All Purpose Analytics and Extended Stats Discussion

[g00n]

MOD NOTE: We recognize that not everyone is into deep statistical analysis. This is a thread for "extended stats" discussions. Stats can still be used in the general roster discussion to make a single point, but if someone wants to debate a player's statistical value ad nauseum, or engage in back and forth about the value of X or Y stat, it must go here. If you have extended analysis that applies in another thread, rather than jam up that thread with large charts and so forth you may want to link to a post here, or use a collapse tag in your original post. Contact a mod if you do not know how to do this.

 

[twabby]

Here is more evidence that Nate Schmidt should be considered for a tougher role soon. Red = more shots than league average, Blue = fewer shots than league average.

Shots for with Nate Schmidt on the ice:

Shots for without Nate Schmidt on the ice:

Shots against with Nate Schmidt on the ice (this is the important chart):

Shots against without Nate Schmidt on the ice:

So basically he gives up very few high danger shots against, and compared to his team helps create more high danger shots for. Isn't this what a defenseman ideally should do? Of course this is impacted by his softer deployments but if someone excels this much in soft deployments doesn't it make sense to test them in tougher deployments? Orlov had the soft deployments last year and was tested this year in tougher deployments and so far is passing the test. Why not do the same with Schmidt?

This is part of the reason why it irritates me that Schmidt is the one being sat. He's been fantastic on the third pairing and he's more of a 4/5 defenseman than a 6/7 IMO.

 

[g00n]

Probably because of who's above him right now. He'll get his chance, imo. Maybe not much this season, but eventually.

 

[twabby]

Probably because of who's above him right now. He'll get his chance, imo. Maybe not much this season, but eventually.

I get it, but maybe instead of being promoted to the second pairing I'd give 44-88 slightly tougher assignments and slightly more ice time and that way a more honest analysis can be conducted. They're killing the easy assignments.

 

[g00n]

I get it, but maybe instead of being promoted to the second pairing I'd give 44-88 slightly tougher assignments and slightly more ice time and that way a more honest analysis can be conducted. They're killing the easy assignments.

If Orpik stays healthy I think Trotz will use him a ton in the playoffs. If that pairing stays together Schmidt will see more time, and more critical time.

In the Leafs event log you posted I saw an awful lot of 44-4 for some key plays.

 

[36kap36]

Here is more evidence that Nate Schmidt should be considered for a tougher role soon. Red = more shots than league average, Blue = fewer shots than league average.

Shots for with Nate Schmidt on the ice:

Shots for without Nate Schmidt on the ice:

Shots against with Nate Schmidt on the ice (this is the important chart):

Shots against without Nate Schmidt on the ice:

So basically he gives up very few high danger shots against, and compared to his team helps create more high danger shots for. Isn't this what a defenseman ideally should do? Of course this is impacted by his softer deployments but if someone excels this much in soft deployments doesn't it make sense to test them in tougher deployments? Orlov had the soft deployments last year and was tested this year in tougher deployments and so far is passing the test. Why not do the same with Schmidt?

This is part of the reason why it irritates me that Schmidt is the one being sat. He's been fantastic on the third pairing and he's more of a 4/5 defenseman than a 6/7 IMO.

I should include the fact that I agree with you, and the luck (/PDO, or any stat you look at for them) of Chorney/Orpik is remarkable. I know I was arguing with you, but that was more or less for the sake of argument. At this point, there is no reason to not give him a bigger chance, and I like the idea of keeping pairings as-are. The problem is the HC, who couldn't give a crap less about stuff like this, as valuable as it is/can be.

What we have in the meantime is a pairing who should be able to turn games for us. If Schmidt continues to play at this level, especially in a sheltered role with Orpik, he should be able to come in and at bare minimum push the play into the offensive zone. If he continues swaying shot attempts like this, even in a limited role, he can hopefully be a game-changer for us. Unfortunately, it doesn't exactly play out that easily.

 

[Hivemind]

I miss Chibi.

