By The NumbersHockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.
Actually, upon looking at the numbers again, their last few years don't make the difference. In fact, from 1988 onward when McCrimmon had left for Calgary, McCrimmon had a higher adjusted +/- than Howe (+129 to +118, although in 3.5 seasons worth more of games).
The difference really comes from their first five years. Howe was a +76 in Hartford, and McCrimmon was -17 in Boston. In their first 2 years in Philly, Howe was +42 and McCrimmon was -3, suggesting that they didn't play together those years, or at least spent a lot of time apart. (All +/- are adjusted)
The Non adjusted +/- was McCrimmon +43, Howe +77 for a difference of +34. Playing on the same team and conditions and being subjected to the same adjusted methods, how does McCrimmon lose so much ground?
Quote:
Yes he is, but I'm not sure Lehtinen is the best comparable for Sakic. He had his best numbers from 1996-2003 playing with Mike Modano. Since there isn't any way to separate the contributions of linemates with this metric, Lehtinen probably ranks higher than he should. I think Modano is the better comparison for Sakic, and I think they were closer in value than most people think.
I am not too sure about that. Lehtinen was one hell of a defensive player on his own. He had a better +/- in less games in 96/97 then he did the first year he played with Modano in 97/98
The Non adjusted +/- was McCrimmon +43, Howe +77 for a difference of +34. Playing on the same team and conditions and being subjected to the same adjusted methods, how does McCrimmon lose so much ground?
Part of the difference is from the fact that Howe played more on the PP, and McCrimmon more on the PK. I estimate SH goals and remove them in the adjustment process, and that makes a difference. I also normalize the numbers to a 200 ESG goal season, so that brings the numbers down a bit for both. After these adjustments, but before the team adjustment, I have Howe at +67 and McCrimmon at +28 for these seasons.
The team-adjustment process gives an expected +/- of +31 for McCrimmon, and +24 for Howe. That's because the players are compared to their off-ice results, so Howe's baseline includes McCrimmon and McCrimmon's baseline includes Howe. As a result, the method may slightly underrate McCrimmon in these two seasons - by 7 if you give him the same baseline as Howe - but it's not significant in the career totals. This could be seen as a flaw in the method - but the method of comparing to off-ice numbers works best for very good to great players, which is what I'm most interested in. It works less well for players who were on the same team as great players.
Finally, the team-adjusted +/- is the normalized +/- with SH goals removed minus expected +/-. For Howe, that's 67-24=+43, and for McCrimmon it's 28-31=-3.
Quote:
Originally Posted by Thornton_19
I am not too sure about that. Lehtinen was one hell of a defensive player on his own. He had a better +/- in less games in 96/97 then he did the first year he played with Modano in 97/98
If I were dividing credit for that line's plus-minus results from 1997-2003, I'd give more of the credit to Modano. He was a considerably better even-strength scorer than Lehtinen, meaning that their goals-for were driven mostly by Modano. I do agree that Lehtinen had a lot to do with their success. He was far from a passenger on that line, and probably should get much of the credit for their strong goal prevention record.
... He had his best numbers from 1996-2003 playing with Mike Modano. Since there isn't any way to separate the contributions of linemates with this metric, Lehtinen probably ranks higher than he should..
Lehtinen not worthy of his six (6) time finalist nominations for the Selke trophy or his three wins? ... or are you saying Sakic and Modano are more worthy?
Lehtinen not worthy of his six (6) time finalist nominations for the Selke trophy or his three wins? ... or are you saying Sakic and Modano are more worthy?
Or maybe he meant Lehtinen ranks higher in adjusted +/- than he should because Modano drove up their even strength scoring?
People come on! Lehtinen is probably the greatest scandinavian defensive forward ever. His only competition is fellow countryman Kurri or swede Tomas Steen. Modano is probably the one gaining +/- on their line, Like Gretzky and Kurri during the 80s.
How is it that Potvin falls 40 behind Trottier and Bossy when they all played for the same team(Short of Trottier's 3 Pens years)?
+/-
Bossy:381
Trottier:452
Potvin:460
Adjusted +/-
Bossy:293
Trottier:292
Potvin:253
Potvin played his first 2 years without either of them, and the Isles team was, well, not great back then.
