Adjusted Even-Strength Plus-minus 1960-2017

[Dark Shadows]

For those specific players, I'd take it with a grain of salt named Eric Lindros or Marcel Dionne.

Yeah, every stat needs to be interpreted, and this one does have some funny results (I'm not sure what to make of Dmitri Khristich or Joe Reekie in the top 50.) But I do think it's far better than unadjusted plus-minus, as the Francis example shows.

By that same token, It would be interesting to see Francis numbers in the years he played in Pittsburgh as opposed to his other teams.

I would be particularly interested in the years he played with Jagr and Lemieux. 1994-95, 95-96, 96-97 and 97-98

 

[seventieslord]

Joe Reekie was a great defensive player who frequently had a very good +/- compared to the rest of his team, despite having very few points.

 

[overpass]

By that same token, It would be interesting to see Francis numbers in the years he played in Pittsburgh as opposed to his other teams.

I would be particularly interested in the years he played with Jagr and Lemieux. 1994-95, 95-96, 96-97 and 97-98

Player | Year | SFrac | ESGF/G | ESGA/G | R-ON | R-OFF | XEV+/- | EV+/- | AdjEV+/- | /Season
Ron Francis | 82-91 | 8.93 | 0.82 | 0.79 | 1.04 | 0.80 | -101 | 23 | 124 | 13.9
Ron Francis | 91-94 | 3.03 | 0.90 | 0.85 | 1.06 | 1.21 | 27 | 12 | -15 | -5.0
Ron Francis | 95-98 | 3.83 | 1.28 | 0.89 | 1.43 | 0.99 | -3 | 130 | 134 | 34.9
Ron Francis | 99-04 | 5.90 | 0.75 | 0.76 | 0.99 | 0.85 | -45 | 2 | 47 | 8.0
Career | 82-04 | 21.68 | 0.89 | 0.82 | 1.09 | 3.85 | -123 | 166 | 290 | 13.4

In the table above, the first time period is the Hartford years only, and the second time period includes his Pittsburgh games from 1991 and runs through 1994.

Ron Francis does have a couple of factors that would cause the numbers to rate him incorrectly. First, during the 1991-1994 Pittsburgh years, the -15 adjusted plus-minus probably underrates him. Why? Mario Lemieux was part of his off-ice baseline for comparison. If I was going to do an in-depth analysis of Francis, I'd run the numbers again with Mario's numbers removed from the off-ice sample. Second, the fact that he played on a line with Jagr from 1995-1998 means that his +134 adjusted probably overrates him. Notice the spike in even-strength goals per game, no doubt influenced by Jagr.

I'm not aware of any particular biases on his Hartford or Carolina numbers, as there were no superstars of the Lemieux/Jagr type there. His adjusted +/- was 171 in ~15 years there.

 

[Dark Shadows]

Player | Year | SFrac | ESGF/G | ESGA/G | R-ON | R-OFF | XEV+/- | EV+/- | AdjEV+/- | /Season
Ron Francis | 82-91 | 8.93 | 0.82 | 0.79 | 1.04 | 0.80 | -101 | 23 | 124 | 13.9
Ron Francis | 91-94 | 3.03 | 0.90 | 0.85 | 1.06 | 1.21 | 27 | 12 | -15 | -5.0
Ron Francis | 95-98 | 3.83 | 1.28 | 0.89 | 1.43 | 0.99 | -3 | 130 | 134 | 34.9
Ron Francis | 99-04 | 5.90 | 0.75 | 0.76 | 0.99 | 0.85 | -45 | 2 | 47 | 8.0
Career | 82-04 | 21.68 | 0.89 | 0.82 | 1.09 | 3.85 | -123 | 166 | 290 | 13.4

In the table above, the first time period is the Hartford years only, and the second time period includes his Pittsburgh games from 1991 and runs through 1994.

Ron Francis does have a couple of factors that would cause the numbers to rate him incorrectly. First, during the 1991-1994 Pittsburgh years, the -15 adjusted plus-minus probably underrates him. Why? Mario Lemieux was part of his off-ice baseline for comparison. If I was going to do an in-depth analysis of Francis, I'd run the numbers again with Mario's numbers removed from the off-ice sample. Second, the fact that he played on a line with Jagr from 1995-1998 means that his +134 adjusted probably overrates him. Notice the spike in even-strength goals per game, no doubt influenced by Jagr.

