All Purpose Analytics and Extended Stats Discussion

[twabby]

@twabby I need some analytics backing up my love for Orlov! No trolling. I think he’s a stud and so underrated.

Over the last 20 games (basically since he was broken up from his pairing with Carlson) he's been the second best Capital defenseman in terms of expected goals at 56%, and his actual goals are at 60%. Nick Jensen of course leads the Capitals in this regard right now at a massive 60% xGF%.

Orlov is kind of a jack of all trades type of player at this point in his career: decent to good at everything, not fantastic at anything. He has a higher than average relative expected goals for per 60 (+0.04 Rel xGF/60), a lower than average relative expected goals against per 60 (-0.03 Rel xGA/60, negative in this case is good), has a good penalty differential, can play on the PP but doesn't excel there, can play on the PK but is only used sparingly there, can play in 4v4 and 3v3 situations, can play a physical game but doesn't rely on it much, etc.

I don't think he's the best at any one thing really: Jensen is probably better defensively according to the numbers, Carlson is better offensively and on the PP, Chara and Jensen are probably better penalty killers, Dillon and Chara are probably more physical, etc. But there isn't really a game situation where he is weak.

His main detractors used to point to his giveaways as a negative mark against him, but again that is a big failing of the eye-test: most people remember the rare big events and let it weigh their opinions way too much, and don't really notice the numerous small things that add up to usually mean more than the big things when taken in totality.

Discussing Orlov was more interesting back in the day when a lot of people were asking for him to be exposed to the Vegas expansion draft in favor of protecting Karl Alzner instead. Brian MacLellan chose wisely, and I think nowadays he's a far less polarizing figure than he used to be.

 

[traparatus]

I wonder where Orlov stands in terms of value for cap dollar. I've always considered his contract to be of fair value but I've seen many people disagree with this assessment.

@twabby, do you know if anyone has done a performance vs cap comparison across the league? Something like value above replacement or xGF% compared to cap hit for all non-ELC defensemen.

 

[twabby]

I wonder where Orlov stands in terms of value for cap dollar. I've always considered his contract to be of fair value but I've seen many people disagree with this assessment.

@twabby, do you know if anyone has done a performance vs cap comparison across the league? Something like value above replacement or xGF% compared to cap hit for all non-ELC defensemen.

I think Dom Luszczyszyn has something like this over at The Athletic, but I'm not a subscriber so I don't have access to his numbers.

My suspicion is that his contract isn't really a bargain, but I don't think the Capitals have many bargains on the team except for Chara, Dowd, and Jensen. Like most of the team I think Orlov's contract is probably right in line with what he should be making, but again I'm not sure what some models think he should be making.

 

[francaisvolantsparis]

I wonder where Orlov stands in terms of value for cap dollar. I've always considered his contract to be of fair value but I've seen many people disagree with this assessment.

@twabby, do you know if anyone has done a performance vs cap comparison across the league? Something like value above replacement or xGF% compared to cap hit for all non-ELC defensemen.

You need the Expected WAR stat. Look to the good exaple of how to use it here: The cost of trading a bad Contract

 

[francaisvolantsparis]

This season estimation (really aproximate, based on hockey-reference site stat):

Player	POS	Point Shares	Correct cap number $M
Nicklas Backstrom	C	3.7	8.891
John Carlson	D	3.1	7.449
Jakub Vrana	LW	2.8	6.728
Alex Ovechkin	LW	2.5	6.008
Justin Schultz	D	2.5	6.008
Nick Jensen	D	2.3	5.527
Zdeno Chara	D	2.2	5.287
Tom Wilson	RW	1.9	4.566
Brenden Dillon	D	1.8	4.325
T.J. Oshie	C	1.6	3.845
Daniel Sprong	RW	1.3	3.124
Evgeny Kuznetsov	C	1.1	2.643
Conor Sheary	LW	1.1	2.643
Garnet Hathaway	RW	1.1	2.643
Lars Eller	C	1	2.403
Nic Dowd	C	0.9	2.163
Dmitry Orlov	D	0.8	1.922
Carl Hagelin	LW	0.5	1.202
Richard Panik	RW	0.4	0.961
Trevor van Riemsdyk	D	0	0.725

[TBODY] [/TBODY]

 

[CapitalsCupReality]

Heh, so Oshie is overpaid, Wilson too..that’s all I really need to see from that chart for validation.

not even worth calling out the other high priced eye sores listed....

