All Purpose Analytics and Extended Stats Discussion

twabby

Registered User
Mar 9, 2010
13,623
14,439
And because his individual shooting rate was also discussed, here is the rolling 10 game average of his individual ixG/60:

upload_2021-4-3_12-24-20.png


(edit: I originally had all situations ixG/60, instead of 5v5 which is what I intended. I believe this is now fixed.)

He is shooting less as the season is going on, and moreso recently after an uptick in the middle of the season. I'm not sure if his lack of current shooting is because he's choosing to not shoot when he has quality opportunities, or if he's just not been able to create the opportunities to begin with.
 
Last edited:
  • Like
Reactions: crab

g00n

Retired Global Mod
Nov 22, 2007
30,430
14,290
And because his individual shooting rate was also discussed, here are the rolling 10 game average of his individual ixG/60:

View attachment 416782

He certainly is shooting less and less as the season is going on. I'm not sure if it's because he's choosing to not shoot when he has quality opportunities, or if he's just not been able to create the opportunities to begin with.

Totally anecdotal "eye test" suggests he's passing up shots, as was seen last night and noted in the GDT when he penetrated the high slot and curled off. He has to take shots like that. It's been a thing for much of his career. I wonder how his early season shooting compared to his career numbers, as it seemed like he was on fire early on when it came to scoring.
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
Totally anecdotal "eye test" suggests he's passing up shots, as was seen last night and noted in the GDT when he penetrated the high slot and curled off. He has to take shots like that. It's been a thing for much of his career. I wonder how his early season shooting compared to his career numbers, as it seemed like he was on fire early on when it came to scoring.

His career average is 0.53 ixG/60 at 5v5. He only really outperformed this number during the middle of the season this year. Even at the beginning he was underperforming his career average a bit, and of late has been significantly underperforming it.
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
What about other individuals? How big is the delta between their O and D plots? Is it particularly unusual to be only slightly better defensively than you underperform offensively? Or is it supposed to be 1:1?

It depends on the player. Players who are better than the team on average overperform in one area more than they underperform in the other area (or they overperform both offensively and defensively). Players who are worse than the team on average underperform in one area more than they overperform in the other area (or they underperform both offensively and defensively).

Examples:

Overperform offensively and defensively: Zdeno Chara, TJ Oshie, Lars Eller, Garnet Hathaway, Jakub Vrana
Overperform defensively more than they underperform offensively: Nick Jensen, Richard Panik, Carl Hagelin, Conor Sheary
Overperform offensively more than they underperform defensively: Evgeny Kuznetsov, Dmitry Orlov
Underperform defensively more than they overperform offensively: Alex Ovechkin, Tom Wilson, Justin Schultz
Underperform offensively more than they overperform defensively: Nic Dowd, Nicklas Backstrom
Underperform offensively and defensively: Brenden Dillon, John Carlson, Daniel Sprong

Also, I want to be clear that underperform and overperform are not synonyms for "good" and "bad" because these numbers do not have context like quality of competition and zone starts factored in, though they do take into account score-effects.

But it is instructive because if you do take into account on-ice context, it's still concerning for a player like Backstrom because he is getting easier deployments than he has in several years. Contrast that with Nic Dowd who is getting the absolute toughest deployments on the team and yet he's still able to be above water defensively. I'd say he's doing really good, despite his xGF% being lower than the team's on average.
 

g00n

Retired Global Mod
Nov 22, 2007
30,430
14,290
It depends on the player. Players who are better than the team on average overperform in one area more than they underperform in the other area (or they should overperform both offensively and defensively). Players who are worse than the team on average underperform in one area more than they overperform in the other area (or they should underperform both offensively and defensively).

