New Stats on Behindthenet.ca: Zone Finish
by Hawerchuk on May 19, 2010 9:01 AM EDT
25 comments
I put zone starts on the site at the beginning of the season, but due to server space limitations, I never posted zone finishes. Here's the link:
http://www.behindthenet.ca/2009/new_5_on_5.php?sort=§ion=zonestart&mingp=&mintoi=&team=&pos=
You can see from this chart that hockey swings very hard towards its equilibrium - no player ever finishes in the defensive or offensive zone to the extremes that players are started in those zones by their coaches:
Bill James once noted a similar effect in baseball - no pitcher gives up home runs as often as Albert Pujols hits them, or as rarely as Willie Bloomquist hits them.
0 recs |
25 comments
|
Email
Print
Comments
Pretty sweet.
Just doing some sanity checks.
Pearson correlation between ZS and ZF: 0.62
Pearson correlation between ZF and Corsi%: 0.54
All from 09/10.
The former clearly speaks to the impact of coaching, you start more in the offensive zone and you end up more in the offensive zone. Some people question this basic aspect of hockey, which is lunacy.
The latter is interesting. Both ZF and Corsi are logically measures of territorial advantage, I instinctively would have expected this to line up better. Still, a pretty strong relationship, I imagine it would be even stronger if we could come up with a measure for shifts-ending-on-the-fly.
And this is great for making specific arguments. A quick look at Pitts over the last three years gives an idea of the separation between Crosby and Malkin (hint: it ain’t close, Crosby is the man on the Pens and Malkin is just flat out bad at 5 on 5, in context).
RO,
What stat are you looking at regarding the separation between Crosby and Malkin? I’m having trouble understanding the tables and what the data means. Any help would be appreciated.
Sid for the Hart
Where their shifts end, relative to where they begin. One of the players is driving the bus, and the other is getting run over by it.
By that standard, every worthwhile Chicago forward other than Versteeg and Madden should shoot themselves in the head. Conversely, every significant Florida forward should be seen as superb at flipping the zone, since every last one of them improved from their starting point.
by Robert Cleave on May 19, 2010 7:42 PM EDT up reply actions
well, you need to adjust by the amount by which zones shift to the mean. i think that you can really only compare players with very similar o-start percentages
No doubt. I was just tweaking Richard, Gabe.
by Robert Cleave on May 19, 2010 9:39 PM EDT up reply actions
And kidding aside, you’d want to match quality of comp as well.
by Robert Cleave on May 19, 2010 9:48 PM EDT up reply actions
hope this is both factually correct and helpful.
start with isolating Pitt forwards, then use the pull-down tab thing to go to “zone start” .
way over on the right, there are two columns marked “OPCT”. this stands for (I assume) offensive zone percentage, which is basically “offensive zone starts(or finishes) / [offensive zone starts(or finishes) + defensive zone starts(or finishes)]”. note that this ignores neutral zone starts totally, because theoretically nobody’s at an advantage on a neutral zone draw.
The leftmost of the two “OPCT” columns represents zone start (>50% means that that player is used for more offensive zone faceoffs than he is used for defensive FOs) while the rightmost “OPCT” column represents zone finishes (this new info).
So, Sid’s numbers are:
-56.7% of his non-neutral-zone FOs are in the offensive zone (offensive zone draws (238 wins+197losses) divided by OZ+DZ (238+197+189+143), or 435/767).
-56.8% of his shifts end in the offensive zone.
Geno’s numbers are:
-58.1% of faceoffs are in the offensive zone.
-53.2% of shifts end in the offensive zone.
All in all, both Sid and Geno are used a lot in the offensive zone. This year, the puck tended to leave the offensive zone with Geno out there, which is fairly standard as the chart at the top of the page represents. Crosby, however, managed to keep his shit in the O-zone, which is a feat unto itself.
by Passive Voice on May 19, 2010 7:56 PM EDT up reply actions
Vic may do it differently than I do. A ‘zone finish’ is a sequence of play ending with a whistle, whether or not a player comes off the ice.
I tried adding in the EV goals for and against, those would naturally end up being faceoffs if they don’t go in. When I do this I get a new correlations:
Pearson correlation between ZS and ZF: 0.62
Pearson correlation between ZF and Corsi%: 0.60
I don’t know, it might make a difference but these are fine differences and we’re painting with the big brush here. The pattern is there.
Does comparing ZF to ZS have any value in determining how a player gets the puck moving? Or is it all sort of noise around the bigger equilibrium effect you describe?
no matter the answer, these numbers don’t do anything to help my Evanderection. that kid’s a monster.
by Passive Voice on May 19, 2010 1:53 PM EDT up reply actions
i wonder if...
there’s some way to find a connection between zone start and corsi/fenwick such that we could find some adjusted fenwick that normalizes zone start to 50%. i was looking at vladimir zharkov’s fenwick, which is like 56%, but his zone start is high (60%) and his zone finish is relatively low (49%). it stands to reason that if his zone start was 50%, his fenwick would be lower.
Vic and Jlikens have correlated zonestart to Corsi and Fenwick. A positive zonestart is worth .8 Corsi and .6 Fenwick.
I don’t have links handy though.
Editor of The Copper & Blue, and leader of The Cult Of Hartikainen.
thanks
i’ll go digging for the jlikens and vic links.
looking at that exceeding expectations link -
i’d also be curious about the correlation between zone start and fenwick wrt faceoffs in the defensive zone generally. just intuitively (and i think brodeur is a fraud found this), the devils have less faceoffs in their zone than most teams – this is because brodeur (sometimes foolishly) will always choose to play the puck if it’s possible. i guess what i’m asking is too complicated and there’s not enough data anyway – is it better, fenwick-wise, to freeze the puck or to keep it in play?