Trottier must be getting some bonus points from his last 2 Isles years, although not much since he was middling/Worst on his team for +/- those years, But I cannot see him getting points for those Pens years because the team +/- was not bad for those teams, but Trottier was twice among the worst for +/-.
Bossy, well, always played for the Isles when they were a great team.
How is it that Potvin falls 40 behind Trottier and Bossy when they all played for the same team(Short of Trottier's 3 Pens years)?
+/-
Bossy:381
Trottier:452
Potvin:460
Adjusted +/-
Bossy:293
Trottier:292
Potvin:253
Potvin played his first 2 years without either of them, and the Isles team was, well, not great back then.
Trottier must be getting some bonus points from his last 2 Isles years, although not much since he was middling/Worst on his team for +/- those years, But I cannot see him getting points for those Pens years because the team +/- was not bad for those teams, but Trottier was twice among the worst for +/-.
Bossy, well, always played for the Isles when they were a great team.
Good question. There are a couple of reasons for this.
1. Bossy didn't play on the penalty kill. This means that he has SH goals against in his raw +/-, but not SH goals for. After I estimate and remove these, he gains +33, and Trottier drops -1 and Potvin -6. This brings Bossy within 40 of the other two.
2. Trottier actually had a better adjusted +/- during Bossy's career than Bossy did (+311). Before 1978 he had a raw +/- of 56, but adjusted +/- of 31. After 1987 he had a raw +/- of -26, and adjusted +/- of -56. These adjusted down because these teams were still very good when he was off the ice, even during these years. So you are right that the Pens years are actually dragging him down. If you look at his career adjusted +/- through 1987, it is +342.
3. Potvin has a larger team adjustment to his +/- than Trottier or Bossy do. This is because he was a defenceman, played more ice time, and was on the ice for many more goals for and against. His expected plus-minus is higher for that reason.
For example, Potvin in 1978 and Bossy in 1979 have similar plus-minuses: +63 for Bossy and +57 for Potvin. But in 1978, Potvin was on the ice for 200 ESGF and ESGA combined. In 1979, Bossy was on the ice for 141 ESGF and ESGA combined. As a result, Potvin has an expected +/- of +26, and Bossy has an expected +/- of +18. After all adjustments, Potvin is +30 for 1978 and Bossy is +43 for 1979.
Potvin had a significantly larger team adjustment for his career, with an expected +/- of +166 compared to Bossy's expected +/- of +78, and the main reason for this is the fact that Potvin was a defenceman and played a lot more ice time.
Giving this great thread a bump with a couple of questions.
overpass, is there a way to separate the expected numbers into the constituent GF and GA components? For example, Joe Reekie scores very well by this metric but it would be interesting to know if the reason is that he was on the ice for many fewer goals against relative to what was expected or because he was fortunate enough to play with good forwards who scored a lot of GF while he was on the ice.
Do you think that the adjustments you make to ESGF and ESGA are influenced by the era in which a player played (maybe because there is more/less parity in different eras or something else)? For example, most of the best ESGA numbers are from players who spent at least part of their career during the late seventies or just prior to the recent lockout.
First, I'll note that I've posted the full season and career numbers for this metric on the HAG yahoo group, if anyone is interested in getting the full numbers.
Quote:
Originally Posted by pitseleh
overpass, is there a way to separate the expected numbers into the constituent GF and GA components? For example, Joe Reekie scores very well by this metric but it would be interesting to know if the reason is that he was on the ice for many fewer goals against relative to what was expected or because he was fortunate enough to play with good forwards who scored a lot of GF while he was on the ice.
Yes, I think so, but only for years where ice-time was recorded. Without recorded ice time, I don't think it's possible to determine from the numbers whether scoring or goal prevention is responsible for a player's plus-minus. In some cases we can make a pretty good guess from knowing how the player played, but that means the numbers don't really give us any more information.
I wouldn't feel confident using an ice-time estimator, either. I understand those are based heavily on the on-ice scoring numbers, so I can't see how they could provide any new information.