Examples:

Overperform offensively and defensively: Zdeno Chara, TJ Oshie, Lars Eller, Garnet Hathaway, Jakub Vrana
Overperform defensively more than they underperform offensively: Nick Jensen, Richard Panik, Carl Hagelin, Conor Sheary
Overperform offensively more than they underperform defensively: Evgeny Kuznetsov, Dmitry Orlov
Underperform defensively more than they overperform offensively: Alex Ovechkin, Tom Wilson, Justin Schultz
Underperform offensively more than they overperform defensively: Nic Dowd, Nicklas Backstrom
Underperform offensively and defensively: Brenden Dillon, John Carlson, Daniel Sprong

Kuz and Orlov are not surprising to me. Vrana's position (and Chara's) will likely raise eyebrows in some. As will Carlson.

Top lines and PP guys seem to be in the most precarious position aside from Osh-babe.
 
  • Like
Reactions: ALLCAPSALLTHETIME

Leksand

Registered User
Oct 30, 2013
663
327
Northern VA
Kuz and Orlov are not surprising to me. Vrana's position (and Chara's) will likely raise eyebrows in some. As will Carlson.

Top lines and PP guys seem to be in the most precarious position aside from Osh-babe.
Well, Carlson is the opposite to an analytics darling, eg seen in major skepticism about his contract and resistance to being in Norris contention or even favorite last year. I have always seen weaknesses defensively and can see where the mostly "counting stats" narrative comes from, but boy is it great to get the puck in the net for a team - at any level it just makes it easier than to always have to worry about letting even one in. I realize there's an ES vs PP thing etc., but would any GM take Chara over Carlson for the rest of this season if given the choice?
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
Just watched the talk by Meghan Chayka at the Ottawa Hockey Analytics conference because it seems like it could be pertinent to the Capitals and their inability to hold leads.



One of the main takeaways is obvious: trailing teams tend not to get as many rush chances as leading teams. It makes sense intuitively of course. Teams that are protecting a lead are more conservative and tend to prioritize staying between the puck and their own net, while teams that are behind are more likely to take chances that may lead to odd-man breaks against.

Numerically, she shows the following:

upload_2021-4-5_13-59-9.png


upload_2021-4-5_14-0-3.png


I think the second graph in particular is useful for the Capitals. When the Capitals are leading they will enter a game-state where they will face cycle chances against but will be provided with more opportunities to score off the rush. So in this case it seems like the correct strategy would be to play the players who are either good at defending against the cycle, or who are good at scoring off the rush, or ideally both.
 
Last edited:

twabby

Registered User
Mar 9, 2010
13,623
14,439
As a follow-up to the previous post, who are players that seem good at defending the cycle via my eye-test? My initial thoughts are Chara for sure, Jensen, and to a lesser extent Orlov. Neither Schultz nor Carlson impress me in this regard, and Dillon seems somewhat neutral. The 4th line, Backstrom, Eller, and Oshie seem like they would be good at this from the forward POV. So what do the numbers say? The following table shows players' xGA/60 when the Capitals are leading.

PlayerxGA/60
Nick Jensen1.4
Zdeno Chara1.49
Carl Hagelin1.67
Conor Sheary1.73
Garnet Hathaway1.9
T.J. Oshie1.9
Nicklas Backstrom1.99
Lars Eller2.05
Nic Dowd2.06
Tom Wilson2.09
Jakub Vrana2.1
Dmitry Orlov2.11
Richard Panik2.13
Evgeny Kuznetsov2.18
Justin Schultz2.19
John Carlson2.4
Alex Ovechkin2.43
Daniel Sprong2.43
Brenden Dillon2.47
[TBODY] [/TBODY]
This makes sense to me. At the top are many I perceive to be good defending in their own zone against the cycle, while at the bottom I see many weaker-appearing in-zone players.

And eye-test wise who are players who seem to create rush chances? Kuznetsov, Vrana, and Ovechkin are all players I see as good at generating rush chances. Backstrom, Wilson, Orlov, and Carlson to a lesser extent. And how do these players rate in terms of xGF/60 while ahead?