 

[twabby]

Driven from the discussion in the Kuznetsov thread, here is Kuznetsov's spider-chart for this year. It basically shows his score-adjusted shot-attempts taken and allowed for various combinations of teammates that he has been on the ice with:

Link for larger version, since HFBoards seems to limit image sizes in posts

The diagonal red line is basically break-even. The further up and to the right, the better the combination has been in terms of shot-attempt ratio. Looking at the Kuznetsov + Orlov/Niskanen pairing is pretty striking: they obliterate the competition in terms of shot-attempts. Remember that these are score-adjusted, so garbage time is already factored in.

http://hockeyviz.com is a great resource for data visualization. I'd recommend clicking around and just exploring because it's much easier to interpret data when they are presented like this than tables full of numbers.

 

[g00n]

HF is resizing that pic so it's too small to see the numbers. Is it possible to upload the image to a hosting site and link that?

 

[twabby]

I put a link beneath the image for a larger version.

 

[CapitalsCupReality]

So, can the stats gurus prove or dismiss Randy's suggestion that some significant system change occurred from one game to the next, that suddenly put 92 in a position to succeed instead of failing?

Is there a game to game analysis that would highlight this shift in system that's being suggested happened to suddenly "make it all work again"? I feel dirty...

 

[twabby]

So, can the stats gurus prove or dismiss Randy's suggestion that some significant system change occurred from one game to the next, that suddenly put 92 in a position to succeed instead of failing?

Is there a game to game analysis that would highlight this shift in system that's being suggested happened to suddenly "make it all work again"? I feel dirty...

Shot-location data would probably be best for this. The heat maps I posted earlier apply for the entire season, unfortunately. You can't break them down by time period (or game by game).

Honestly I think the shift came in early December when the defensive pairings were changed, as I mentioned above. His deployment with a talented transition pairing in Orlov-Niskanen seemed to help spark his game IMO. Further to his spider plot posted above, his rolling expected goals for % is below:

Expected goals basically tries to take everything into account: shot location (distance and angle), shot type (snap shot, slap shot, wrist shot, etc.), whether is an off-side shot, and other factors.

I don't think it's coincidence that the valley in this chart corresponds to when the defensive pairings were changed. Before he hit rock bottom he had spent most of his time skating with Orlov-Carlson, afterwards most of his EV time was spent with Orlov-Niskanen and his xGF% rose dramatically. The former has had mixed results as a pairing, the latter has had incredibly great results.

I'm not sure if a system change also occurred on December 7 or just a change of the defensive combinations, but it seems clear that that is when things turned around for Kuznetsov (and others, for that matter. Ovechkin's xGF% has risen dramatically since December 7 as well).

 

[CapitalsCupReality]

Is Kuzy a Byproduct of playing with more compatible players, or byproduct of a seemingly magical system change....hmmm. Is Orlov Niskanen's success a byproduct of Kuzy's line playing better, or a cause of why they are? Are they all just playing better as players do during a nice streak of games often?

There's no doubt 70 is playing at an elite level. How does that all factor in when talking about team successes and why guys have upticks and downswings in production. Success breeds confidence throughout the lineup.

Too bad you can't see game by game #s. thanks for looking.

 

[RandyHolt]

Good stuff Twabby. Defensemen have been more active offensively to my eye. Almost a bit of the Finnish Five thing going on.

It may just be Kuzy unleashed from the back boards, however. He looks like a different player, almost overnight.

We still have the low to high offense only without going down below the goal line nearly as much. If we dont have a good look off the rush, the pucks are moving back up top much quicker, while the opposition Fs are still getting back / scrambling.

Good afternoon, Mr. Twabby. Your mission Twab, should you choose to accept it, I am curious how Alzner's advanced stats have been over the current hot streak we have been on. Has he been a Schultz like passenger racking up the pluses, or are his stats still lacking.

As always, should you or any of your H.F. Force be warned or banned, the Mod will disavow any knowledge of your actions. This post will self-destruct in five seconds. Good luck.

 

[twabby]

Good stuff Twabby. Defensemen have been more active offensively to my eye. Almost a bit of the Finnish Five thing going on.

It may just be Kuzy unleashed from the back boards, however. He looks like a different player, almost overnight.

