Random Thoughts – Dawning of the Age of Fancy Stats

It’s been a pretty weird year for those of us who were around the the beginning of hockey’s “advanced stats” movement. At the start, there were never any thoughts about popularizing their use or getting them into the NHL executive offices. It was simply a collection of enthusiasts asking questions, gathering data and arguing over the implications. That’s why the nomenclature is so odd – Fenwick, Corsi, PDO – the stat names were simply shorthand monickers that stuck. There was no consideration about accessibility (ie; marketing).

Now national broadcasts are mentioning these odd words during games. Pundits argue their value and utility between periods. The esoteric debates of a handful of guys with silly handles on blank template blogspot sites has somehow trickled upwards into popular consciousness. It’s somewhat gratifying to know that the off-hand dismissals and lots (and lots) of hate mail which was the coin of the realm for stats writers for many years wasn’t all for nothing.

– The big test case which seems to have moved the needle in favour of the nerds is, of course, the Toronto Maple Leafs. Toronto came into the season with inflated expectations after an outburst of percentages during the lock-out shortened year saw them break into the post-season. Their awful underlying numbers caused the stats oriented amongst us to be far more circumspect about their chances, of course. Bodog had the Leafs season point total at about 96 before the puck dropped on October. I happily bet the under. 

– It’s not just that Toronto finished the season with just 29 regulation wins, 84 points and a -25 goal differential…it’s because they failed in such dramatic fashion. Their last quarter collapse wasn’t specifically predicted by their lousy possession rate (aside from the fact that getting horribly out-shot simply makes a collapse more likely), but the NHL’s biggest franchise wheezing to a halt five feet from the finish line seemed to put an end to the debate with an emphatic exclamation point.

– Of course, the debate isn’t over as far as some last holdouts are concerned (see: Steve Simmons, Damien Cox, Dave Shoalts and the two thirds of FAN960 morning show), but it is over in all the ways that matter. The utility of possession can be ably demonstrated with studies and evidence. People like Ray Ferraro are going on record as in-favour of the new stats. Journalists like James Mirtle and Bruce Arthur with national publications frequently integrate evidence-based analysis in their work. Nate Silver’s 538 was scooped by ESPN and is at the forefront of “data journalism”, where testing assumptions with quantitative inquiries and making predictions is the new normal.

– Perhaps more telling is that the insights gleaned from online Corsi talk has started to infiltrate not just the newspapers and TV broadcasts but the executive offices in hockey arenas as well. The last, best rejoinder in the stats debate for those opposed was always “why don’t the people who are paid to do this use this stuff?” For a long time, the only response was to point at the evidence and shrug. But that’s changing.

The Hawks don’t call it Corsi
or Fenwick. They have their own terminology and methodology. Bowman
hinted that the fundamental goal of the Hawks’ analytics is the same as
the more commonly accepted ones, but he wouldn’t go into any more
specifics.



“What we do is different,”
Bowman said. “I think it’s better, but I guess it’s a matter of opinion.
It’s also a competitive advantage. That stuff’s readily available, but
what we have is more proprietary. Which is why I’m really trying not to
talk about it. I think what we do gives us an advantage over other
teams. They might say I’m wrong, but we’re pretty confident that what we
have works.”

That’s from a recent Chicago Sun-Times article noting how analytics have become a big part of the way the Blackhawks do business.

– Related – if you missed it, Josh Weissbock published something on the possession rates of AHL teams entering the post-season. I can tell you he was contacted by at least one curious NHL organization about his method for determining the numbers.

– Kyle Dubas, the young GM of the Sault St. Marie Greyhounds, discussed yesterday how his organizations’ focus on possession has led them to change their emphasis from dump-and-chase hockey to carrying (and keeping) the puck as much as possible. Justin Bourne sums it up like this:

[T]he best teams remaining all have stars who can handle the puck.
Their defensemen can (Doughty, Keith, Chara, Subban, Letang), and their
forwards can (Kane, Kopitar, Crosby, Getzlaf, Bergeron). You can go
pretty far down most lineups before coming across a name that makes you
skeptical. You have to be able to handle the puck to play in the NHL
today (very few exceptions), and it’s going to be more important going
forward.

If you want to make players better, if you want to
make teams better, you have to understand that possession is at a
premium. Players are too good these days to just give them the puck and
assume that they’ll lose it back to you shortly.

When you have it, keep it. That’s the way it’s going to be in the new NHL.

– The emphasis on carrying the puck across the blueline rather than the “chip and chase” strategy comes directly from the work of Eric T, whose zone entry project and studies showed that controlled zone entries led to double the possession rate as the classic bang and crash method. This, of course, is the next frontier of advanced stats – not simply broad brush correlations and general, long scale predictions, but actionable insights born from a marriage of quantitative analysis and scouting/observation.

For another example of this kind of stuff, be sure to check out Tyler Dellow’s recent work with how different face off tactics can have strong effects on possession rates. 