PlayerxGF/60
Evgeny Kuznetsov2.09
Brenden Dillon2
Alex Ovechkin1.95
Carl Hagelin1.86
Garnet Hathaway1.84
Lars Eller1.84
John Carlson1.84
Justin Schultz1.71
Nic Dowd1.66
Dmitry Orlov1.66
Tom Wilson1.63
T.J. Oshie1.57
Jakub Vrana1.55
Daniel Sprong1.44
Conor Sheary1.41
Richard Panik1.37
Nick Jensen1.33
Nicklas Backstrom1.27
Zdeno Chara1.23
[TBODY] [/TBODY]
There are a few more surprises here than I thought. Dillon near the top is surprising, though I'm not sure that it's because of him instead of who's on the ice with him. Backstrom near the bottom is somewhat surprising, though his lack of speed makes it less surprising I guess. Vrana near the bottom is also surprising. Kuznetsov at the top is not surprising at all.

So what to make of it all? Well, once again I think the 4th line needs to be praised. They're being asked to do a lot and for the most part are doing it. Maybe Jensen and Chara should be called on for shutdown duty late in games instead of Carlson and Dillon. Carlson in particular has struggled holding the lead. Kuznetsov is their best player at creating chances and I think it's also fair to say he's one of their best players in terms of actually converting rush chances. If the Capitals want to keep the opposition honest he needs to be out there getting regular shifts when ahead, otherwise the opponent may cheat even more than usual in the offensive zone. Backstrom from a pure-shutdown perspective is fine but is not doing anything to blow the game open.

One thing I would like is to see how the team is doing specifically in the third period, though I'm having trouble finding that data. It would probably be more illuminating since that is where the team seems to break down the most.
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
Are the Caps' shooting percentages unusually high and possibly be due for a regression?

Any analytics which would help with that?

I'd say they are likely due for a regression because they are blowing their expectation out of the water at 5v5: they are scoring 3.1 goals/60, but only averaging 2.1 expected goals for/60, beating their expectation by a full 1.0 goals/60. The next best team since 2007-08 in terms of outperforming expectation was the 09-10 Capitals at +0.74 goals above expected/60.

But I'd add that the Capitals have always had unusually high shooting percentages in the regular season. They simply have always seemed to have way above average shooters and playmakers, at least in the regular season. Every year since 2011-12 they have exceeded their expected goals at 5v5.

However, the concerning thing is that while the Capitals have always shot way above expectation in the regular season, that same success has not translated to the postseason with only a few notable exceptions: 2009 (when Ovechkin lit it up), 2010 (skewed mainly by the first 4 games in the series against Montreal), and 2018 (a cool thing happened this year). Every other postseason they have shot near expectation or below at 5v5.

This leads me to believe that whatever the Capitals are doing in the regular season that "breaks" the expected goals model is not happening in the postseason except in 2018 and a few other years. My suspicion is they are more easily able to generate rush chances in the regular season or able to make the extra pass in the offensive zone, something that the xG model doesn't handle particularly well. This is something they were finally able to do in the 2018 postseason, where it seemed like a ton of their goals were scored off the rush compared to postseasons prior.

And for all of the years that they have run into a "hot" goalie, it seems like they weren't really able to get a lot of odd-man rushes, especially against Lundqvist when he was a Ranger. Though this is completely anecdotal and don't have the data to back that up.

Perhaps this suggests that their formula for winning this postseason isn't to grind it out and play heavy, cycling low-to-high hockey, but rather to win the transition battle. Prevent odd-man rushes against, and develop an effective counter-attacking strategy to create odd-man rushes for.
 
Last edited:

HTFN

Registered User
Feb 8, 2009
12,173
10,703
No genius, but Trotz was a "quality over quantity" shots guy who was fine with like 20 shots if 10 were high danger, which explained the shooting percentage, and then when the laser focus comes in and the playoff defense ramps up those chances weren't always there in a way that you can't really predict or adjust for. The opponent suddenly trying harder with more consistency isn't really a systems failure, it's just the dumb luck you have to push in the playoffs while you're hoping for bounces.

