Category Archives: Fancy Stats

Beer League Hero, Edmonton Oilers, Fancy Stats, G Money, NHL

G-Money on Dealing Nugent-Hopkins

5 Comments

***Yesterday we posted an article speaking to the disappointing seasons that Ryan Nugent-Hopkins and Jesse Puljujarvi were having and I asked the boys for their thoughts on what we should do with the Nuge. You can read that article HERE.

I did the same on Twitter with this poll and I’m not surprised with the results. It’s not like my followers are wrong 🙂

It’s funny, each time I’ve decided to post an article detailing how unimpressed I am with a specific player, they decide to go ahead in the following game and put up some points… Maybe I should do it more often.

Now, as I mentioned in the article yesterday, G-Money’s (@oilersnerdalert) excerpt was lifted from a longer reply to my question and now I’d like to share G-Money’s full reply with you. Bon Appetit! – BLH***

Q: What’s wrong with Ryan Nugent-Hopkins, and Should the Oilers Trade Him?

A: a. Nothing other than a run of bad luck, and b. God no!

On point a, one of my lenses for looking at a player is “WoodMoney“, the matchup-based quality of competition methodology that @Woodguy55 and I put together. Here’s a look at how much time Nuge is spending facing the various levels of competition, and how he is doing so far this year as compared to the other two main centres (I’ve included both Corsi and DangerFen for the Elite tier, but only CF% vs the others so as not to turn an intimidating table of numbers into an overwhelming one ):

Nuge

% TOI vs Elite – 41%
CF% –  45.9%
DFF%  – 44.9%
% TOI vs Middle – 39%
CF% – 53.9%
% TOI vs Gritensity – 20%
CF% – 59.7%

McDavid 

% TOI vs Elite – 32%
CF% – 53.4%
DFF% – 56.0%
% TOI vs Middle – 46%
CF% – 55.1%
% TOI vs Gritensity – 22%
CF% – 56.8%

Draisaitl 

% TOI vs Elite – 27%
CF% – 49.5%
DFF% – 48.0%
% TOI vs Middle – 49%
CF% – 50.0%
% TOI vs Gritensity – 24%
CF% – 55.0%

Conclusions

On point a:

1. McDavid is stupid good. He destroys everyone.

2. Nuge is being used by TMc as the shutdown power vs power centre this year. Not McDavid. Not Draisaitl. Nuge is the guy spending 41% of his time against the best players in the NHL. That’s creating a ton of clear air for McDavid and Draisaitl. If you’re comparing things like points, you better take that into account. Nuge’s points are being sacrificed to give the other two a chance to score more.

3. When Nuge is up against those great players, it’s true he’s struggling to keep his head above water.  Moreso than in years past.  And he’s not the only one. My suggestion: give him Eberle and Pouliot on an ongoing basis. Let those two (who are both struggling) right their ships. Nuge’s ship will get fixed right along with them.

4. When Nuge is not against those great players, against pretty much every one else, he runs roughshod.  The Nuge is Yuuuuuuuge!

5. So there is nothing wrong with Nuge, except:

On point b:

Nuge is shooting at 5.1% this season. He has a career average of 11.2% prior to this season. So he’s shooting at less than half of his career average.

He’ll find his groove again, guaranteed.

Every player’s sh% varies wildly above and below their long-term average. And it’s more or less random (if a player could control it, they’d always shoot above their average, which would raise their average, which means they’d shoot at random above and below that average, which…)

That’s just how it goes. Sh% controls you, you don’t control sh%.

Now as for trading Nuge … well, my thought process is always that whether it makes sense to trade a player is based entirely on the return.  Anyone is on the block if what you’re getting back is good enough.

But you know what would be stupid though? Trading a player at what would in effect be the maximum possible discount because of one of those sh% lows.

***With G-Money’s balanced analysis and down-to-Earth reasoning, it’s hard, for me at least, to want to move Nuge ASAP because I’m curious as hell as to what the Oilers might look like if they have all three of McDavid, Draisaitl, and RNH humming along on the offense.

What do you think? Let us know in the comments below!***

Grab a Nugent-Hopkins 16-Bit Superstar shirt by clicking on the pic above!!

Edmonton Oilers, Fancy Stats, Michael Sifeldeen, NHL

The Case *For* Kris Russell

2 Comments

There has been a lot of noise made in the last couple days in the Oilogosphere about Kris Russell and how he’s literally the worst hockey player to ever hockey. Many of the advanced stats purveyors just have a field day when it comes to Russell and his (real or perceived) deficiencies, mostly centering around his basement-dweller numbers in categories like Corsi- and Fenwick-For. They really, truly fail to understand what professional hockey general managers see in the guy. The fact that his on-ice shooting percentage and save percentage is historically through-the-roof and currently 2nd amongst Oilers d-men (along with his PDO, too), are often brushed aside as “luck” or “flukey.”