 

[francaisvolantsparis]

the last season:

Player	POS	Point Shares	Correct cap number $M
John Carlson	D	10.7	10.977
Alex Ovechkin	LW	8.7	8.925
Jakub Vrana	LW	6.1	6.258
Braden Holtby	G	6	6.155
T.J. Oshie	C	5.5	5.642
Evgeny Kuznetsov	C	5.1	5.232
Nicklas Backstrom	C	4.6	4.719
Tom Wilson	RW	4.2	4.309
Dmitry Orlov	D	4	4.103
Michal Kempny	D	3.9	4.001
Lars Eller	C	3.6	3.693
Radko Gudas	D	3.2	3.283
Richard Panik	RW	2.5	2.565
Carl Hagelin	LW	2.3	2.360
Garnet Hathaway	RW	1.5	1.539
Nic Dowd	C	1.5	1.539
Nick Jensen	D	1.3	1.334
Travis Boyd	C	1.1	1.128
Brendan Leipsic	LW	0.5	0.700
Chandler Stephenson	C	0.4	0.700
Tyler Lewington	D	-0.1	0.700

[TBODY] [/TBODY]

 

[francaisvolantsparis]

2018/19

Player	POS	Point Shares	Correct cap number $M
Alex Ovechkin	LW	11	9.307
John Carlson	D	10.5	8.884
Braden Holtby	G	10.2	8.630
Evgeny Kuznetsov	C	6.8	5.754
Nicklas Backstrom	C	6.6	5.584
T.J. Oshie	C	5.6	4.738
Michal Kempny	D	5.3	4.484
Brett Connolly	RW	5.2	4.400
Tom Wilson	RW	4.4	3.723
Dmitry Orlov	D	4	3.384
Pheonix Copley	G	3.9	3.300
Matt Niskanen	D	3.6	3.046
Lars Eller	C	2.3	1.946
Andre Burakovsky	LW	2.2	1.861
Nic Dowd	C	2.1	1.777
Christian Djoos	D	2	1.692
Brooks Orpik	D	2	1.692
Travis Boyd	C	1.7	1.438
Carl Hagelin	LW	1.1	0.931
Dmitrij Jaskin	RW	0.1	0.675
Chandler Stephenson	C	0	0.675

[TBODY] [/TBODY]

 

[francaisvolantsparis]

2017/18

Player	POS	Point Shares	Correct cap number $M
Alex Ovechkin	LW	11.2	8.926
John Carlson	D	9.8	7.810
Evgeny Kuznetsov	C	8.8	7.013
Braden Holtby	G	8.3	6.614
Nicklas Backstrom	C	6.9	5.499
Matt Niskanen	D	6.6	5.260
Philipp Grubauer	G	6.5	5.180
Dmitry Orlov	D	6.2	4.941
T.J. Oshie	C	4.4	3.506
Lars Eller	C	3.4	2.710
Christian Djoos	D	3.3	2.630
Tom Wilson	RW	3.2	2.550
Brett Connolly	RW	2.6	2.072
Andre Burakovsky	LW	2.6	2.072
Alex Chiasson	RW	1.6	1.275
Chandler Stephenson	C	1.6	1.275
Jay Beagle	RW	1.4	1.116
Brooks Orpik	D	1.3	1.036
Michal Kempny	D	0.8	0.650
Devante Smith-Pelly	RW	0.5	0.650
Travis Boyd	C	0.1	0.650

[TBODY] [/TBODY]

 

[Carlzner]

Clicking on a thread for the first time since it was bumped just to see a bunch of nonsense charts from Francais, classic

 

[CapitalsCupReality]

This comparison vs contract value might not paint the picture some want to see lol....some big name/big $ guys underperforming year to year.

That would seem to be further validation as to the accuracy of some of the “eye tests” that are often casually dismissed around these parts by the math crew.

People that know this sport don’t always need a chart to make the determination that a player is underperforming, but I guess if math is the only thing you believe....read the charts lol.

 

[AlexBrovechkin8]

@twabby you probably know this -- where can I find how much time a player spends with other players as linemates? My old go-to was taken down and I'm not sure I've seen a quality replacement.

 

[francaisvolantsparis]

@twabby you probably know this -- where can I find how much time a player spends with other players as linemates? My old go-to was taken down and I'm not sure I've seen a quality replacement.

Line Stats - Natural Stat Trick

 

[twabby]

@twabby you probably know this -- where can I find how much time a player spends with other players as linemates? My old go-to was taken down and I'm not sure I've seen a quality replacement.

Player List - Natural Stat Trick

You can search for a name, a state (5v5, PP, PK, etc.), and there's a "Teammates" link.

 

[twabby]

Line Stats - Natural Stat Trick

This will also work, but you need to select the players you're interested in. This link is also useful if you want to see how multiple players do at the same time. For instance, you can see how Ovechkin-Backstrom-Wilson do compared to Ovechkin-Kuznetsov-Wilson, or similar.

 

[kicksavedave]

Heh, so Oshie is overpaid, Wilson too..that’s all I really need to see from that chart for validation.

not even worth calling out the other high priced eye sores listed....

Wait, I thought people were only paid for point production?

 

[francaisvolantsparis]

It is division points. Two points for the win. One point for the OT loss.

 

[CapitalsCupReality]

Wait, I thought people were only paid for point production?

“only”, that’s wrong, but you damn sure get paid when you DO produce.

 

[Random schmoe]

(Oof, my apologies in advance for breaking anyone's display. I don't know what I'm doing with uploading tables. It is what it is. I wrapped in spoiler tags in the hopes it helps.)

I got curious. Instead of comparing Kuznetsov's defensive prowess to Dowd's (or some other), or saying Dowd isn't very good because he doesn't score as much as Kuzy, etc. I wanted to see how the Caps players performed in super simplistic methods based on how/where they were deployed. i.e., did 'defensive' (Dowd, Chara, etc) players actually do OK limiting goals when starting in the defensive zone, did 'offensive' (Carlson, Kuzy, etc) players actually do OK scoring goals when starting in the offensive zone.