– Although the fight over whether Corsi and such is relevant is mostly over, the truth is we’re really only scratching the surface when it comes to rally understanding this stuff. In addition, the game constantly changes and evolves, so the need to challenge assumptions and inspect the market for inefficiencies will never ebb. 

– Finally, Corey Sznajder of Shutdown Line is embarking on a project to watch over 1,000 games this summer and record detailed zone entry information which he will eventually make publicly available. This kind of info will be immensely useful, so I urge you to swing by this link and drop a few bucks to help him fund this thing. 

  • Lordmork

    Kent, if you were to peer into a crystal ball, how do you think this advent of advanced stats could change hockey over the next five years? Will we stop seeing goons on the fourth line? Will there be a new focus on players who can possess the puck? I’d be curious on your thoughts, maybe in another post. I enjoyed this one.

    • That’s a good question. I think we will definitely see more puck carrying and less emphasis on “bang and crash” and “keeping it simple”. A lot of the emerging analytics suggest coaches are entirely too conservative when it comes to certain neutral zone tactics and there is increasing awareness that toughness for toughness sake doesn’t really deliver value on the ice.

      I think you’ll also see a lot more puck moving from the back-end. Guys like TJ Brodie will become what scouts look for in blue line prospects. I don’t think you’ll ever get away from size concerns entirely for defenders (and that’s probably sensible0

    • Mort

      There there WW, I know advanced stats hurt your brain, but just follow me down this ladder for a second:

      Nobody disagrees that goals/goals against are the direct cause of wins/losses. But how do you get goals? By shooting the puck (that’s where Corsi comes in!). How do you prevent goals against? By stopping the opponent from shooting.

      Now, WW, I hope you’re still following. How do you get shots? Well, if you don’t have the puck, you can’t shoot it, so the most important thing in that regard is possession. Now, how do you stop your opponents from getting shots? By maintaining possession! Here we are! We made it! We figured out why possession is important! Yay!

      • Skuehler

        Dear Mort.

        My point is not that there are no causal relationships in this world,
        My point is; just because there is a correlation between 2 events (A and B occur together a lot) does not mean that A causes B.

        There are other possible explanations, for example:
        Maybe B causes A, or C causes both A and B, or A equals B, or your sample size is not big enough and the correlation is a coincidence………

        Let me give you an example:” studies have shown that people who have more birthdays tend to live longer!!!!”
        Are we to conclude that birthdays are good for your health because there is a strong correlation?
        You have just found the fountain of youth my friend!!!

        The problem with advanced stats is not the “advanced stats”……..it’s that the people using the stats may not be “advanced”……

        Who’s brain is hurting now??

        WW

        • Skuehler

          you’re being purposefully obtuse about this. please stop.

          these arguments, on both sides, will continue ad infinitum – but it doesn’t matter. the war’s over. if you don’t think this stuff has more value than what you’ve thought about hockey all of your life, you’re wrong.

          • Skuehler

            No , I don’t” think this stuff has more value” than what I have thought about hockey all my life…….I think it “adds to” what I have thought about hockey all my life!

            Why do you run a piece on advanced stats if you don’t want us to debate the merits of them???

            WW

        • Mort

          “‘Studies have shown that people who have more birthdays tend to live longer!!!!’ Are we to conclude that birthdays are good for your health because there is a strong correlation? You have just found the fountain of youth my friend!!!”

          Maybe you just can’t tell whether goals produce shots or shots produce goals. Fortunately, WW, I’m here to tell you that shots definitely cause goals, and not the other way around! 🙂

          Anyway, I would hope that anyone who isn’t a moron could see that living longer will bring about more birthdays. So, if for some reason, you made it a goal to have more birthdays, you would have to live longer to achieve that. How do you live longer? Different people say different things, and you can go with whatever you think is right.

          The point is, birthdays do not beget birthdays. Goals do not simply occur from nothing. For every effect, there is a cause. Advanced stats is an effort to examine the causes.

          “The problem with advanced stats is not the ‘advanced stats’……..it’s that the people using the stats may not be ‘advanced’…”

          Have you ever heard of “argumentum ad hominem”? 🙂

  • I don’t think anyone is claiming that stats hold all the answers or that they will capture 100% of what makes a hockey player useful. But just because they dont tell us EVERYTHING doesn’t mean it can’t tell is ANYTHING.

  • BurningSensation

    I am going to go out on a limb and suggest that the Flames will be one of the first five teams to get ‘SportsVu’ cameras for improved data collection.

    Burke likes having an edge and its a way to spend money for one that won’t affect the cap.

  • Truculence

    Coming from a Bachelor degree in Industrial/Environmental Psychology, I totally see Analytics evolving into individual teams formatting. Right now the stats are being viewed from an observational perspective of what is happening on the ice by individual players. Ultimately, everyone wants stats to work as a predictor of a players performance, but there are too many variables that underlie performance. What I think is a reasonable goal in the use of analytics is to use this to create profiles similar to profilers that work in the law enforcement. These profiles based on the analytic data will ultimately help teams make decisions on drafting, free agency signings & trades. I can see how teams like the Hawks will come up with their own processes & analytics platform that in their view equate into the best chance for long term success.