We still have the low to high offense only without going down below the goal line nearly as much. If we dont have a good look off the rush, the pucks are moving back up top much quicker, while the opposition Fs are still getting back / scrambling.

Good afternoon, Mr. Twabby. Your mission Twab, should you choose to accept it, I am curious how Alzner's advanced stats have been over the current hot streak we have been on. Has he been a Schultz like passenger racking up the pluses, or are his stats still lacking.

As always, should you or any of your H.F. Force be warned or banned, the Mod will disavow any knowledge of your actions. This post will self-destruct in five seconds. Good luck.

Alzner's still concerning IMO. I checked from December 7 until now (including last night's game), and here are some of his and the rest of the defense's individual stats (all possession-stats are score, zone, and venue adjusted from corsica.hockey):

Player|Shot-attempt %|Expected goals %|Actual goals for %
Matt Niskanen|60.82%|60.03%|70.83%
Dmitry Orlov|59.63%|56.39%|68.00%
Nate Schmidt|58.12%|61.93%|88.89%
Brooks Orpik|54.21%|53.03%|64.71%
John Carlson|47.77%|49.66%|60.87%
Karl Alzner| 45.07% | 46.97% |66.67%

Every player has a great actual goals for % (thank you Braden Holtby, a more dynamic offense, and some honestly lucky shooting that will eventually go away). It's not surprising to see actual goals be above expected goals for each player since Holtby is great and the team probably has above average shooting talent, but the difference is too large to be sustainable.

Niskanen, Orlov, and Schmidt are killing it in terms of shot-attempts and expected goals %. This is a good indicator of future success. They are certainly getting lucky but even with break-even luck their goals-for % is likely to be on the positive side.

Carlson and Alzner are still concerning in that they get hemmed in their own zone way too often for more liking. Their expected goals for % paints a better picture of that pairing as they are closer to 50% in that regard, but still underwater. Certainly having tougher deployments hurts their cause, but it doesn't explain away the vast difference between them and the other 4 defensemen.

So to answer your question: Alzner really hasn't been much better over this stretch and it's an ongoing concern for me (being paired with Carlson certainly doesn't help things either). He's no Jeff Schultz because Schultz was a tire fire in his own zone and Alzner is good in his own zone and on the PK, but IMO Alzner lacks in the neutral and offensive zones just like Schultz and it hasn't really been getting better.

The other two pairings are too good right now to break up so I guess the only thing to hope for is improvement in their play. I have some faith in Carlson's ability to do this because he has shown in the past flashes of great play, but I'm not sold on Alzner yet who is the definition of a non-streaky player so I wouldn't suddenly expect anything different than what he has done his entire career. Like Hivemind mentioned in another thread, it might just be that this is the best configuration the team can have and that we will have to live with one of the pairings being underwater. I certainly see no need to break up the pairings as constructed.

 

[RandyHolt]

@Twabby Mission Impossible Accomplished - thanks much! I used to think exposing Orlov to expansion was our only choice. Now I am not so sure. Keep up the reporting on anything of interest.

It's interesting that the D pairs got shuffled at the same time I noticed what I call the funnel offense resurrected after a 12 month hiatus. Trotz clearly made significant changes.

 

[CapitalsCupReality]

What do the numbers say about 1st star Alzner tonight?

 

[RandyHolt]

Do games played count as stats? If so, maybe that's what they say.

I am interested in hearing about Willy's advanced stats before and after the turnaround ~12/1. I know its a small sample size but I expect to see a pop.

More painfully, I am also interested in seeing if those same stats mirror last year; Game 1 - Blizzard, and Blizzard - year end.

Maybe the big picture would be more telling; those same time periods, but for the entire team. My theory of course, sans the hitting, this is the same offense from early last season. I am curious if the advanced stats just happen to mirror the different approaches to winning Barry deploys.

 

[CapitalsCupReality]

his shots are down since 12/1. .86 shots per game over 21 games vs 1.05 shot per game the first 21 games.

 

[twabby]

Do games played count as stats? If so, maybe that's what they say.

