicated in brackets.
|Corsi Rel QoC||Corsi QoC|
|Slater||0.85 (1)||0.05 (1)|
|Glass||0.76 (2)||-0.08 (5)|
|Little||0.52 (3)||-0.26 (7)|
|Thorburn||0.51 (4)||0.05 (2)|
|Ladd||0.50 (5)||-0.07 (4)|
|Wheeler||0.23 (6)||-0.35 (8)|
|Wellwood||0.00 (7)||-0.38 (9)|
|Meat||-0.01 (8)||-0.17 (6)|
|Antropov||-0.2 (9)||0.03 (3)|
|Burmistrov||-0.19 (10)||-0.55 (11)|
|Kane||-0.24 (11)||-0.58 (12)|
|Fehr||-0.53 (12)||-0.41 (10)|
|Steam Punk||-0.79 (13)||-0.86 (13)|
|Corsi Rel QoC||Corsi QoC|
|Byfuglien||0.69 (1)||-0.19 (5)|
|Toby||0.59 (2)||0.21 (2)|
|Bogosian||0.59 (3)||-0.14 (4)|
|Oduya||0.35 (4)||0.47 (1)|
|Hainsey||0.33 (5)||-0.13 (3)|
|Stu||-0.19 (6)||-0.32 (6)|
|Flood||-0.73 (7)||-1.09 (8)|
|Clitsome||-0.78 (8)||-0.92 (7)|
|Jones||-1.2 (9)||-1.53 (9)|
I'm going to stick with Corsi Rel QoC for now. There's discussion on the differences between it and Corsi QOC further down.
For the Forwards
- The toughest matchups have gone to the GST line, followed by Little, Ladd & Wheeler.
- Wellwood & Miettinen have seen the mid range competition.
- The easiest matchups have gone to Antropov, young players Kane & Burmistrov and a couple guys who've had a hard time cracking the lineup, Fehr and Stapleton.
For the Defense
- The toughest matchups have gone to our big 3 on D of Dustin, Toby, and Bogosian.
- Oduya, Hainsey have gotten the mid range matchups. Oduya's numbers will include a few games with Chicago. Stuart also fits in here although he's close to being in the next group down.
- Jones got the easiest matchups followed by Flood. Clitsome's numbers are shown here but they're obviously based on his time with Columbus. Comparing them to the other Columbus defenseman indicates he probably belongs in this grouping as well.
Corsi Rel QoC vs Corsi QoC:
Corsi QoC is indicated in the second column. The rankings are pretty close to the Corsi Rel QoC. However there's a few exceptions which I find interesting:
- Little drop from 3rd to 7th
- Antropov jumps from 9th to 3rd
- Dustin drops from 1st to 5th
- Oduya jumps from 4th to 1st.
So of the two Quality of Competition metrics which one do you look at and why? Also, how do you explain the difference for the four players above?
Most people who know their way around this stuff use Corsi Rel QoC. But the "why" of that isn't immediately clear to me.
I found a nice quote from Gabe about the difference between the two taken from this article:
Incidentally, this list uses what's known as "Corsi Rel Quality of Competition" - instead of averaging a player's raw Corsi number, we average his Corsi number relative to his team. If we use unadjusted Corsi, the Quality of Competition leaders come from just a handful of teams: primarily Columbus, St. Louis and Nashville, all of whom played significant numbers of games against dominant Detroit and Chicago teams.
This makes immediate sense to me if you're comparing players on different teams. But does it matter if we're only looking at the Jets players? And why the change in ranking for the 4 guys mentioned above?
Ben made the following point in our previous discussion:
..because coach’s line-match quite a bit you typically end up facing a line that has a similar place within their organization (you: 1st line, them: 3rd or 1st line, you: 2nd line, them: 2nd line, you: 4th line, them: 4th line). Since not all teams 2nd lines are equally talented, you try to adjust for it by looking at how they perform relative to teammates.
I don't think I understood how Corsi Rel QoC accounted for this until I started looking at a some of the numbers.
Here's a comparison of Corsi values between 2 Jets defenseman:
|Corsi On||Corsi Off||Corsi Rel|
The 2 have somewhat similar Corsi On values. Yet Jones' Corsi Off is much better than Bogosian's. Why is that? The most obvious explanation is that Jones is not a good hockey player and the place he does the Jets the most good is sitting on the bench. That's may be true, but if it's the only explanation than why is his Corsi On almost as good as Bogosian's?
Another explanation is that Bogosian is producing similar on ice results as Jones, except he's doing it against much tougher competition.
But if we were to do a QoC calculation for teams playing the Jets, Corsi QoC would indicate both Bogosian and Jones have roughly equal skill levels. It's only by looking at Corsi Rel instead of Corsi On do the results show that Bogosian is a better hockey player. That's part of the "how" to the point Ben was making above.
I still have no idea why the change in QoC rankings occur for to Little, Antropov, Byfuglien & Oduya. If anyone's got any theories I'd love to hear them.