I wonder if this is why they've maintained that percentage well into Laviolette's emphasis on shot volume, but then again I also see them frequently out shot so I don't know that Laviolette is seeing those 50+ attempts that he wants either. Their ability to convert is pretty clearly learned and sustainable at some level so looking for playoff success is really just going to come down to Laviolette in the playoffs. I don't think there's a measurable answer there.
 

ALLCAPSALLTHETIME

Great Dane! Love that Eller feller.
Oct 10, 2009
9,234
4,898
British Columbia, Canada
I'd say they are likely due for a regression because they are blowing their expectation out of the water at 5v5: they are scoring 3.1 goals/60, but only averaging 2.1 expected goals for/60, beating their expectation by a full 1.0 goals/60. The next best team since 2007-08 in terms of outperforming expectation was the 09-10 Capitals at +0.74 goals above expected/60.

But I'd add that the Capitals have always had unusually high shooting percentages in the regular season. They simply have always seemed to have way above average shooters and playmakers, at least in the regular season. Every year since 2011-12 they have exceeded their expected goals at 5v5.

However, the concerning thing is that while the Capitals have always shot way above expectation in the regular season, that same success has not translated to the postseason with only a few notable exceptions: 2009 (when Ovechkin lit it up), 2010 (skewed mainly by the first 4 games in the series against Montreal), and 2018 (a cool thing happened this year). Every other postseason they have shot near expectation or below at 5v5.

This leads me to believe that whatever the Capitals are doing in the regular season that "breaks" the expected goals model is not happening in the postseason except in 2018 and a few other years. My suspicion is they are more easily able to generate rush chances in the regular season or able to make the extra pass in the offensive zone, something that the xG model doesn't handle particularly well. This is something they were finally able to do in the 2018 postseason, where it seemed like a ton of their goals were scored off the rush compared to postseasons prior.

And for all of the years that they have run into a "hot" goalie, it seems like they weren't really able to get a lot of odd-man rushes, especially against Lundqvist when he was a Ranger. Though this is completely anecdotal and don't have the data to back that up.

Perhaps this suggests that their formula for winning this postseason isn't to grind it out and play heavy, cycling low-to-high hockey, but rather to win the transition battle. Prevent odd-man rushes against, and develop an effective counter-attacking strategy to create odd-man rushes for.
Thank for the detailed answer. It was sort of what I figured.

Sounds like what the Pens did to win their recent Cups. Control odd man breaks and win the transition battle. That plus a lot of luck.
 
  • Like
Reactions: twabby

g00n

Retired Global Mod
Nov 22, 2007
30,430
14,290
As a follow-up to the previous post, who are players that seem good at defending the cycle via my eye-test? My initial thoughts are Chara for sure, Jensen, and to a lesser extent Orlov. Neither Schultz nor Carlson impress me in this regard, and Dillon seems somewhat neutral. The 4th line, Backstrom, Eller, and Oshie seem like they would be good at this from the forward POV. So what do the numbers say? The following table shows players' xGA/60 when the Capitals are leading.

PlayerxGA/60
Nick Jensen1.4
Zdeno Chara1.49
Carl Hagelin1.67
Conor Sheary1.73
Garnet Hathaway1.9
T.J. Oshie1.9
Nicklas Backstrom1.99
Lars Eller2.05
Nic Dowd2.06
Tom Wilson2.09
Jakub Vrana2.1
Dmitry Orlov2.11
Richard Panik2.13
Evgeny Kuznetsov2.18
Justin Schultz2.19
John Carlson2.4
Alex Ovechkin2.43
Daniel Sprong2.43
Brenden Dillon2.47
[TBODY] [/TBODY]
This makes sense to me. At the top are many I perceive to be good defending in their own zone against the cycle, while at the bottom I see many weaker-appearing in-zone players.

And eye-test wise who are players who seem to create rush chances? Kuznetsov, Vrana, and Ovechkin are all players I see as good at generating rush chances. Backstrom, Wilson, Orlov, and Carlson to a lesser extent. And how do these players rate in terms of xGF/60 while ahead?