I'm not aware of any particular biases on his Hartford or Carolina numbers, as there were no superstars of the Lemieux/Jagr type there. His adjusted +/- was 171 in ~15 years there.

Thanks for posting that

 

[Stonefly]

That's possible, although Howe did lead the Wings in scoring by a fair clip in 1968 without Mahovlich, and only Howie Young has a better unadjusted +/- on the team (and that team included Norm Ullman most of the year). In 1971 Howe was also finally breaking down some and retired after the season because of severe arthritis in his wrists that limited him all year. In 1970 Howe had a better +/- than Mahovlich by a decent margin as well. How was also still playing on the PK despite being 41. I'm sure Mahovlich had an impact, but I don't see him as the reason Howe's +/- relative to the team would be inflated.

As far as his prime +/- I think it would be much higher than you assume. Take 1952-53 for example. The team had 222GF and 133GA, a +89 margin. Howe finished that season with 95 points. He was on a stacked line there as well. I don't think it's outside the realm of reason to assume that his +/- that season was extremely high, even in relation to his teammates. Nobody on his team finished withing 24 points of him, leading me to believe he was double or triple shifted quite a bit that season (or he was scoring a ridiculous amount of unassisted goals). Sadly there will likely never be a way to know for sure, but I'd bet his adjusted +/- per season through his prime and even mid 30s was far higher than you might expect.

I wouldn't be at all surprised if this were true. Until Orr came along Howe was the most dominant player there was. For us kids Gordie was it. He was the man. It was incredible how he could control so much of what happened on the ice. When he had the puck it was virtually impossible to take from him. Because of that I have no doubt he would have had an excellent +/- in his prime. That being said, I don't think for a second that he would surpass Orr's numbers. As much as Howe dominated the ice it just wasn't on the same level Orr did. And it has nothing to do with the expansion of the league. It was evident in Orr's first season he was already better than everyone else and that as you know was a six team league.

 

[FissionFire]

I wouldn't be at all surprised if this were true. Until Orr came along Howe was the most dominant player there was. For us kids Gordie was it. He was the man. It was incredible how he could control so much of what happened on the ice. When he had the puck it was virtually impossible to take from him. Because of that I have no doubt he would have had an excellent +/- in his prime. That being said, I don't think for a second that he would surpass Orr's numbers. As much as Howe dominated the ice it just wasn't on the same level Orr did. And it has nothing to do with the expansion of the league. It was evident in Orr's first season he was already better than everyone else and that as you know was a six team league.

I find that hard to believe since he wasn't exactly setting himself apart from the pack as a rookie. Maybe it was evident that he would be better than everyone once he matured but I find it hard to believe he was clearly the best player in the NHL as an 18yr old rookie. Harry Howell scored about the same as Orr did that season. Gary Bergman on a bad Detroit team was only 6 points back. A 35yr old Pilote put up 11 more points. Orr's future greatness may have been evident but I strongly disagree that he was the best in the league as an 18yr old rookie in the 6 team NHL. Expansion didn't affect his talent, but it undoubtedly had a huge impact on his numbers.

EDIT: The voters that season would seem to agree with me since he was behind Pilote and Howell in All-Star balloting (probably Horton too).

 

[seventieslord]

I find that hard to believe since he wasn't exactly setting himself apart from the pack as a rookie. Maybe it was evident that he would be better than everyone once he matured but I find it hard to believe he was clearly the best player in the NHL as an 18yr old rookie. Harry Howell scored about the same as Orr did that season. A 39yr old Kelly was a few points behind. Gary Bergman on a bad Detroit team was only 6 points back. A 35yr old Pilote put up 11 more points. Orr's future greatness may have been evident but I strongly disagree that he was the best in the league as an 18yr old rookie in the 6 team NHL. Expansion didn't affect his talent, but it undoubtedly had a huge impact on his numbers.

You beat me to it. All true.

Orr was not the best player in the league, or the best defenseman, immediately. That happened the year after.

 

[MadArcand]

Thanks for posting that

Still need a grain of salt? He rates better in his HFD/CAR years than Oates and Hawerchuk do combined.

Another thing I noticed, I'm surprised at how well does Sundin do in this (behind Modano but well ahead of Gilmour and Roenick). Even more, Turgeon (no clue what could account for this. Can't recall him playing with anyone so great it'd skew those numbers much.) And yet more, Palffy (OTOH I guess being the best player on his team pretty much his whole career helps... but still).

