Don’t look now: Last season’s worst by ESTR
Even Strength Total Rating, Regular season 2008-9
Rank Name Team Pos ESTR
1 Colton Orr NYR RW -1.60
2 Chris Stewart COL RW -1.53
3 Brendan Witt NYI D -1.52
4 Kris Draper DET RW -1.45
5 Kyle Turris PHX C -1.32
6 Thomas Pock NYI D -1.32
7 Darroll Powe PHI C -1.29
8 Jay Pandolfo NJD LW -1.27
9 Patrick Lalime BUF G -1.26
10 F. Sjostrom NYR RW -1.25
over 2 years ago
MarioD
23 comments
0 recs |
Email
PrintComments
Here’s an explanation of what ESTR is.
http://puckprospectus.com/article.php?articleid=254
Even Strength Total Rating (ESTR) is a more advanced version of Goal Difference per 60 minutes (GD/60), taking into account the quality of teammates and quality of opposition for all non-empty net Even Strength goals scored while a player is on the ice. ESTR is a sum of its components, Even Strength Offensive Rating (ESOR) and Even Strength Defensive Rating (ESDR), which are Goals For per 60 minutes (GF/60) above average and Goals Against per 60 minutes (GA/60) above average, when adjusted for teammates and opposition. To get ESTR and its components for each player, goal difficulties for all goals over the course of a regular season (or postseason) are calculated, weighted by the base GF/60 and GA/60 of the players on the ice. For each player’s GF and GA, the average goal difficulty the player was on-ice for is then applied against each player’s base GF/60 and GA/60 to get their ESOR, ESDR and ESTR.
In short, ESTR is the number of goals per 60 minutes of ESTOI that the player is worth at Even Strength, given NHL-average players as teammates and opposition.
Ie: Darrol Powe, last season, was the seventh worst even strength player in all of the NHL.
One issue
This is trying to factor “goal difficulty” – a phrase used twice – but is eliminating empty net goals. Is that because they are seen as easy or because it is essentially a shorthanded situation, and therefore not in the even-strength equation?
Broad Street Hockey -
Makin' it look mean since 1967.
I’ve actually discussed EN with Timo a couple times, so I think I can answer this without looking up and quoting explanations.
In short, EN goals are complete bullshit and really tell us nothing about scoring ability or defensive ability.
The most difficult empty net goal requires the scoring team merely to gain the blue line and shoot. So EN should never really be counted in anything. They’re the equivalent of a sacrifice bunt in baseball.
By “goal difficulty” I believe that means adjusted for the skill level of the players on the ice.
Wait...
…did Patrick Lalime score a goal last year? What is he doing on this list?
Do you see what happens when you find a stranger in the Alps?
This is interesting
The Top 10?
1 Patrik Berglund STL C +2.02
2 T.J. Oshie STL C +1.55
3 Tomas Holmstrom DET LW +1.53
4 Tim Connolly BUF C +1.52
5 Blake Wheeler BOS RW +1.48
6 Alex Tanguay MTL LW +1.45
7 Michael Ryder BOS RW +1.45
8 David Perron STL LW +1.44
9 Bobby Ryan ANA RW +1.43
10 David Krejci BOS C +1.42
I’m not so sure that represents the 10 best players in the NHL…
Broad Street Hockey -
Makin' it look mean since 1967.
Also
Using ESTR, Evgeni Malkin is rated 8th best ON HIS OWN TEAM. Behind such great players as Bill Guerin, Maxime Talbot, and Hall Gill.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 1, 2009 4:18 PM EDT up reply actions
First off, it evaluates only Even Strength play.
Second, it incorporates time on ice into the equation.
Malkin logs a ridiculous amount of time, and doesn’t produce. It’s the same criticism of Ovechkin double-shifting himself. He’s just not going to be productive.
You lose a little bit
when you say neither Ovechkin or Malkin produce. They were the top two in Even Strength points in the league. Yes they play a lot, but they also are 1 and 2 in points per game.
But the top ten’s ESTOI/G:
1. 11:08
2. 11:59
3. 11:57
4. 13: 09
5. 11:35
6. 11:59
7. 11:49
8. 11:17
9. 12:42
10. 12:40
Basically, I’m not so sure what the point in that stat is. It is obviously a stat that gives preference to players receiving third line minutes, who just happen to get some special teams time. The top ten is a good reference point to list when you list the bottom ten to prove a point.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 1, 2009 8:12 PM EDT up reply actions
Basically, I’m not so sure what the point in that stat is. It is obviously a stat that gives preference to players receiving third line minutes, who just happen to get some special teams time. The top ten is a good reference point to list when you list the bottom ten to prove a point.