PlayerxGF/60
Evgeny Kuznetsov2.09
Brenden Dillon2
Alex Ovechkin1.95
Carl Hagelin1.86
Garnet Hathaway1.84
Lars Eller1.84
John Carlson1.84
Justin Schultz1.71
Nic Dowd1.66
Dmitry Orlov1.66
Tom Wilson1.63
T.J. Oshie1.57
Jakub Vrana1.55
Daniel Sprong1.44
Conor Sheary1.41
Richard Panik1.37
Nick Jensen1.33
Nicklas Backstrom1.27
Zdeno Chara1.23
[TBODY] [/TBODY]
There are a few more surprises here than I thought. Dillon near the top is surprising, though I'm not sure that it's because of him instead of who's on the ice with him. Backstrom near the bottom is somewhat surprising, though his lack of speed makes it less surprising I guess. Vrana near the bottom is also surprising. Kuznetsov at the top is not surprising at all.

So what to make of it all? Well, once again I think the 4th line needs to be praised. They're being asked to do a lot and for the most part are doing it. Maybe Jensen and Chara should be called on for shutdown duty late in games instead of Carlson and Dillon. Carlson in particular has struggled holding the lead. Kuznetsov is their best player at creating chances and I think it's also fair to say he's one of their best players in terms of actually converting rush chances. If the Capitals want to keep the opposition honest he needs to be out there getting regular shifts when ahead, otherwise the opponent may cheat even more than usual in the offensive zone. Backstrom from a pure-shutdown perspective is fine but is not doing anything to blow the game open.

One thing I would like is to see how the team is doing specifically in the third period, though I'm having trouble finding that data. It would probably be more illuminating since that is where the team seems to break down the most.

Have not read further to see if this was covered, but it would be interesting to see what the difference is in GA/GF for these tables per player. That might make the decision easier.

Regarding Kuzy, I get what you're saying since he's at the top of the GF list but he's also 6th worst in the GA table. So that's why I'd like to see the delta, if you think it's worth the time.

Overall it does seem like Lavi probably favors the GA breakdown when making his late-game decisions. This seems like a PHILOSOPHICAL difference rather than a statistical one. He and his coaches seem to be aware of who's more likely to shut down the other team late in the game but they don't prioritize additional scoring opportunities.

I also wonder if, as discussed recently in another thread, they would benefit from style changes in the 3rd to protect the leads they build in their afiak league-best 2nd periods. Simply going by the personnel doesn't seem to be working as well as hoped.
 
  • Like
Reactions: ALLCAPSALLTHETIME

g00n

Retired Global Mod
Nov 22, 2007
30,430
14,290
My suspicion is they are more easily able to generate rush chances in the regular season or able to make the extra pass in the offensive zone, something that the xG model doesn't handle particularly well. This is something they were finally able to do in the 2018 postseason, where it seemed like a ton of their goals were scored off the rush compared to postseasons prior.

I think this is exactly what happens (bold). It's typical of teams with a lot of skill up front to get "too fancy" and pull of shit that won't fly in the postseason, especially with teams far more willing to block shots in the playoffs (presumably), or more likely to cheat/float defensively in general in the regular season.

They also got much luckier in 2018 than any playoff run I've ever seen in Caps history. So that helps.
 

kicksavedave

I'm just here for the memes and gifs.
Sponsor
Apr 29, 2009
10,599
13,239
Fallbrook, CA
www.tiasarms.org
I think this is exactly what happens (bold). It's typical of teams with a lot of skill up front to get "too fancy" and pull of shit that won't fly in the postseason, especially with teams far more willing to block shots in the playoffs (presumably), or more likely to cheat/float defensively in general in the regular season.

They also got much luckier in 2018 than any playoff run I've ever seen in Caps history. So that helps.