I'd love to be able to separate the expected numbers into GF and GA, as I think it would be considerably more informative and accurate, but I don't think the data is available to do so until the last 10 years.
In the specific case of Reekie, we could examine his last few years in Washington with ice-time data and see if it yields any insight.
Quote:
Originally Posted by pitseleh
Do you think that the adjustments you make to ESGF and ESGA are influenced by the era in which a player played (maybe because there is more/less parity in different eras or something else)? For example, most of the best ESGA numbers are from players who spent at least part of their career during the late seventies or just prior to the recent lockout.
It's likely that there are era influences. First, I adjust based on league even-strength scoring levels. I could also have adjusted for league scoring levels, which would probably be more accurate from a "value" perspective, but less so from an "ability" perspective. So that method of adjustment could be done differently, but I'm not sure that would cause consistent era effects.
Parity is certainly a big deal. Many of the best numbers come from the seventies, and those must be influenced by the fact that there were many great teams and terrible teams. Distribution of even-strength icetime may also make a difference - did teams roll four lines, or play their top line half the time? I'm not sure what would cause the effects just prior to the recent lockout, but it may be a change in coaching tactics. Possibly some of the top players were sent out for a lot of offensive zone faceoffs and few defensive zone faceoffs, and the nature of the 5-on-5 game at the time, with the neutral zone completely clogged up, meant that you were unlikely to be scored on if you started in the offensive zone.
It appears to me that star forwards in the 1980s and early 1990s generally have worse ESGA and adjusted plus-minus than their counterparts of the post-lockout period. I don't think that comes from the way I adjust, it's in the raw data. My guess is that it may have been a change in the way coaches matched lines and how they leveraged the ice time of their offensive players, but I'm not sure of that.
One more thought - goalies will influence the numbers of specific players. Maybe the fact that there are were more standout goalies in the seventies and nineties plays a part?
Anyway, I can't say for sure what the causes of the apparent era effects are, but I agree that they may be there.
Yes, I think so, but only for years where ice-time was recorded. Without recorded ice time, I don't think it's possible to determine from the numbers whether scoring or goal prevention is responsible for a player's plus-minus. In some cases we can make a pretty good guess from knowing how the player played, but that means the numbers don't really give us any more information.
Thanks for the quick reply. It's unfortunate that we've only really been getting good data for such a short period of time because I do think it could give you a lot of information. With guys like Marek Malik we know that they rode the coattails of a great forward line in front of them but it's harder to figure that type of stuff out for historical defense/forward pairings because I don't think anyone's memory is really good enough to be sure about something like that.
For example, Joe Reekie scores very well by this metric but it would be interesting to know if the reason is that he was on the ice for many fewer goals against relative to what was expected or because he was fortunate enough to play with good forwards who scored a lot of GF while he was on the ice.
I was curious about Reekie, so I ran his numbers by ice time for the four years from 1997-98 until 2000-01. His career adjusted +/- is 185, and in these 4 years it was +78, so this breakdown should provide some insight into a substantial portion of his career.
He played all 4 years on a Washington team that had a very consistent defence corps. Calle Johansson, Sergei Gonchar, Brendan Witt, and Ken Klee also played on that team for the same 4 years, so I'll compare his numbers to their numbers.
Year
Player
ESGF/60
ESGA/60
1998
Joe Reekie
2.37
1.88
1998
Ken Klee
1.79
2.28
1998
Calle Johansson
1.73
2.52
1998
Sergei Gonchar
2.68
2.57
1998
Mark Tinordi
3.29
2.79
1998
Phil Housley
2.29
2.84
1998
Brendan Witt
2.01
2.91
Year
Player
ESGF/60
ESGA/60
1999
Brendan Witt
1.55
2.11
1999
Calle Johansson
2.69
2.13
1999
Joe Reekie
2.59
2.25
1999
Sergei Gonchar
2.52
2.30
1999
Ken Klee
2.00
2.50
1999
Dmitri Mironov
2.19
2.60
1999
Mark Tinordi
1.98
2.62
Year
Player
ESGF/60
ESGA/60
2000
Joe Reekie
3.01
1.76
2000
Ken Klee
2.23
2.00
2000
Calle Johansson
2.52
2.18
2000
Brendan Witt
2.23
2.20
2000
Sergei Gonchar
3.63
2.32
2000
Dmitri Mironov
2.87
2.47
Year
Player
ESGF/60
ESGA/60
2001
Calle Johansson
2.37
1.91
2001
Joe Reekie
2.64
2.02
2001
Sergei Gonchar
2.85
2.09
2001
Brendan Witt
2.12
2.17
2001
Ken Klee
2.05
2.52
2001
Sylvain Cote
2.43
2.65
2001
Dmitri Mironov
2.11
2.94
Finally, the average numbers of the D-men who played multiple years in Washington during this time. This is simply the average of the seasonal numbers.