Boy oh boy, isn’t there a word for selectively applying the statistics that help reinforce your preconceived notions and support your argument, while simultaneously ignoring or underplaying the ones that work against you?

From what I can tell, from watching almost every one of the Oilers’ 25 games so far this season, Kris Russell is an enigma that defies conventional stat-tracking. Watching him, seeing his play, I just cannot understand how this guy keeps getting such shade thrown on him by the pundits. He’s like the hockey player equivalent of that kid in your high school class who shows up looking sloppy and unkempt and gets sneered about by his teachers who don’t expect anything from him, but keeps getting good marks on his tests and constantly outperforms his perceived shortcomings. The things that Russell does well aren’t very easy to measure in Corsi derivatives, and the things he isn’t strong at are hyper-inflated by those very same metrics.

But the sheer arrogance of some of these writers who, I fear, legitimately feel their armchair analysis is superior to that of a Stanley Cup winning general manager and the teams and teams of people he surrounds himself with is astounding. In fact, Chiarelli himself noted when questioned about his initial pickup of Russell that they knew all about the reviews and numbers on him. And according to the internal, unconventional statistics the Oilers have access to (read: non-Corsi-derivative), Russell is currently sitting 2nd in the NHL amongst all defenders in clean offensive and neutral zone entries. And in the pressure system the Oilers employ, this type of skillset is immeasurably valuable.

You know how you can see how valuable it is? By watching Russell play hockey for the Edmonton Oilers. By seeing how he gels with Sekera, and how relatively smoothly and quickly he’s picked up playing on his off-side. (A feat many of the naysayers are unwilling to acknowledge the difficulty of.) Perhaps listening to Sekera himself –who praised Russell from Day 1 saying how “easy” he makes his life — would be another good place to start. Maybe acknowledging the fact that, while yes, an NHL-high number of blocked shots might indicate a propensity for having a history of being stuck in the defensive zone too long, it’s also an indicator of a warrior mentality – someone who is routinely willing to sacrifice his body as a last-ditch effort to stop a scoring opportunity and according to Woodguy, blocked shots cut down on dangerous fenwick. It’s also just simply a fact that shots against are going to happen, and when they do, it sure must be nice for the rest of the Oilers knowing they have a guy who is absolutely going to jump in front of them and take the punishment they entail. It must be nice finally having a couple veteran defenders on our second pairing with very recent playoff experience who can chew up minutes and help guide the next generation in their development. It must be nice having a quick-skating D-man who stays poised in PK situations and never really looks flustered or panicked.

Listen, I was as anti-Russell as the next guy when we first signed him. I saw all the stats and heard all the doomsday predictions that came along with them. And I’m a big fan of fancystats! I use them poorly and with an air of superiority all the time! But here are the objective, observable facts: we’re an above-.500 hockey team that is pretty seriously in the playoff conversation at the quarter-mark of the season, which has been so obviously out of the question by this time during the Decade of Darkness(TM), it’s not something we should understate the importance of. If you don’t think Russell can be said to be partially responsible for that, you haven’t been paying attention. You haven’t been watching him play. If we sign him for another 3 or 4 years at some value <$4.5M/per, I take that deal every day, and I think Peter Chiarelli does, too. And he’s smarter than us and gets paid far more handsomely than us to make that call.

Come fight me on Twitter @sife

Click the pic and grab a 16-bit McDavid tee!
Beer League Hero, Edmonton Oilers, Fancy Stats, NHL

Matt Benning: The Oilers’ Best Defenseman… Really? Maybe. #SmallSampleSize

Before we get into the meat of this beast I’d like to preface it by letting you know that you don’t have to believe what you’re about to read and you should probably do yourself a favor and check out the data for yourself and draw your own conclusions.

  • corsica.hockey (here)
  • stats.hockeyanalysis.com (here)

Also… #SmallSampleSize

The Oilers have a handful of very nice young dmen patrolling their blue line this season. Oscar Klefbom, Darnell Nurse, Andrej Sekera, Eric Gryba, Kris Russell, and most notably Adam Larsson but has Matt Benning been the best of the bunch since being called up on the 21st of October?

Let’s toss some numbers out there to gander upon.