You’re confused.
First off, it does not count special team time, so your second sentence is completely off the mark. It evaluates purely even strength (I believe this includes 4-4 time).
Second, you’re talking about points, but this isn’t about points. It’s about total play. When has Evgeni Malkin ever been known for his defense? Same for Ovechkin who spends most of his D time parked at the other blue line.
Let me stop here because I just found this: In the post with the top 10, Timo responded in the comments to explain the stat:
The purpose of Even Strength Total Rating, ESTR, is essentially the same as plus/minus ratingto give an overall offensive/defensive value to a playerbut it’s a quantum leap forward. The problems with plus/minus? Doesn’t account for TOI, strength of teammates, or strength of opponents. Pretty big problems.
ESTR is the most advanced measurement of its kind that I know of, to deal with strength of teammates and strength of opponents on every goal that the player was on the ice.
…
In short, ESTR is Goal Difference per 60 minutes of ESTOI, adjusted for strength of opponents and strength of teammates.
…
The fact that it’s a rate stat "per 60″ accounts for TOI. I like rate stats, most folks out there love the counting stats.
The starting point is looking at GF/60 and GA/60 of all the players on the ice, for each goal.
In essence, you’re saying “Hey look, the top ten in that stat doesn’t match up with conventional wisdom! Therefore, the stat is bad.” Timmo introduces it by saying:
Who was the best overall player in the NHL last season? Pavel Datsyuk, of course. Right? No? Evgeni Malkin? Alexander Ovechkin? Sidney Crosby? Here’s your first look at the Even Strength Total Rating for the 2008-9 regular season:
Timo explains this a bit in his Prospectus post yesterday on this topic:
Where are the Big Three, you might ask? Evgeni Malkin was close to cracking the list at +0.98 ESTR, but Sidney Crosby and Alexander Ovechkin do not rank as highly, due to negative ESDR.
Shit, HTML fails. There should be no strikethrough and this was all supposed to be the same block quote:
The purpose of Even Strength Total Rating, ESTR, is essentially the same as plus/minus rating to give an overall offensive/defensive value to a player but it’s a quantum leap forward. The problems with plus/minus? Doesn’t account for TOI, strength of teammates, or strength of opponents. Pretty big problems.
ESTR is the most advanced measurement of its kind that I know of, to deal with strength of teammates and strength of opponents on every goal that the player was on the ice.
…
In short, ESTR is Goal Difference per 60 minutes of ESTOI, adjusted for strength of opponents and strength of teammates.
…
The fact that it’s a rate stat "per 60″ accounts for TOI. I like rate stats, most folks out there love the counting stats.
The starting point is looking at GF/60 and GA/60 of all the players on the ice, for each goal.
Ok
That all makes sense. The GFON/60 and GAON/60 are stats I enjoy a lot. As are qualcomp and qualteam. However, I think I am having a problem with the combination of the four of them. Each of them alone I get (seeing as how qualteam and qualcomp are based off of the previous two) so I’m wondering how the combination of them makes sense.
I’m certainly having a problem with the results. That isn’t to say that I think the stat is crap because I don’t like the results, but not one of those guys has ever been known to be superior in any aspect of the game. When I saw virtually all of them received so little ice time, that reinforced the notion that something is wrong.
I like the idea, but I just don’t see how the parts of the equation fit together to produce the result. Unless it isn’t a combination of the four stats I said earlier, in which case I just flat out missed the point again and feel dumb.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 1, 2009 11:03 PM EDT up reply actions
Two things. First, let’s break down the four stats.
A) GFON/60 – How many goals were scored when that player was on the ice
It’s an individual stat so we don’t care about goals allowed while that player was on the bench
B) GAON/60 – How many goals were given up when you on the ice
Same above
The above two stats are also rate stats, to they equalize time on ice.
C) Qualcomp
Here’s a game the Flyers played against the Sabres and Lalime was in net. As we see in the original post, Lalime was attrocious last season. So, shouldn’t those six goals scored on Lalime be worth statistically less than a goal scored against Tim Thomas last season?
D) Qualteam
By the same token, Thomas Vanek is on ice when Lalime allows a goal from the redline. Shouldn’t that hurt Vanek less than when he’s on ice for a Ryan Miller GA?