I am interested in hearing about Willy's advanced stats before and after the turnaround ~12/1. I know its a small sample size but I expect to see a pop.

More painfully, I am also interested in seeing if those same stats mirror last year; Game 1 - Blizzard, and Blizzard - year end.

Maybe the big picture would be more telling; those same time periods, but for the entire team. My theory of course, sans the hitting, this is the same offense from early last season. I am curious if the advanced stats just happen to mirror the different approaches to winning Barry deploys.

In terms of shot-differential, the team has been incredibly consistent this season vs. last season:

]

I think this paints a good picture of how good the defense has been by limiting shot quantity consistently this season vs. the more erratic quality of the team defense last season. Of course I don't think it paints a great picture of the team offense because the team hasn't really been very good until recently. Expected goals for paints a better picture of the team offense IMO:

This matches my eye test a little better. Last season's team started out strong, had a lull, improved in January, then the snowstorm, until finally improving at the very end of the season and playing pretty well in the playoffs. This season's team had a decent enough start but the play steadily got worse until the beginning of December and now it's on a ridiculous upward trajectory.

And then just as a comparison, here are the Penguins over the same time period:

Since the coaching change they have been consistently above 50% xGF% and close to 65% at times over 10 game samples. This is what the Capitals need to strive to do: maintain this consistency throughout the regular season. There is no "hot switch". Trotz needs to make sure he doesn't outsmart himself. Don't go back to Niskanen and Alzner as a top pairing. Don't turtle. Don't discourage high risk plays in favor of "playing it safe." Keep the foot on the gas like they have for a few weeks straight.

(As a side note, looking at the Penguins graph, their only real valley last season after Sullivan took over was around the time they played the Capitals in the playoffs. The Capitals were very good last season but they are even better right now.)

 

[RandyHolt]

Thanks Twabby. I agree the xGF seems to match the eye test. If you post stats that don't match an eye test, folks will instantly question data and cast doubt.

Interesting that we knocked Pitt down a peg, before continuing on to the cup. It seemed like the PP had a big hand in helping us hang with Pitt (not 100% sure), and I am not sure how significantly PPs etc are reflected in the graph, if at all. We all know our PP had been a huge part of our offense, before this year.

Maybe the PPs failures made Barry re-think things. Grind em down, force them to take penalties, Ovi one timers FTW - win 3-2 - game planned or not, was our recipe for victory. Not anymore. Some blame Carlson. I personally think PKers finally learned. You could see it yesterday, as soon as Nick went up top; they stormed both Nisky and Ovi at the same time.

 

[Stewie G]

It's always been my understanding is that more stats are done solely at ES. To follow up on Randy's post, are there any stats that have been found to be worthwhile when evaluating PPs/PKs?

 

[Sam Spade]

Very happy with MOST of the advanced stats this year, but that PDO is way too high. Not a deal breaker for me, getting CF% above 52% is more important, but I still don't like it.

https://stats.hockeyanalysis.com/teamstats.php?disp=1&db=201617&sit=5v5&sort=PDO&sortdir=DESC

 

[twabby]

It's always been my understanding is that more stats are done solely at ES. To follow up on Randy's post, are there any stats that have been found to be worthwhile when evaluating PPs/PKs?

Typically the measure used in the past to measure special teams effectiveness was unblocked shot attempts per 60. There's more defensive structure on the PK so blocking shots becomes more of a strategy rather than desperation like it is at even strength, which is why Fenwick (unblocked shots) is used more than Corsi (all shots).

Unsurprisingly, Washington, SJ, and Pittsburgh are all in the top 5 over the past few years in terms of FF60 on the PP. And on the PK, only the Kings are better than the Capitals in terms of limiting unblocked shots against this season.

There's more work being done to measure shot quality on the PP including expected goals (Corsica.hockey has a description of expected goals on their blog).

 

[twabby]

Further to the tweet Langway posted in the GDT. It's a disturbing trend but it's hard to blame it all on Carlson's absence when the coaching staff IMO hasn't done a sufficient job identifying their best even strength defensive pairings.