For this part, I enlisted the help of our in-house stats guru, G-Money (@oilersnerdalert) and asked him this:

If you were to take a handful of stats from corsica.hockey in search of the best dman on the Oilers, which stats would you take?

To which G-Money replied:

First and foremost would be the ‘against’ metrics: CA (raw shot attempts), FA (unblocked shot attempts), and xGA (expected goals against).  I like them as rates, so you adjust for ice time.  CA/60 etc. To my mind, that’s a primary marker. 

Secondary are the balanced metrics e.g. CF%/FF%/xGF%, since a high ‘against’ number can be offset if the defender also creates a lot.  And either way, you want those to be above 50% if you can, and if you can’t, at least the “Rel” version to be positive so he’s showing better than his team. 

And of course, the WoodMoneys! Which show you how the numbers break down vs comp. I have the numbers generating for this season so I can supply you with those if you like, though the sample sizes are still smaller than I’d like.

So… We are here and here we are. The numbas!!

Matt Benning (9gp 0g 2a 2pts +3)

CA 101 (1)
FA 74 (1)
xGA 4.41 (1)
CA/60 47.53 (1)
FA/60 34.82 (1)
xGA/60 2.08 (1)
CF% 57.2% (1)
FF% 34.82% (1)
xGF% 60.4%(1)

Stats courtesy of Corsica.Hockey.

Benning’s rank within the Oilers defense is the number in brackets.

We still need to remember that he has only played a handful of games and according to Corsica.Hockey, he’s only been on the ice for 127 minutes. The next player with the least amount of regular ice time is Kris Russell and he is last in most of the categories above… Could that be due to missing training camp? Who knows?…

So apart from that FF% (which we want over 50%), we find Benning at the top of every major statistic G-Money has instructed us to use in our search to determine the Oilers best dman.

But how about to our eye?

For me, his best game was the last one versus the Kings and that’s probably because he was smashing bodies left and right and didn’t make any major mistakes.

I like what Bruce McCurdy had to say about that performance in the Cult of Hockey’s post-gamer:

#83 Matt Benning, 6. Joined the back end hitting spree with his most physical game as an Oiler. Really cranked a couple of guys, even as one booming hit took him out of position to enable the Nolan breakaway that his partner Sekera thwarted. Later crushed Kings banger Kyle Clifford with a solid wallop. Excellent shot and scoring chance shares, and the only Oiler to post a +2 on the night.

WoodMoney

G-Money introduced us to Dangerous Fenwick last year and that is basically the Fenwick stat (unblocked shots for/against) but it takes into consideration where the shot was released. Read more about it HERE.

What WoodMoney does is it calculates a player’s fancy stats versus different “calibers” of player. Elite, Middle, and Gritensity. Read more about that HERE.

I just want to focus on Benning’s Dangerous Fenwick For/Against (DFF/DFA) here because I feel like we’ve already gone over the other fancies above. Also, these numbers are good up to last Thursday.

Stat Elite Middle Gritensity
DFF 16.9 (7) 50.3 (6) 28.7 (6)
DFA 17.4 (1) 39.6 (1) 12.6 (1)
DFF% 49.3 (1) 56 (7) 69.5 (1)
DFF/60 32.1 (6) 55.3 (1) 66.7 (1)
DFA/60 33 (1) 43.6 (6) 29.3 (1)
TOI 31.37 54:32 25.50

Again, numbers in brackets are where Benning ranks within the Oilers defense. That defense being Klefbom, Larsson, Nurse, Sekera, Gryba, Russell, and Benning himself.

So my understanding is that for the “against” numbers, we want that to be as low as possible and the “for” numbers to be as high as possible. Naturally. what I’m reading in Benning’s numbers here is that

Now, what I’m reading in Benning’s numbers:

  • Versus “Elite” competition, he struggles… As he should.
  • Versus “Middle” competition he does better
  • Versus “Gritensity” competition he does worse than versus “Middle”…

I think that since the rookie has been playing the majority of his hockey versus “Middle” competition, the numbers are less up-and-down. I mean look at his DFA. Very low versus “Elite” and “Gritensity” but much higher versus “Middle”…

If you can’t tell, I’m not all that versed in the fancies and reading G-Money’s numbers is like having a driving wheel in my zipper… It’s driving me nuts.