Conversely, some of their epic playoff failures had a pretty large sum of bad luck against. In 2018 it all went our way, but a single bounce here or there and it could have just been more of the same, just like a bounce here or there in the past and we could have already had a Cup before 2018.
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
Have not read further to see if this was covered, but it would be interesting to see what the difference is in GA/GF for these tables per player. That might make the decision easier.

Regarding Kuzy, I get what you're saying since he's at the top of the GF list but he's also 6th worst in the GA table. So that's why I'd like to see the delta, if you think it's worth the time.

Overall it does seem like Lavi probably favors the GA breakdown when making his late-game decisions. This seems like a PHILOSOPHICAL difference rather than a statistical one. He and his coaches seem to be aware of who's more likely to shut down the other team late in the game but they don't prioritize additional scoring opportunities.

I also wonder if, as discussed recently in another thread, they would benefit from style changes in the 3rd to protect the leads they build in their afiak league-best 2nd periods. Simply going by the personnel doesn't seem to be working as well as hoped.

I’m on my phone now so I can’t make a table on my own, but the data is here for Capitals players when ahead by any amount:

Player Season Totals - Natural Stat Trick

And then ahead by 1 goal exactly:

Player Season Totals - Natural Stat Trick

In situations where they are leading, Kuznetsov is the best on the team in terms of actual goal differential and in terms of actual goals allowed, and is also very good with a 1 goal lead.

So hopefully Laviolette isn’t making the decision to reduce Kuznetsov’s ice time based on results, because his results are very good.

I agree, I think some strategic and stylistic changes probably need to be made. I think this may include trying to attack and counter-attack more than they do now. They also need to get more saves in general.
 

Random schmoe

Random fan with their own opinions
Sponsor
Feb 13, 2019
911
938
PlayerShots2021%Career%
Nicklas Backstrom6221.012.2
Alex Ovechkin13414.212.8
John Carlson9510.56.3
T.J. Oshie7215.314.4
Jakub Vrana6515.413.0
Tom Wilson4520.011.0
Evgeny Kuznetsov5512.711.8
Justin Schultz504.06.2
Conor Sheary4219.012.5
Lars Eller3713.510.1
Garnet Hathaway419.89.7
Daniel Sprong 34 20.6 11.8
Nick Jensen375.42.2
Brenden Dillon323.13.3
Dmitry Orlov4810.46.0
Carl Hagelin557.37.9
Nic Dowd4316.311.3
Richard Panik407.511.5
Zdeno Chara454.46.2
Trevor van Riemsdyk 8 12.5 3.5
Jonas Siegenthaler30.03.1
[TBODY] [/TBODY]
 
  • Like
Reactions: ALLCAPSALLTHETIME

Random schmoe

Random fan with their own opinions
Sponsor
Feb 13, 2019
911
938
(wow, tables suck. I can't seem to add text to the message outside of the table structure....)

Someone smarter than me with more time can probably figure out net weights of shooting percentage differentials vs career numbers based on shot counts..
 
  • Like
Reactions: ALLCAPSALLTHETIME

twabby

Registered User
Mar 9, 2010
13,623
14,439
Overall it does seem like Lavi probably favors the GA breakdown when making his late-game decisions. This seems like a PHILOSOPHICAL difference rather than a statistical one. He and his coaches seem to be aware of who's more likely to shut down the other team late in the game but they don't prioritize additional scoring opportunities.

Getting back to this from yesterday: I actually give Laviolette more credit than that. Even if it is just a philosophical difference, it is also the correct statistical choice in many cases to prioritize the defensive option when ahead, even if another option might have a better overall expected goal differential. And it makes sense right? It's an extreme example, but if the Capitals are up by 2 goals with say 5 minutes left, would you prefer to play the player who averages 10 goals for and 7 goals against per 60, or would you rather play the player who averages 0.5 goals for and 1.0 goals against per 60? Even though the first player is going to average out in the long run to have a much better goal differential, in this case it's probably better to play the second player because not allowing a goal is so much more valuable at that point than scoring another goal.