 

[arrbez]

Palffy (OTOH I guess being the best player on his team pretty much his whole career helps... but still).

I don't think Palffy ever got the fanfare he deserved. Amazingly consistent no matter who he played with, but spent his whole career in the wastelands with NYI and LA. Not to mention that outside of his half season in Pittsburgh, his career perfectly spanned what most would consider the "dead puck era" (1996-2004). He was over a PPG every season but one (.94 in 2002) during this span. His production was pretty much in line with the Sundin's, Modano's, etc.

 

[Stonefly]

I find that hard to believe since he wasn't exactly setting himself apart from the pack as a rookie. Maybe it was evident that he would be better than everyone once he matured but I find it hard to believe he was clearly the best player in the NHL as an 18yr old rookie. Harry Howell scored about the same as Orr did that season. Gary Bergman on a bad Detroit team was only 6 points back. A 35yr old Pilote put up 11 more points. Orr's future greatness may have been evident but I strongly disagree that he was the best in the league as an 18yr old rookie in the 6 team NHL. Expansion didn't affect his talent, but it undoubtedly had a huge impact on his numbers.

EDIT: The voters that season would seem to agree with me since he was behind Pilote and Howell in All-Star balloting (probably Horton too).

You beat me to it. All true.

Orr was not the best player in the league, or the best defenseman, immediately. That happened the year after.

Coming from two guys who no doubt were not even born yet. No offence, but you will never understand how Orr dominated an entire game without having seen it. Stats and awards don't tell you that part. From the first game I ever saw him play he had the puck more than anyone else and controlled what happened when he was on the ice. And it was like that pretty much every game I saw him play until about 74/75 when it was clear his knee was affecting his skating.
As for the award voting, that he was recognized as much as he was is astounding. Do you have any idea how hard it was for rookies to get any kind of respect back then?

 

[Dark Shadows]

Still need a grain of salt? He rates better in his HFD/CAR years than Oates and Hawerchuk do combined.

Another thing I noticed, I'm surprised at how well does Sundin do in this (behind Modano but well ahead of Gilmour and Roenick). Even more, Turgeon (no clue what could account for this. Can't recall him playing with anyone so great it'd skew those numbers much.) And yet more, Palffy (OTOH I guess being the best player on his team pretty much his whole career helps... but still).

Its just more evidence that this needs to be taken with a grain of salt.

From what I can see, it really over-rewards playing with a weaker team.

There is still no way in hell anyone is going to convince me Francis is way better than Gilmour or Oates.

 

[MadArcand]

Its just more evidence that this needs to be taken with a grain of salt.

From what I can see, it really over-rewards playing with a weaker team.

There is still no way in hell anyone is going to convince me Francis is way better than Gilmour or Oates.

Not trying to do that. I see Oates close if behind Francis anyway. Same with Killer. And didn't even mention Gilmour, who scores fairly well here anyway. Just shows that the cries of 'Oates was Francis' equal defensively' are inane.

Tho Hawerchuk is indeed far and away behind Ron.

/end OT

 

[overpass]

Another thing I noticed, I'm surprised at how well does Sundin do in this (behind Modano but well ahead of Gilmour and Roenick). Even more, Turgeon (no clue what could account for this. Can't recall him playing with anyone so great it'd skew those numbers much.) And yet more, Palffy (OTOH I guess being the best player on his team pretty much his whole career helps... but still).

Sundin is very high, but I think he may deserve it. I think he's actually been underrated at times, although he played in Toronto. He's never really had the big season where he was top-5 in scoring, but he's very consistently played with very ordinary wingers against the opponents best lines and come out well ahead in his matchups. I think he added a huge amount of value to Toronto while he was there.

I'm not so sure about Turgeon. Before I ran the numbers I would have guessed that he'd be the type that wouldn't rate highly on this. Still, he played in a number of different teams and for different coaches, as a number one centre for most of his career, and is fairly consistently a plus player on his team. He probably takes a bigger knock than he should for his lack of physical play.

Palffy does look very good by these numbers. He was an excellent scorer who was underrated by the fact that he played in the dead puck era and he missed a lot of time with injuries. I believe he looks very good using adjusted points. Does the fact that he played on terrible teams cause this stat to overrate him? Possibly. I've tried to find the right amount to adjust but I may be wrong, and every situation is different so a one-size-fits-all adjustment won't be perfect. Even so, his EV plus-minus before adjusting for his terrible teammates is +149 in 684 games. I think he belongs on the leaderboard somewhere, although top-25 does seem a bit high.