By combining the four stats, you are taking team Goal Differential and factoring out all the variables. First, we only care about goals when the specific player is on the ice. Second, get rid of special teams play and empty net goals, because those are either not true indicators of ability (EN Goals, for instance don’t tell us anything about ability) or at least are of a different ability (PP Goals for, there’s a skill to PP but not really the same as there is 5-5).
Now that we’re down to ES goes scored while on the ice, who else was on the ice. If Richards was on the ice with Cote/Powe/Nodl and Sbisa do you really think the goal against could possibly have been Mike’s fault? So, third, we normalize teammates. Now, normalize opposition, looking at goaltenders as the easiest illustration, it would be silly to value a goal scored against Jose Theodore the same way we value a goal scored against Henrik Lundquist.
You come out with the:
Rate of Even Strength Goals While Player X Was On the Ice Normalized for Opponent Skill Level and Teammate Skill Level.
Again, that all makes sense
Are you aware of the actual equation? In other words, how do they come up with the +2.02? That might only confuse me, but it would be interesting to see how they go about factoring in all of those things to come up with a specific number.
To do a (possibly horrible) job of summarizing the stat, it is trying to compute who does the most with the longest odds (scoring on Tim Thomas while playing with Cote/Nodl/Parent/Sbisa or preventing a goal from being scored on Patrick Lalime while those same 4 are also on the ice)?
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 2, 2009 1:23 PM EDT up reply actions
To do a (possibly horrible) job of summarizing the stat, it is trying to compute who does the most with the longest odds (scoring on Tim Thomas while playing with Cote/Nodl/Parent/Sbisa or preventing a goal from being scored on Patrick Lalime while those same 4 are also on the ice)?
No no no. Don’t take my examples too literally. I’m just using the extreme examples to show why each normalization needs to occur.
I don’t know how familiar you are with sabermetrics, but let me see if this helps:
Total Bases: OPS+; Plus/Minus: ESTR
Both take a counting stat, turn it into a rate stat, and then normalize it (but for different factors, obviously).
Its attempting to tell you the same thing as +/- is attempting to tell you. But in a much more fair way. (Not counting EN goals into the equation, normalizing for the players on the ice, and recording it as a rate, not overall totals.)
As to your “why aren’t any star players up there” issue, lets look at Evgeni Malkin:
2008-09 NHL +/-
Malkin(17), Ovechkin(8), and Crosby(+3) are 45th, 111th, and 204th in the league in +/- last season.
So, in response to your earlier criticism of top players not being there, it is clearly an improvement upon +/-. Also note that Malkin was only 3rd on his team in +/- last season (in response to the criticism that he’s 8th on his team in ESTR).
We can both agree +/- sucks. Even the special teams included +/- sucks, though I prefer that to the former.
I am aware your example was the most far fetched possible. It was helpful.
The analogy is a bit different. I see what you mean, and it makes sense. Accepting that, the difference though is that OPS+ doesn’t have to factor in teammates and opponents. Also, OPS+ tends to give you the best players in the game. Pujols and Mauer are leading the NL and AL this year in that, whereas here, the top 10 includes a bunch of no-names and typically known average players.
Either way, +/- isn’t used often as it is, so where will this fit in? Maybe this is something better equated to defensive fielding stats in baseball, since there never has been a good way to evaluate those. You would need to factor in a player’s reaction time, range, positioning, arm, etc. Until the new invention of cameras above the fields this year, that might change. But in hockey, there never has been a good way to numerically value the defensive specialists in the game. A name I was more surprised to find absent from this list was Pavel Datsyuk since he plays both zones very well.
Good point about the star players – I knew Crosby sucked, this is just yet another proof of that.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 2, 2009 7:13 PM EDT up reply actions
I don’t entirely disagree with anything you said, but let me clarify and correct a couple points.
I made the OPS+ comparison as a sort of “how the sausage is made” comparison, not as dead accurate. The major difference is that OPS+ is an offense only statistic, whereas ESTR accounts for both offense and defense. So, again, your perception of the best players isn’t necessarily dispositive of the stat. (Your Datsyuk example is great though, give me a moment to return to that.)
OPS+ is different because as you said it doesn’t factor opposition and teammates. However, it does factor in ballparks, which ESTR doesn’t have to. (Though I wonder if ice conditions could be demonstrated to consistently affect play. If not by arena, perhaps by whether there was an NBA game the night before…. Maybe five years from now we’ll have that stat?)