However, in practice the margins between the abilities of players is much tighter than the example above, and also it seems like Laviolette has in the past benched players like Kuznetsov and Vrana for almost the entire third period when defending a lead, instead of only reducing their ice time toward the very end of the period where defensive leverage becomes much more important.

Going to dig into the numbers more in a bit.
 

twabby

Registered User
Mar 9, 2010
13,623
14,439
Ok, so long post ahead.

Goal-scoring and allowing goals in hockey is generally modeled to be a Poisson Process. Here's a primer on the Poisson Distribution and how it can model goal-scoring in the NHL: http://www.hockeyanalytics.com/Research_files/Poisson_Toolbox.pdf

It's not a perfect model, but it's probably close enough for practical purposes.

So assuming goal-scoring and conceding each follow a Poisson distribution, and that they are independent of each other (this is not true, but again it's likely close enough for our purposes), we want to know what the likely goal differential will be when a certain player plays.

To model this, we use the Skellam Distribution, which is the difference in two Poisson random variables. Skellam distribution - Wikipedia

Using players' current expected goals for and against/60 while leading, combined with the Skellam Distribution, we can calculate the probabilities of scoring or conceding in a 1 minute period of time on the ice for each of the 4 centers on the team. I chose 1 minute mainly because the math was easiest and also because it's "close enough" to the average shift length of an NHLer, which is about 45-48 seconds. Here are the results:

PlayerxGF/60xGA/60P(-1 goals)P(0 goals)P(+1 goals)
Nicklas Backstrom1.271.993.14%94.78%2.01%
Evgeny Kuznetsov2.092.183.39%93.25%3.25%
Lars Eller1.842.053.20%93.82%2.88%
Nic Dowd1.662.063.23%94.08%2.60%
[TBODY] [/TBODY]
(Note that I ignored the probability of scoring or conceding 2+ goals because the probabilities are negligible. I doubt including them will have much of an impact on what follows.)

The probabilities in the table seem to make sense. On a vast majority of 1 minute increments no goals are scored. Backstrom is the lowest event center on the team in terms of xG, so he's more likely to have a 1 minute period of scoreless hockey than any of the other players. And on the other end, Kuznetsov is the highest event center in terms of xG, so he is least likely to have a 1 minute period of scoreless hockey. But since Kuzy has the best xGF/60 it makes sense that he has the best chance of having a 1 minute period where the Capitals net +1 goals. And since Backstrom has the worst xGF/60 while leading, it also makes sense that he is the least likely to have a 1 minute period of play where the Capitals have a net +1 goals. Similar results for xGA, where Kuznetsov fares worst and Backstrom fares best since they are the worst and best, respectively, in terms of xGA/60.

Given these probabilities above, it's also important to consider what effect scoring a goal, conceding a goal, or doing neither have on win probability given the current game state. I did a little Googling and found the following paper on modeling win probability based on time remaining in the game and score-state (ahead by 1, behind by 2, tied, etc.): https://dlib.bc.edu/islandora/object/bc-ir:108029/datastream/PDF/view

Based on Christophe Bernier's model of win probability, and based on the the probabilities of scoring or conceding calculated for the top 4 centers on the team in the above table, here are the results for each of the players above in terms of expected win probability added for a 1 minute period of play in the third period, assuming the Capitals are ahead by 1:

upload_2021-4-6_13-51-17.png


So until about 6 or 7 minutes into the 3rd period it looks like Kuznetsov would be the best option to play under this model, which makes some level of sense since the game still isn't close to ending. Preferring the player with the best xG differential seems sensical enough when the end isn't near since getting a 2 goal lead still has a lot of value at this point in the game.

However, as the game gets closer to ending Kuznetsov becomes the least preferable player, and Backstrom, who has the best xGA despite having the worst xGF, becomes the best option with about 2 or 3 minutes left in the game. Again, this makes a lot of sense because conceding a goal will hurt win probability much more than scoring a goal will help win probability toward the end of a game when leading by 1.