Player
Years
ESGF/60
ESGA/60
Joe Reekie
1998-2001
2.65
1.98
Calle Johansson
1998-2001
2.33
2.18
Sergei Gonchar
1998-2001
2.92
2.32
Ken Klee
1998-2001
2.02
2.33
Brendan Witt
1998-2001
1.98
2.35
Dmitri Mironov
1999-2001
2.39
2.67
Mark Tinordi
1998-1999
2.64
2.71
Reekie had the best defensive numbers of the group, and this was a very solid group of defencemen. He played a major role on the penalty kill as well, suggesting that his coach relied on him to play a defensive role. By the way, Washington in 1998-99 had one of the great penalty kills of all-time, with 14 shorthanded goals against only 39 PPGA, and Reekie was on the first unit.
His offensive numbers are better than one would expect - only Gonchar is better. Either Reekie was an underrated offensive defenceman (unlikely considering his points scored and reputation), or he was fortunate in the forwards he played with.
I think he was a very good defensive defenceman, and likely underrated. I also think that he was probably lucky to some degree in that he was on the ice with good offensive players or other players simply had offensive success when he was on the ice. A list of the top adjusted +/- players will select for two things - the best players and the luckiest players. I think most of the players in the top 50 are the best, but it's likely that at least a few were lucky, and Reekie probably falls into this category. I still think he was very good and underrated, but I wouldn't take his overall adjusted +/- at face value.
Anyway, maybe that's more than you really wanted to know about Joe Reekie, but that's the kind of analysis that can be done when ice time is included. I've fooled around with including ice time in the adjusted plus-minus stat in the past, but for simplicity's sake I won't introduce a whole new stat just to answer this question.
I changed the way I calculate the numbers slightly, so there's a minor change in the numbers from last year's post.
Stats Glossary Seasons: Season Fractions. 1.00 is a full season. I prefer it to games played because it gives a 48 game season, a 74 game season, an 80 game season or an 82 game season the same weight. $ESGF: Even-strength goals for, normalized to a 200 ESG scoring environment and with estimated SH goals removed. $ESGA: Even-strength goals against, normalized to a 200 ESG scoring environment and with estimated SH goals removed. R-ON: Even strength GF/GA ratio when the player is on the ice. R-OFF: Even-strength GF/GA ratio when the player is off the ice. XEV+/-: Expected even-strength plus-minus, which is an estimate of the plus-minus that an average player would post with the same teammates. Based on R-OFF. EV+/-: Even –strength plus-minus, which is simply plus-minus with estimated shorthanded goals removed and normalized to a 200 ESG environment. AEV+/-: Adjusted even-strength plus-minus, which is even-strength plus-minus minus expected even-strength plus-minus. This is the final number.
First, here are the top numbers for players who were active in 2009.