Look, I could be reading all of these numbers all wrong. I could be jumping the gun too because Benning has only played such a low number of games. But can we take into consideration that he basically came on when the Oilers losing streak started (It’s all HIS fault!! BURN THE WITCH!!) and the fact that he’s putting up these kinds of numbers is pretty bloody rad! One only wishes he could have a cannon of a slapshot and was putting up Weber-like numbers…

I know that Russell and Gryba are coming back but I’d be hard pressed to send this guy back to Bakersfield with the way he’s playing right now. Benning is a right-handed defender (something the Oilers need), he can skate and move the puck, he can hit and I believe he’s been getting some time on the powerplay as well.

I vote he stays ON the island and maybe it’s time we shipped Gryba out. At the very least you have Benning and Nurse playing together. I know they had a shift or two against LA together and I was pleasantly surprised at how well they did in that limited time. #SmallSampleSize

What do you think? If I’ve missed anything or misunderstood something, please let me know in the comments below or on the Twitter machine @beerleagueheroe!

If you’re a fan of Lowetide, you need this shirt! Click the pic and get yours today!
Click the pic and grab a 16-bit McDavid tee!

 

Edmonton Oilers, Fancy Stats, G Money

It’s way too early for this dashboard

It’s not enough data!

Normally for a shot metrics dashboard like this, you’d prefer to have at least 10 and preferably 20 games of data.

But I’m showing it to you anyway!

The reason is because, well, it actually does have a few interesting things embedded in it, and I’m going to walk you through what I see. Would be interested to hear if you observe anything else!

So without further ado …

Here is the dashboard, annotated with numbered points of interest (click to embiggen, an explanation of the dashboard is at the end if you haven’t seen it before):


Analysis

So referring to the numbers above, and remembering the caveat about how early as it is, here’s what I see of interest:

  1. Oh what a surprise, the McDavid line tops the board!  The biggest thing to note here is that big discrepancy between McDavid’s blue Corsi line and his purple Danger line. Most players I expect will have those converge over the season. I expect with McDavid that divergence is going to stay.
  2. Mark Fayne only played the one game but hey, it was really good!
  3. I still think the best thing for his development (and the cap situation) is for Jesse P to spend a good chunk of time in the AHL.  But these are nice numbers for a raw rook.
  4. Oscar and Adam aka KlefLar aka AdamBom are giving up an unhealthy edge in shot attempts, but are doing very nicely on the danger aspect.  As I noted in point #1, this is a mild cause for concern, as these two tend to converge over time.  However!  Larsson has shown for the last couple of seasons that he is a better danger suppressor than he is a shot suppressor, so I expect that we may see this ‘poor shot, good danger’ being a persistent effect too. I can live with that!
  5. You can clearly see the recent uptick for Gryba and Nurse – they’ve played two really solid games in a row.
  6. Russell and Sekera started very strongly but have been falling off the pace the last while. This is what a lot of the “Russell warners” suggested was likely to happen.  I don’t think it’s overly indicative either way, and have very much liked both Russell and Sekera’s game so far.  But it’s something to keep an eye on.
  7. Like Nurse and Gryba, Nugey had a rough start to the season, but the last couple of games have been really solid, and against Vancouver I thought Nugey was the best of the Oiler forwards.

The other notable issue is that performance of the team outside of the McDavid line is a significant concern.

Those are the main things I see on this dashboard this early in the season.  There are other things of note (like Pitlick’s shot metrics vs his 20% sh%!, but those are exactly the kinds of things that it really is too early to comment on. I’m more interested in some of the early trends around the key players and prospects).

How about you?  See anything else interesting?

PS. Dashboard Explanation

If you haven’t seen this dashboard (I tweet it out regularly through the season, and will do more of these walkthroughs here at BLH every 20 or so games), here is how to read it:

  1. Every player with even a single game will show up.  Blue background is a forward, yellow is for defensemen.
  2. The list is sorted by raw Corsi differential
  3. The blue/brown trace shows you raw Corsi differential over the course of the season so far. Blue means above breakeven, brown is below.
  4. The purple line shows my Dangerous Fenwick metric, which takes unblocked shots and adjusts them for danger (based on shot type and distance)
  5. Obviously, the farther above the line, the better!
  6. But really what you’re looking for is things like trends, hot and cold streaks, big divergences between raw and danger metrics, and so on
  7. I call it a dashboard because in a single image it gives you a really clear look on how every player on the team is doing over the course of the season, and that’s not easy to do!
  8. Feel free to tweet questions at me @OilersNerdAlert. You can also post questions in the comment section below, but for some reason my comments get eaten on a regular basis, so I might not be able to respond!
Edmonton Oilers, Fancy Stats, G Money, NHL

Vital Signs – Late October 2016 Edition

1 Comment

Every once in a while, I do a ‘vital signs’ analysis on the team.  I look at a variety of indicators that give us a look under the hood of how the team is doing on a number of important indicators.