Just another note on the Oates/Francis discussion - recent research has shown that players who take a higher percentage of defensive zone faceoffs than normal will have a lower plus-minus. This makes complete sense to me. Unfortunately, there's no data to adjust for that, so that's another subjective variable. Oates and Francis were both known as fine defensive players and faceoff men, so their plus-minus numbers may underrate them a bit relative to other star centres who didn't have that role. On the other side of this discussion, I believe Teemu Selanne is an example of a player who has taken many more offensive zone faceoffs since the lockout. His plus-minus post-lockout probably overrates him, although I can't say if this is the case for his whole career.

 

[canucks4ever]

Palffy is very underrated, he wont make the hall of fame but there is no doubt he was better than some guys in it. Steve Shutt was just based off Lafluer's success and he didn't do **** as soon as the dynasty was over. Mike Gartner was just good forever, and I dont care if Clark Gillies gets dubbed as a good leader, he is based off dynasty success too. Palffy scores regardless of what team he plays for, um lol these guys don't. Steve Shutt was garbage as soon as Lafluer's prime was over.

 

[seventieslord]

Those far right columns, the PP% and SH% are mighty interesting. But what should I read into them? Besides the obvious, what do they tell us?

 

[overpass]

Those far right columns, the PP% and SH% are mighty interesting. But what should I read into them? Besides the obvious, what do they tell us?

First, I think they can show how large a role a player had on special teams. I guess that's obvious, but it's important. The reason I included it here, although it has nothing to do with adjusted plus-minus, is to remind people that these players contributed more than their even-strength play, and to give a rough estimate of their role on special teams.

If you're looking for stats on a player's penalty killing value, SH% is about as good as you'll get for most of the time from 1968 until 1998, when icetime started to be recorded. Use SH% along with shorthanded goals and the overall performance of the team while shorthanded for an estimate of their shorthanded play. For example, you could say "Serge Savard has a SH% of 58%, so he played about 58% of the shorthanded minutes for an excellent penalty killing team, therefore he provided a great deal of value while shorthanded." It's a crude method of value, and I think anyone who has seen two players play will be able to make a better estimate of their penalty killing play. The advantage this has is for those of us who haven't seen all the players play to make quick and dirty comparisons between any player.

PP% doesn't have as much value, since power play points are also available. It does help to put the power play points in context. Why didn't Gary Roberts put up big power play numbers? He only had a PP% of 38%, so for most of his career he wasn't playing on the first unit powerplay. Why didn't Niklas Lidstrom score as many power play points as other top defencemen, even when adjusted for scoring level? He only had a PP% of 72%, meaning that he got 1st unit PP minutes, but he didn't play the whole powerplay like Bourque, Leetch, and MacInnis did for much of their career (PP% of 88%, 87%, and 87% respectively).

 

[seventieslord]

Would this be an accurate statement?

"under this adjusted system, it is easier to score high if on a good team or low if on a bad team, just not as easy as it is in regular +/-."

 

[overpass]

Would this be an accurate statement?

"under this adjusted system, it is easier to score high if on a good team or low if on a bad team, just not as easy as it is in regular +/-."

I hope not. My intention in designing this was to recognize the plus-minus contributions of players on bad teams and on good teams equally.

That said, it may be the case. The method doesn't compare a player's on-ice plus-minus to his team's off-ice plus-minus, exactly, it compares it to somewhere between the off-ice numbers and league average. If it compared to teammates only, the 1976-77 Canadiens would have numbers equal to the 1974-75 Capitals, and that clearly isn't right. As a result, the average player on the Canadiens will score better than the average player on the Capitals. I think that's correct, as the average player on those Canadiens was undoubtedly far better than the average player on those Capitals.

The method may still favor players on good teams to some degree. However, another poster thought it favored players on bad teams. I hope it doesn't do either. If you think it is biased against players on bad teams or good teams, you can look at the R-ON and R-OFF columns in the tables above. Those are the GF/GA ratios for the team when the player was on the ice or off the ice. For example, Larry Robinson has an R-ON of 1.60, and an R-OFF of 1.33, so he played on excellent teams and outperformed them. Borje Salming has an R-ON of 1.15 and an R-OFF of .82, so he played on terrible teams and outperformed them by even more. Which is more impressive? My final number has them as almost even, but if you disagree with the method of comparing against the team you can just look at the R-ON and R-OFF.