Rk
Player
Seasons
$ESGF/G
$ESGA/G
R-ON
R-OFF
XEV+/-
EV+/-
AEV+/-
/82
1
Teemu Selanne
14.15
1.00
0.81
1.24
0.84
-130
223
352
25
2
Nicklas Lidstrom
16.57
1.19
0.83
1.43
1.28
239
490
250
15
3
Keith Tkachuk
14.20
0.95
0.84
1.13
0.85
-116
127
243
17
4
Sergei Fedorov
15.60
0.92
0.68
1.35
1.09
61
303
242
15
5
Joe Thornton
10.20
1.00
0.75
1.33
0.94
-34
207
241
24
6
Mike Modano
17.36
0.86
0.69
1.25
1.03
22
244
222
13
7
Jarome Iginla
11.49
0.93
0.79
1.17
0.87
-80
130
210
18
8
Milan Hejduk
9.55
0.99
0.69
1.43
1.05
22
232
210
22
9
Alex Tanguay
8.04
1.07
0.70
1.52
1.09
33
243
209
26
10
Simon Gagne
7.39
0.89
0.57
1.56
0.95
-16
192
209
28
11
Patrik Elias
10.02
0.88
0.57
1.56
1.13
51
259
208
21
12
Mats Sundin
16.82
0.96
0.85
1.13
0.94
-56
152
208
12
13
Jere Lehtinen
9.96
0.81
0.52
1.57
1.09
32
239
207
21
14
Daniel Alfredsson
11.37
0.97
0.73
1.32
1.05
29
218
189
17
15
Chris Chelios
20.59
0.99
0.78
1.28
1.18
173
362
189
9
16
Sergei Gonchar
11.60
1.02
0.88
1.17
0.93
-48
139
187
16
17
Michael Nylander
11.23
0.87
0.71
1.23
0.94
-33
151
184
16
18
Paul Kariya
11.55
0.97
0.85
1.14
0.89
-68
114
182
16
19
Chris Pronger
12.81
1.01
0.84
1.20
0.99
-4
177
181
14
20
Marek Malik
8.44
0.94
0.71
1.32
0.95
-20
158
177
21
21
Brendan Shanahan
19.04
0.85
0.70
1.22
1.09
73
243
170
9
22
Brian Rafalski
8.44
1.08
0.76
1.42
1.13
52
220
168
20
23
Jason Arnott
12.97
0.89
0.74
1.21
1.00
0
167
167
13
24
Pavel Datsyuk
6.41
1.02
0.61
1.67
1.17
48
215
167
26
25
Pavol Demitra
10.12
0.96
0.73
1.31
1.04
20
186
166
16
26
Dany Heatley
6.18
1.10
0.91
1.21
0.83
-65
97
162
26
Not surprisingly, Nicklas Lidstrom leads this category. He's certainly the most accomplished skater in the league, and has posted some of his best even strength numbers late in his career. While his career +/- of 419 was certainly helped by his teammates, even after adjustments it's still very impressive.
Edit: Oops! I missed Selanne on the active list. He jumps to the top of the active players list. I think Lidstrom's plus-minus may be more impressive than Selanne's, as Lidstrom has generally played against the league's best while Selanne has played fairly easy minutes, at least later in his career.
Here are the top post-lockout performers in AEV+/-.
Rk
Player
Seasons
$ESGF/G
$ESGA/G
R-ON
R-OFF
XEV+/-
EV+/-
AEV+/-
/82
1
Joe Thornton
3.99
1.08
0.65
1.67
0.88
-25
141
166
42
2
Pavel Datsyuk
3.87
1.15
0.58
2.01
1.10
19
183
164
43
3
Alexander Ovechkin
3.95
1.19
0.94
1.27
0.81
-50
83
133
34
4
Dany Heatley
3.87
1.15
0.78
1.47
0.99
-2
117
119
31
5
Nicklas Lidstrom
3.83
1.22
0.69
1.77
1.30
55
167
112
29
6
Sidney Crosby
3.54
1.12
0.85
1.32
0.85
-33
79
112
32
7
Jaromir Jagr
3.00
1.09
0.68
1.61
0.95
-8
102
109
36
8
Ryan Getzlaf
3.62
0.82
0.50
1.63
0.97
-5
94
99
27
9
Teemu Selanne
3.09
0.89
0.52
1.72
0.97
-4
94
99
32
10
Daniel Sedin
3.99
0.83
0.57
1.46
0.92
-13
85
98
25
11
Henrik Sedin
4.00
0.87
0.61
1.42
0.92
-14
84
98
24
12
Daniel Alfredsson
3.70
1.07
0.74
1.45
1.02
4
101
97
26
13
Simon Gagne
3.07
0.93
0.70
1.32
0.76
-39
56
96
31
14
Henrik Zetterberg
3.56
1.04
0.61
1.72
1.23
35
128
93
26
15
Jason Arnott
3.57
1.01
0.72
1.39
0.98
-4
83
87
24
16
Jason Spezza
3.57
1.09
0.79
1.38
1.02
3
88
85
24
17
Marian Hossa
3.76
0.93
0.70
1.32
0.92
-14
70
84
22
18
Tom Preissing
3.09
0.91
0.65
1.40
0.89
-17
66
82
27
19
Brian Rafalski
3.84
1.03
0.76
1.35
1.02
4
83
79
21
20
Jarome Iginla
3.85
1.03
0.78
1.31
0.99
-2
77
79
21
Thornton and Datsyuk stand out as the cream of the crop here. Datsyuk has had better raw numbers, but his teams have been significantly better than Thornton's as measured by off-ice numbers.