Charts below presented without conclusions - because honestly there are still a ton of variables that this doesn't include, and I don't know of any formula that can factor them in without being subjective. I basically just took Goals For divided by Offensive Zone Starts, and Goals Against divided by Defensive Zone Starts. Minimum 200 minutes. This is also showing all situations, because I never understood why you only want to measure ~80-90% of a game. (I do know the logic used to focus on 5v5, I just disagree with it.)

I do think some of the things this shows is interesting - like Nick Jensen has been on the ice for 21 GF, while only getting 26 offensive zone starts. By far the best ratio among D-men on the team, compared to Carlson's 48 GF despite 146 offensive zone starts.

Showing all NHL players also reveals something interesting, that of all qualifying forwards, Dowd has been on the ice for 16 GA despite 158 D Zone starts - that's the fifth best ratio in the entire league.

As evidenced by Jensen's numbers, there are obvious sample size questions on some of those. So be it, I'm not pretending its great or all meaning.

Filter is applied to show only Caps players, but anyone is welcome to download it (link in each title), play with the filters, see NHL as a whole, etc. I won't pretend to be much of an advanced stats guy - I think expected goals is a wholly stupid way of evaluating performance over a season - so yeah, whatever, nevermind.

Spoiler: Forwards

Forwards:

Spoiler: Defensemen

Defensemen:

 

[francaisvolantsparis]

(Oof, my apologies in advance for breaking anyone's display. I don't know what I'm doing with uploading tables. It is what it is. I wrapped in spoiler tags in the hopes it helps.)

I got curious. Instead of comparing Kuznetsov's defensive prowess to Dowd's (or some other), or saying Dowd isn't very good because he doesn't score as much as Kuzy, etc. I wanted to see how the Caps players performed in super simplistic methods based on how/where they were deployed. i.e., did 'defensive' (Dowd, Chara, etc) players actually do OK limiting goals when starting in the defensive zone, did 'offensive' (Carlson, Kuzy, etc) players actually do OK scoring goals when starting in the offensive zone.

Charts below presented without conclusions - because honestly there are still a ton of variables that this doesn't include, and I don't know of any formula that can factor them in without being subjective. I basically just took Goals For divided by Offensive Zone Starts, and Goals Against divided by Defensive Zone Starts. Minimum 200 minutes. This is also showing all situations, because I never understood why you only want to measure ~80-90% of a game. (I do know the logic used to focus on 5v5, I just disagree with it.)

I do think some of the things this shows is interesting - like Nick Jensen has been on the ice for 21 GF, while only getting 26 offensive zone starts. By far the best ratio among D-men on the team, compared to Carlson's 48 GF despite 146 offensive zone starts.

Showing all NHL players also reveals something interesting, that of all qualifying forwards, Dowd has been on the ice for 16 GA despite 158 D Zone starts - that's the fifth best ratio in the entire league.

As evidenced by Jensen's numbers, there are obvious sample size questions on some of those. So be it, I'm not pretending its great or all meaning.

Filter is applied to show only Caps players, but anyone is welcome to download it (link in each title), play with the filters, see NHL as a whole, etc. I won't pretend to be much of an advanced stats guy - I think expected goals is a wholly stupid way of evaluating performance over a season - so yeah, whatever, nevermind.

Spoiler: Forwards

Forwards:

Spoiler: Defensemen

Defensemen:

Many GF are scored from neutral zone starts or even from the def zone starts not only from off zone stats. The same is true for GA. You are dividing your stats by the zero here. If you count all goals from all starts but only the off starts to divide. Imagine someone who has only played def starts in all the season and scored one goal in 82 full games. One goal in a whole season is awful and this player will never see NHL contract again. But in your metric he will have 999999999999999999999999% efficasity.

 

[Random schmoe]

I know it excludes information that is relevant. But I specifically wanted to focus on offensive and defensive zone deployments. AFAIK, there is no stat to isolate goals per zone start. So this includes and shows exactly what i wanted it to.

 

[kicksavedave]

Interesting article about how the NHL is capturing and using data for analytics. I did not know they are now using chips in shoulder pads to track players. Obviously having more data (vs recording data using only the eyeballs) leads to more opportunity to use AI to crunch that data.

https://blogs.oracle.com/bigdata/how-hockey-is-embracing-big-data-and-analytics

Wonder what this all means for Kuzy? *grin*

 

[twabby]

Bumping this thread so a few forums regulars can give it a quick read.

 

[Hivemind]

A comment from the closed GDT caught my eye - Billy Beane is not out of baseball. He's still with Oakland. He's the Vice President of Baseball Operations.

 

[hb12xchamps]

A comment from the closed GDT caught my eye - Billy Beane is not out of baseball. He's still with Oakland. He's the Vice President of Baseball Operations.

He’s leaving baseball to take a role with Fenway Sports Group as of October 2020. I’d venture this is his last season. Article I read said he most likely shifts his focus to European soccer