Why do I call them important?

Because they tend to predict how well the team is going to do in the future much better than do the above the hood indicators like goals and win-loss record. The under the hood indicators tell us whether recent success (or failure) is likely to sustain.

Now that said, it is very early days, and so these numbers are going to shift around a lot (unstable condition, a symptom of life). Even so, they do carry a little bit of information right now, so the peek under the hood isn’t entirely futile. And it supports the idea of staying level headed – something that becomes all the more important when the team rockets off to the best start in many years!

The Vital Signs Monitor

To make the measures a little easier to review, I take my inspiration from the Star Trek vital signs monitor, to wit:

The main difference when using this idea for hockey is the switch in colour scheme. Unlike vital signs where normal is good and high or low is bad, with hockey the middle tends to be good (for a middle pack team), the high numbers are often unsustainable, and the bad numbers you hope will rebound. So we adjust.

Let’s see how that looks for Our Oilers (click to embiggen; data from corsica.hockey):

Down below, I explain the rationale for the indicators I’ve used.  If you’re not interested in such arcanery, just pay attention to the rankings for overall, EV offense, EV defense, special teams, and what I call ‘unpredictables’ (sh% and sv%, which are volatile and for most teams without either elite or terrible goaltending tend to move strongly towards the midline as the season progresses).

What’s the Frequency, Neal?

First, before we mull on this chart specifically, take a minute to think on what you think the chart should say by the end of the year. Are the Oilers a top 10, middle 10, or bottom 10 team?

In my view, there are some terrific strengths (McDavid, C depth, the new top pair) married to some real flaws (backup goalie, right wing depth, not enough righties on defense). I’d call this a mid pack team – I don’t expect the current record to sustain, but I don’t expect a fall all the way back to historic depths either.

IMO

In that context, this vital signs set actually looks OK to me.

  • Sh% is high and will likely come back to earth.
  • Sv%? Well, it’s not out of the question that the Oilers have a Top 10 goalie if Talbot stays healthy.
  • The PP shot rates give me great cheer. These are excellent numbers, and exactly what you’d hope to see a team loaded at forward able to do. Huzzah!
  • The PK shot rates cause some concern, and that SA60 number needs to come up big time.
  • The EV CF% is not very good, but most of that is the poor shot rates against. Some of that is score effects, as the Oilers have opened up the score a few times. But it is concerning. Like a few others, we’d like to see it move up towards the middle as the season progresses.
  • The xGF numbers are higher than the CF numbers. Both have comparable predictivity, so I’d expect the two of them to converge in the middle. And middle is good.  By the way, in December of last year the trend was reversed – the Oilers were having trouble converting their shots to dangerous chances, so this is not a bad place to be.

Bottom Line

I’m usually quite curmudgeonly when I look at the Oilers fancystats.  They haven’t been good over most of the last decade – and they’ve been right.

This year, though, I’m a little encouraged. Overall, this is kind of where the Oilers should be if they are to be a middle tier team.  As long as the raw shot metrics (CF/CA) move towards the danger metrics (xG) and stay somewhere in that middle lane, life is good.  We’ll keep our eyes peeled.

But there’s a bright side to the great start, even if it’s unsustainable. Because the early points are banked, they can’t take them away, and other teams have ridden an early hot streak and late season mid-tier play to a playoff berth.

So why not us?

Happy i.e. Ungloomy Halloween!

Addendum – Metrics Rationale

Ferda boys.  Ferda!

I use these metrics, for the following reasons.

1 – CF% (5v5 Corsi For %) is the single best predictor of future success we have (it’s not great, but it’s as good as we’ve got).  I break the components of this down into rates to look at the offense and defense separately.

2 – I roll in the corsica.hockey expected goals (xG) to get a sense of whether the shot attempts levels that Corsi is giving us are translating into dangerous chances.  Again, also looking at offense and defense rates broken down.

3 – I look at shot rates on the special teams. I don’t use Corsi in this case, because I’m not interested in possession so much – the team on the powerplay should have possession as a given! Rather, I look at how many unblocked shots the team gets on net (or gives up). This is a sustainable skill on powerplays.  I add in the expected goals, again to incorporate the degree of danger for and against.

4 – Just for fun, throw in the shooting and save percentages. These don’t tend to stick or predict very well, but usually if they are unusually high or low, you can expect them to move back towards the middle as the season progresses.

If you’re a fan of Lowetide, you need this shirt! Click the pic and get yours today!