Thornton and Heatley should have ample opportunity to continue their success in this area if they play together this season in San Jose.
It's clear that the post-lockout era has been dominated by forwards. Lidstrom is the only defenceman in the top 15.
Finally, here are the top 40 in AEV+/- since 1968.
Rk
Player
Seasons
$ESGF/G
$ESGA/G
R-ON
R-OFF
XEV+/-
EV+/-
AEV+/-
/82
1
Ray Bourque
20.30
1.16
0.85
1.37
0.95
-58
523
581
29
2
Bobby Orr
7.68
1.81
0.84
2.16
1.09
51
611
560
73
3
Jaromir Jagr
15.94
1.25
0.91
1.38
0.95
-51
451
502
32
4
Wayne Gretzky
18.82
1.40
1.12
1.26
1.04
47
445
398
21
5
Larry Robinson
17.34
1.31
0.82
1.60
1.34
307
696
389
22
6
Mark Howe
11.70
1.11
0.75
1.49
0.96
-23
350
373
32
7
Eric Lindros
9.63
1.24
0.82
1.52
0.94
-33
336
368
38
8
Bobby Clarke
14.49
0.87
0.48
1.81
1.19
100
462
362
25
9
Al Macinnis
17.72
1.10
0.78
1.41
1.12
105
467
361
20
10
Borje Salming
14.37
1.17
1.02
1.14
0.82
-181
172
353
25
11
Teemu Selanne
14.15
1.00
0.81
1.24
0.84
-130
223
352
25
12
Dave Taylor
13.84
0.88
0.67
1.31
0.84
-106
234
340
25
13
Mario Lemieux
11.29
1.30
1.07
1.21
0.85
-122
207
329
29
14
Peter Forsberg
9.02
1.15
0.66
1.74
1.08
34
362
328
36
15
Marcel Dionne
16.92
0.93
0.84
1.11
0.80
-194
124
317
19
16
Mike Bossy
9.40
1.08
0.60
1.80
1.17
73
370
297
32
17
Bryan Trottier
15.96
0.94
0.63
1.50
1.17
111
407
297
19
18
John Leclair
12.17
1.01
0.67
1.50
1.07
39
335
296
24
19
Guy Lafleur
14.14
1.07
0.64
1.67
1.35
206
496
289
20
20
Larry Murphy
20.35
1.06
0.88
1.20
1.02
22
298
277
14
21
Ron Francis
21.68
0.89
0.80
1.11
0.90
-115
160
276
13
22
Denis Potvin
13.27
1.18
0.79
1.49
1.23
156
420
265
20
23
Zigmund Palffy
8.63
1.02
0.81
1.26
0.79
-106
150
256
30
24
Nicklas Lidstrom
16.57
1.19
0.83
1.43
1.28
239
490
250
15
25
Steve Larmer
12.88
0.84
0.64
1.32
0.94
-36
214
249
19
26
Brad Mccrimmon
15.45
0.94
0.65
1.44
1.18
115
362
247
16
27
Brad Park
14.10
1.21
0.86
1.40
1.20
155
400
245
17
28
Keith Tkachuk
14.20
0.95
0.84
1.13
0.85
-116
127
243
17
29
Sergei Fedorov
15.60
0.92
0.68
1.35
1.09
61
303
242
15
30
Joe Thornton
10.20
1.00
0.75
1.33
0.94
-34
207
241
24
31
Scott Stevens
20.53
1.13
0.86
1.31
1.19
202
443
241
12
32
Brian Propp
12.65
0.85
0.57
1.48
1.12
58
284
226
18
33
Steve Shutt
11.66
0.94
0.52
1.81
1.45
181
405
225
19
34
Mike Modano
17.36
0.86
0.69
1.25
1.03
22
244
222
13
35
Charlie Simmer
8.90
0.82
0.62
1.33
0.83
-69
148
218
24
36
Craig Ramsay
13.44
0.72
0.48
1.50
1.11
50
266
216
16
37
Pierre Turgeon
16.28
0.90
0.72
1.25
1.03
24
236
211
13
38
Jean Ratelle
13.04
0.99
0.65
1.51
1.27
145
357
211
16
39
Gary Roberts
15.08
0.87
0.65
1.34
1.10
62
273
211
14
40
Jarome Iginla
11.49
0.93
0.79
1.17
0.87
-80
130
210
18
No active players are high on the career list. (Edit: other than Selanne, who probably won't move very far.) Lidstrom can move into the top 20 if he continues to defy the aging process and produce. If Joe Thornton continues on his recent pace, he may crack the top 10 in a few years.
It certainly says something for players so good offensively that its hard to score while they are on the ice, because Thornton is nowhere near as good as Datsyuk defensively(Although Thornton is not bad. He backchecks, but makes those boneheaded thread the needle cross ice passes that get intercepted often and lead to turnovers).