As for Datsyuk, and this is purely my own theory: Because it is a rate statistic, I believe this is indicative that overplaying your top players makes them less effective. (Datsyuk average 19:13 TOI last season.)
I have an alternative idea but it’s really not that well fleshed out, so I’m not entirely confident in it. But I’ll espouse it poorly anyhow: Even though we compensate qualcom and qualteam, perhaps, still Datsyuk has the defensive talent to shut down Crosby, but Berglund does not. what I’m saying is, what we know as Datsyuk’s talent is really that he has a top tier Berglund doesn’t have. As an example to explain this, (making this all up, obviously): Ryan Howard and Albert Pujols have the same OPS+. However, Ryan Howard’s homeruns average 405 feet and Pujols’ average home run distance in 380. Hope that makes sense…
Yeah, I noticed the TOI disparity when showing the Top 10. That has to somehow be relevant. Almost all of those guys played between 11 and 13 minutes of ES. I doubt that’s a coincidence.
On your theory: I follow until the example. Rather than home run distance, would a better example be that Albert Pujols can hit Tim Lincecum, Chris Carpenter, Cliff Lee, and C.C. Sabathia (as in the better pitchers in the game) to an OPS+ better than Ryan Howard – however marginal? As in, Pujols “has a top tier [Howard] doesn’t have”. Obviously, Howard can hit those people, but Pujols is just a better player in almost every regard.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 2, 2009 9:17 PM EDT up reply actions
I think you have the right idea, and its a more on the nose hypothetical than mine. I was trying for the easiest to understand example, and the best I came up with was to demonstrate how stats can’t account for all abilities.
The problem with your example, though is that if Howard and Pujols have an equal OPS+ and Pujols OPS+ against Aces is higher, that means Howards has to be better against non-Ace pitching.
The point I’m trying to make is somewhere between the two hypos, I think…. But you and I seem to be on the same page, just neither of us has figured out precisely how to put it into words.
Haha
Yeah, no problem. I get the point even if I can’t articulate it.
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 3, 2009 10:50 AM EDT up reply actions
WHOA WHOA WHOA WHOA WHOA
Re: Malkin being 8th on his team, read your link again!
http://icehockeymetrics.com/?p=3
For our example, we’ll look at the Stanley Cup Champion Pittsburgh Penguins’ postseason ESTR
(emphasis added)
Oops…
Broad Street Hockey -
Makin' it look mean since 1967.
by Geoff Detweiler on Sep 2, 2009 6:55 PM EDT up reply actions
All that discussion about ESTR?
Thanks, I’m honored. Here’s the Penguins’ regular season ESTRs, exclusive here:
Name Pos ESTR
RUSLAN FEDOTENKO L 1.40
TYLER KENNEDY C 1.00
EVGENI MALKIN C 0.98
ALEX GOLIGOSKI D 0.64
CHRIS KUNITZ L 0.62
MIROSLAV SATAN R 0.52
PETR SYKORA R 0.46
PHILIPPE BOUCHER D 0.46
ROB SCUDERI D 0.44
SIDNEY CROSBY C 0.32
HAL GILL D 0.19
BILL GUERIN R 0.13
MARK EATON D 0.04
JORDAN STAAL C 0.00
BROOKS ORPIK D -0.07
KRIS LETANG D -0.23
DANY SABOURIN G -0.30
PASCAL DUPUIS L -0.35
MATT COOKE L -0.40
MAXIME TALBOT C -0.85
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Broad Street Hockey
Related FanShots on Broad Street Hockey
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
SBNation.com Recent Stories
In Association With
Latest Puck Daddy - NHL from Yahoo! Sports
- Wednesday’s Three Stars: Devils win; Kreider ties record; Brodeur’s sick save
- Watch John Tortorella’s Game 2 press conference, as he’ll ‘keep it in the room’ (VIDEO)
- Devils Coach Pete DeBoer’s adjustments make difference in Game 2 win vs. Rangers
- Marty Brodeur’s ‘own goal’ in Game 2 vs. Rangers (VIDEO)
- Coyotes’ Martin Hanzal suspended 1 game for boarding Dustin Brown
Latest stories and headlines from NHL.com
- Rangers coach John Tortorella benches Gaborik for much of third after blunder on tying goal
- Rangers coach John Tortorella benches Gaborik for much of third after blunder on tying goal
- Rangers lose, Gaborik in doghouse
- Devils-Rangers Sum
- Devils D Adam Larsson out of New Jersey lineup for Game 2, replaced by Peter Harrold