But despite this, Kuznetsov still appears to be one of the top 2 center options until about 7 or 8 minutes left in the third period, and is still a top 3 option until about 6 minutes left.

Here are the results when up 2 and up 3:

upload_2021-4-6_13-51-6.png

upload_2021-4-6_16-39-45.png


These results are much less interesting IMO. All 4 centers seem to have a similar impact on win probability no matter how much time is left with a multigoal lead (note the scales are slightly different between these graphs and the Up 1 graph above).

So what do I make of this all in terms of Kuznetsov and third period deployment? If holding a multigoal lead it's probably just best to play all of the centers equally and that there's no need to bench anyone. This will leave everyone the freshest, and probably more able to defend a lead than if a few of them are double-shifting which will likely impact their performance due to exhaustion. If holding a 1-goal lead, then I would play Kuznetsov normally until about 6 minutes are remaining in the game, at which point he might become a riskier option to play and I could see taking shifts away from him being the sensical thing to do.

But based on the model above I would not be restricting his shifts as early as Laviolette has been doing, and especially since he seems to do this with multigoal leads where it doesn't seem sensical at all. His slightly worse defensive impact IMO doesn't make up for how it will hurt the other centers by having them double shift.

I plan on looking at defensemen next.
 
Last edited:

twabby

Registered User
Mar 9, 2010
13,623
14,439
As promised here are the defensemen:

upload_2021-4-6_15-42-17.png

upload_2021-4-6_15-42-24.png

upload_2021-4-6_15-42-31.png


Maybe this is where Laviolette should be focusing instead. In pretty much every game-state in the third period the Chara-Jensen pairing is the preferred one by the model, following by Orlov-Schultz, followed by Dillon-Carlson. In particular, when up by 1 or 2 goals this model likes Chara and Jensen quite a bit more than the other two pairs. But especially early in the third period with a 2 goal lead and late in the third period with a 1 goal lead are where they see their highest impacts relative to the other pairings.

Yet it seems like Carlson is the preferred RD in late game situations over Jensen and Schultz despite having a higher xGA/60 and GA/60 while leading than the other options, and Dillon/Chara are given relatively equally footing late in games despite Chara having the best xGA/60 and GA/60 while leading and Dillon having the worst (though I have been noticing Chara more and more in third periods when ahead, which is good).

Ultimately I think @g00n is right that it's more than just deployment and the team needs to adopt different tactics when holding a lead. But while it's arguable that he's making the correct choice in deployments late in the game, it's not really arguable that he is playing it "safe" late in games by benching Kuznetsov/Vrana and playing his #1D in high leverage defensive situations. These decisions don't seem particularly controversial inside the room and with his GM, I imagine. And given that he's playing it "safe" with the deployments, it indicates to me that Laviolette may be playing it too safe with his tactics as well. Perhaps preferring more dump-and-change instead of trying to spring odd-man rushes against the other teams cheating.
 
Last edited:

twabby

Registered User
Mar 9, 2010
13,623
14,439
From a fancy stats standpoint in a vacuum it seems to be a wash, or close to it. Some models prefer Mantha more, some Vrana. But they're close. What is pretty universal between the models is that Mantha provides quite a bit more defensively than Vrana.

This will make Washington better this year, if for no other reason than I have to imagine Laviolette is going to trust Mantha more than Vrana. For argument's sake if you call the two players a wash talent-wise, then Mantha likely getting somewhere in the neighborhood of 16-18 minutes a night compared to Vrana likely only getting 12-14 minutes a night seems like a lot of additional value added for Washington. Is that value worth an additional first round pick? I don't know, probably? Could Vrana have been utilized better? I don't know, probably?

This seems like one of those scenarios where it probably just comes down to fit, and if Washington thinks Mantha is a better fit then I understand the trade.
 
  • Like
Reactions: AlexBrovechkin8

Ad

Upcoming events

Ad

Ad

-->