I find it shocking that he came #2 post lockout, but I suppose its not so strange given his game controlling style.
But, I'm not sure that all seasons should be weighted the same. I think all games weighted the same worked fine. Now each time a player played a game in the 1995 season, it was like 1.7 games.
This is a minor complaint, mind you, and I always like reading your work.
Do you really only have 368 posts? Or did that get reset at some point? I swear I've read more than 368 of your posts.
But, I'm not sure that all seasons should be weighted the same. I think all games weighted the same worked fine. Now each time a player played a game in the 1995 season, it was like 1.7 games.
This is a minor complaint, mind you, and I always like reading your work.
Do you really only have 368 posts? Or did that get reset at some point? I swear I've read more than 368 of your posts.
Yes, I've started to think recently also that 1995 should be treated differently from other seasons. Maybe in my next update.
First, I'll note that I've posted the full season and career numbers for this metric on the HAG yahoo group, if anyone is interested in getting the full numbers.
It's likely that there are era influences. First, I adjust based on league even-strength scoring levels. I could also have adjusted for league scoring levels, which would probably be more accurate from a "value" perspective, but less so from an "ability" perspective. So that method of adjustment could be done differently, but I'm not sure that would cause consistent era effects.
Parity is certainly a big deal. Many of the best numbers come from the seventies, and those must be influenced by the fact that there were many great teams and terrible teams. Distribution of even-strength icetime may also make a difference - did teams roll four lines, or play their top line half the time? I'm not sure what would cause the effects just prior to the recent lockout, but it may be a change in coaching tactics. Possibly some of the top players were sent out for a lot of offensive zone faceoffs and few defensive zone faceoffs, and the nature of the 5-on-5 game at the time, with the neutral zone completely clogged up, meant that you were unlikely to be scored on if you started in the offensive zone.
It appears to me that star forwards in the 1980s and early 1990s generally have worse ESGA and adjusted plus-minus than their counterparts of the post-lockout period. I don't think that comes from the way I adjust, it's in the raw data. My guess is that it may have been a change in the way coaches matched lines and how they leveraged the ice time of their offensive players, but I'm not sure of that.
One more thought - goalies will influence the numbers of specific players. Maybe the fact that there are were more standout goalies in the seventies and nineties plays a part?
Anyway, I can't say for sure what the causes of the apparent era effects are, but I agree that they may be there.
In response to questions about influences and how the game has changed.