Some teams have fewer faceoffs period. I’ve noticed a tendency for low-event teams to have fewer faceoffs – or for low-faceoff teams to have fewer events (SF and SA) – but have done nothing comprehensive, just random observations.
But here’s one: the team with the fewest total faceoffs this year was New Jersey – a whopping 222 fewer than any other club (4244 to Chicago’s 4466, so about 5% fewer). At the other end was Washington with 4927 – 104 more than anybody else. The range is from <52 per game for NJD to slightly over 60 for WSH, but most of the teams are packed into a range half that wide.
Go back over the past 5 years since the lockout, and NJD were 2nd, 1st, 2nd, 9th, and 1st in fewest faceoffs. The “9th” was the year that Brodeur got hurt and missed a substantial portion of the season. So it’s fairly easy to conclude that he is the (or certainly “a”) driving force behind that stat, by keeping the play flowing as efficiently as he does. Other regular teams in the top 5 include over that span include San Jose, Dallas and NYR. Nabokov and Turco are well-known puckmovers, I can’t say as I’ve pegged Lundqvist one way or the other in this dept. (unlike the others, I have never seen him play a live game which is when I tend to really notice this sort of thing). But at a guess, he likes to keep the play moving.
Anyway, not sure what this all really means, just throwing it out there as something worth considering. I think there is an absolute gold mine out there in faceoff stats, not just in the whos and the wheres but in the how manys.
Writer for The Copper & Blue and primary shareholder of Zorg Industries
"Never be ashamed of who you are" -- Jean-Baptiste Emanuel Zorg
by Bruce McCurdy on May 21, 2010 12:46 AM EDT up reply actions
re: lundqvist
i watch the rangers a lot and haven’t noticed a tendency to play the puck a lot. but lundqvist’s weak point is his glove – he loves letting shots hit him. it’s possible that his style of goaltending just leads to fewer whistles.
solid work on the faceoffs stuff – i knew NJ was up near the top, didn’t know they were that high. some devils fans don’t like this tendency, because brodeur will not take into account, say, that players need a line change, or will sometimes just throw it out into traffic and hope. NJ has also been generally weak down the middle faceoff-wise (08-09 being an exception), so i’m not sure if it’s brodeur helping his team out or just doing what he feels like.
Maybe I’m an idiot, but why do Chicago’s and Florida’s zone finishes, as a team, regress towards the mean? I understand for individual players, but for the team in aggregate, shouldn’t the ratio of Ozone Starts / Dzone Starts ~= Ozone Finishes / Dzone finishes? I was assuming you defined a finish as where the next faceoff starts?
1. There’s definitely some vagaries to the way everything is tabulated. Absolutely, those things should be approx =, but it’s approx.
Scoring an EV goal should be called an OZ finish, but clearly the next shift is an NZ start. There are enough EV goals scored that, given that we’re mostly in the 45%-55% band, these things matter. Have to make sure icings and penalties are properly accounted for, etc… Gabe went to a great school and is obviously smarter than me, but accounting for the neutral zone is tricky.
2. The point about anyone who gets disproportionate OZ or DZ starts (say the outer quintiles) is important. If you try to compare them by counting ‘which way they’re moving the puck’ like they’re apples and apples, you’ll get a misleading result.
3. I love Triumph44’s question. Ever since this whole subject started being discussed in earnest a couple of years ago (i.e. the puck position game) I’ve had a nagging idea that the entirety of benefit from a good puckhandling goalie is keeping it alive. Not helping the breakout per se, but just avoiding the known risk of an own-zone faceoff without introducing countervailing risk.
4. Gabe: I hope RH’s stag was enjoyable.
ah yeah, I should subtract out goals for and against. I thought I’d post the raw numbers, but it seems more “correct” to account for the real shift finishes. Triumph’s question is great. I’m going to but some time into thinking about it.
RH’s party was about what you’d expect from a dozen or so Queen’s engineers in Vegas…
Question
If the determination of zone finish is based upon the faceoff location following a stoppage in play there’s a few issues, although I’m guessing they would be statistically inconsequential.
1. The following rule might detract from the stat a bit:
When a stoppage of play has been caused by any player of the attacking side in the attacking zone, the ensuing face-off shall be made in the neutral zone on the nearest face-off spot. If two rule violations are the reason for the stoppage of play (ie. high-sticking the puck and intentional off-side), the ensuing face-off location shall be determined as the spot that provides the least amount of territorial advantage to the offending team.
This would definitely penalize players (by this measure) on the ice when a teammate causes a stoppage in play.
2. Similar to the above, any time a D-man new player descends beyond the point, his team’s offensive zone finish is going to be bounced out to the neutral zone, which would again screw up zone finish.
3. Goals for and against also present the same problem.
Rather than subtracting goals for/against, wouldn’t it make sense to just add goals for to the offensive zone, and goals against to the defensive zone? It would screw up the percentages obviously, but you could easily deduct the combination from the neutral zone finishes.
"Success is the ability to go from one failure to another with no loss of enthusiasm."
- Sir Winston Churchill
I'm pretty sure he's talking about the Leafs.
by Steve Burtch on Jun 5, 2010 11:10 AM EDT up reply actions
Comments For This Post Are Closed
Secondary Sidebar
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2010 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