 

[seventieslord]

I hope not. My intention in designing this was to recognize the plus-minus contributions of players on bad teams and on good teams equally.

That said, it may be the case. The method doesn't compare a player's on-ice plus-minus to his team's off-ice plus-minus, exactly, it compares it to somewhere between the off-ice numbers and league average. If it compared to teammates only, the 1976-77 Canadiens would have numbers equal to the 1974-75 Capitals, and that clearly isn't right. As a result, the average player on the Canadiens will score better than the average player on the Capitals. I think that's correct, as the average player on those Canadiens was undoubtedly far better than the average player on those Capitals.

The method may still favor players on good teams to some degree. However, another poster thought it favored players on bad teams. I hope it doesn't do either. If you think it is biased against players on bad teams or good teams, you can look at the R-ON and R-OFF columns in the tables above. Those are the GF/GA ratios for the team when the player was on the ice or off the ice. For example, Larry Robinson has an R-ON of 1.60, and an R-OFF of 1.33, so he played on excellent teams and outperformed them. Borje Salming has an R-ON of 1.15 and an R-OFF of .82, so he played on terrible teams and outperformed them by even more. Which is more impressive? My final number has them as almost even, but if you disagree with the method of comparing against the team you can just look at the R-ON and R-OFF.

No problem at all. I thought that was the case, and that's what you should aim to do. An average Hab shouldn't end up with the same rating as an average Capital.

 

[Dark Shadows]

Just out of curiosity, how is it that McCrimmon and Howe end up with such a large gap?
NHL wise
McCrimmon was a career +444, Howe a career + 400. They played on the same teams for 6 years. What gives them such a gigantic difference over the adjusted method?

I am assuming you did not include his WHA years since those would likely have severely skewed the results. He always played for a top team in the WHA and they had very inflated +/-'s due to the weakness of the league.

Heck, the Whalers were 27-30+ points ahead of the league average in their last 2 WHA years, and immediately dropped to a below average team in the NHL, despite having virtually the same team(They kept 17 of their 25 Roster players and replaced the 8 they lost with equal ones)

 

[overpass]

Just out of curiosity, how is it that McCrimmon and Howe end up with such a large gap?
NHL wise
McCrimmon was a career +444, Howe a career + 400. They played on the same teams for 6 years. What gives them such a gigantic difference over the adjusted method?

I am assuming you did not include his WHA years since those would likely have severely skewed the results. He always played for a top team in the WHA and they had very inflated +/-'s due to the weakness of the league.

Heck, the Whalers were 27-30+ points ahead of the league average in their last 2 WHA years, and immediately dropped to a below average team in the NHL, despite having virtually the same team(They kept 17 of their 25 Roster players and replaced the 8 they lost with equal ones)

Basically, Howe was on a poor Hartford team for 3 years while McCrimmon was on a very good Boston team, and later Howe was on Philadelphia after they had dropped off while McCrimmon spent 3 years on a great Calgary team.

They did have their best years together, and both had good numbers outside of these peak years, but Howe had those numbers on average to poor teams, and McCrimmon on very good to great teams.

No, I didn't include the WHA years in here. I haven't done any calculations for the WHA.

 

[Dark Shadows]

Basically, Howe was on a poor Hartford team for 3 years while McCrimmon was on a very good Boston team, and later Howe was on Philadelphia after they had dropped off while McCrimmon spent 3 years on a great Calgary team.

They did have their best years together, and both had good numbers outside of these peak years, but Howe had those numbers on average to poor teams, and McCrimmon on very good to great teams.

No, I didn't include the WHA years in here. I haven't done any calculations for the WHA.

I thought the last few years they had might even it out a bit. McCrimmon was on a laughably bad Hartford team while Howe was on the best team in the league

 

[VanIslander]

Lehtinen is ahead of Sakic

 

[Enstrom39]

This is terrific. You might consider posting it on the yahoo Hockey Analytics Groug (H.A.G.) on yahoo groups. I'm sure you're get some more comments.

 

[overpass]

I thought the last few years they had might even it out a bit. McCrimmon was on a laughably bad Hartford team while Howe was on the best team in the league

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)

Lehtinen is ahead of Sakic

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.