Three lines were the norm from the fifties into the nineties. Four lines is a relatively new phenomena. During this period the extra forward(s) would be rotated thru the top three lines on a need or situational basis. The extra forward(s) also had situational responsibilities PK, PP, desire to change momentum etc. Finally the length of shifts has changed drastically, 30-45 second shifts are often the norm.Going back into the fifties shifts would run well over a minute at times. The top line would often get longer shifts especially in the 1950's thru the 1970's, certain players would get extended shift considerations - Phil Esposito with Boston being the prime example.
Defensemen, especially the 4 - 6 defensemen have to be viewed differently.From the fifties into the seventies it was mainly a two pairing defensive rotation with the fifth defenseman being a sub with situational responsibilities. Once these responsibilities became strategic your plus / minus figures became impacted. Certain coaches - especially with the Canadiens(Blake, Bowman) saw the advantage of having one of the 4 - 6th defensemen play the enforcer role.
Looking at the 1976-79 Canadiens will illustrate this approach. Pierre Bouchard and Gilles Lupien filled this role.Other than Mario Tremblay who could fight a bit, none of the Canadiens forwards could fill the enforcer role. So Bowman would play Bouchard or Lupien, when necessary, with one of the Big 3 to physically support his forward lines. As a result Bouchard had nice +/- totals with the Canadiens but these numbers reversed when they were traded to other teams where they were expected to play.
A few other comments about factors that influence the +/-.
The game is being micro-coached more and more. Players are used situationally - centers on faceoffs given the success rate against other centers,RHS or LHS center, where the faceoff is held, etc.
Changing lines on the go or simply players on the go has evolved.
The coincidental minor rule. 4 on 4 or 3 on 3 hockey is rarer than it used to be. This has an impact on ice time and +/-.
The use of video as a coaching aid and a statistical aid.
Far from convinced that all these factors can be accounted for from a historic perspective.
Last edited by Canadiens1958: 09-21-2009 at 06:41 AM.
Reason: wording
Slightly surprised that the difference between Datsyuk and Zetterberg is so huge since the LO. Only shows how dominant Datsyuk has been in last couple of years.
And talk about Detroit having depth. R-OFF is better than 1.00 for every player.
Also sort of unexpected too see Thornton being #1. But actually it makes sense. His first two seasons were superb.
Hmm, surprised to see Selänne so high on this list, actually higer than any other still active player. I guess I'll have to read the thread a bit to understand how these have been calculated.
The game is being micro-coached more and more. Players are used situationally - centers on faceoffs given the success rate against other centers,RHS or LHS center, where the faceoff is held, etc.
Changing lines on the go or simply players on the go has evolved.
Thanks for your comments. I think this point is very important with regard to plus-minus ratings from the past few years. For example, since the lockout, Teemu Selanne has gone out for many more offensive zone faceoffs than defensive zone faceoffs. He never gets onto the ice while changing on the fly except when the puck is going in the right direction. This is a good use of his talent by the coach, but it does tend to skew the plus-minus to reflect usage as much as talent.
I think this plus-minus method works better for the 70s and 80s, as shifts were generally longer and coaches did less micro-coaching, especially in the regular season. For more recent years it's even more important to know more about the context in which the player put up these numbers, and his role on the team.
Thanks for your comments. I think this point is very important with regard to plus-minus ratings from the past few years. For example, since the lockout, Teemu Selanne has gone out for many more offensive zone faceoffs than defensive zone faceoffs. He never gets onto the ice while changing on the fly except when the puck is going in the right direction. This is a good use of his talent by the coach, but it does tend to skew the plus-minus to reflect usage as much as talent.
I think this plus-minus method works better for the 70s and 80s, as shifts were generally longer and coaches did less micro-coaching, especially in the regular season. For more recent years it's even more important to know more about the context in which the player put up these numbers, and his role on the team.
Would take the micro-coaching even further. Factor in tailored to the opposition, the score, the period (position of the bench, in the offensive or defensive zone), left wing or right wing faceoff in the offensive or defensive zone.
As for the actual stat (adjusted or not), think one has to be very careful how it is interpreted. Describing, evaluating or rating are not interchangeable concepts.