    Analytics is here to stay & will be huge valuable tools with most NHL teams. If you manage to get one of them to tell you their specific analytic system, naturally they will have to kill you afterward. 🙂

    • Bean-counting cowboy

      “profiling” – very interesting concept. Anyone watch Criminal Minds on TV? Not quite the same, but some of the same concepts could be used.

      I like the idea – you lump prospects or players into their given profile, which has been based on pre-populated database info based on all kinds of metrics including analytic stats (zone starts, quality of comp, corsi etc.) This would include lumping them in based on historical data of successful players with a similar profile. Then when you have several players you are looking at with equal profiles and you can’t decide – you look at risk factors.

      Risk factors would be less quantifiable metrics where things could go wrong with the player. These could be such things as any history with drinking, gambling, drugs, etc. in the past; another risk factor could be likelihood to return to homeland if a foreign player, other “risk factors” could include their upbringing, parent’s athletic history/ability, body metrics (potential for weight gain and not the good kind), any history of missing team practices, meetings, commitment issues, etc. It may seem highly subjective and judgmental, but if a weighting was given to multiple “risk factor” areas, based on available data, it could help the decision between two players that are seemingly equal.

      • Bean-counting cowboy

        Each team will no doubt weight what they feel is important in what traits a player should possess or the traits that they would covet. We hear lots about hockey IQ, we hear about top 10 prospects scouting reports outlining strengths & weaknesses. As the statistical data increases, it will be able to assist to identify & calculate probabilities that a player with certain deficiencies or tendencies(decisions made during a game) can become X type of player. It will also assist in the structuring the development programs of young prospects in the system. It wont be perfect but it sure can assist GM’s in hopefully making the right decisions more often. So many young players have talent, that’s why they were drafted in the first place & is why later round picks can become elite. The talent is in all the kids, it’s the other things that NHL teams are trying to get an advantage on each other. I think stats is one way of doing that & will be fun to watch how it evolves. Adapt or be left behind.

    • Truculence

      Totally unrelated, but I’m curious as to what kind of job opportunities are available to people that graduate with a bachelors degree in “Industrial/Environmental Psychology”. More to the fact, what the hell is that?

      • Burnward

        Hahaha! Well it was basically a Psychology degree with many courses in Organizational Behavior & Design, Industrial & Environmental Psychology. It was geared at creating best performance, looking at the work setting & geared in a Human Resource type of career. Looked at personality types ie. Fear of Failure/Need to Achieve profiles. I also was able to take Sport Psychology & was really interested in that field but very limited opportunity in Calgary. Now keep in mind I graduated in 1982 right about the time Trudeau gutted the Oil Industry with the NEP & career opportunities went out the window. 1st departments that were eliminated were the HR & out there stuff I wanted to do. Funny enough I wound up being a licensed Private Investigator (Jim Rockford only I drove a Camaro instead of a Firebird) & then eventually got into the money business & own a Brokerage Company & do some loan sharking on the side. What can I say, I don’t even think the type of courses & programs I took back then are even offered at the University’s anymore. Sometimes I feel like that old guy that never got to bat in Field of Dreams, but I have no regrets. I find this site interesting because I had always been interested in predicting & enhancing performance & the interaction within the workplace. Don’t think analytics in hockey quite have it down yet but there is a place for analytics in hockey. Anyway, you asked, don’t get me going, I like to ramble. :-}

  • TheoForever

    The commentators on tv yesterday were talking about having to play on the edge and be willing to play dirty hockey in order to win the cup.
    Heart, toughness and determination to be successful are important.
    Those factors are measurable. If you look at junior players drafted very often talented kids fail to become Nhlers, yet some less talented bigger players make the show. Hard work and willingness to sacrifice the body come from within. Some people want to be pro players badly enough that they will do anything.
    Talent is not always enough, you need to work hard and you need to be able to take physical punishment.

    • ChinookArchYYC

      the reality some people are more willing than others to make the sacrifices necessary to become “pro athletes”.

      Ryan Howse is a perfect example. Tons of talent and potential but he just couldn’t do what was necessary to become a pro athlete – stuff like Diet… drinking etc… you need a tremendous amount of self control to even build the body of an athlete.

      that’s why I think so many orgs look at a player’s character.

      • Parallex

        Howse did NOT have a ton of talent… he was mediocre back in junior. Nice shot… eff all on all other tools. I remember watching some Chiliwack games the guys I noticed were Horak and Sundhar… Howse was defensively inept, an unremarkable skater, and selfish with the puck. Back in another thread I said Bouma was a guy you could just tell was going to make the NHL… House was a guy you just knew wouldn’t

  • seve927

    I think what Walter is saying, and it would take hours of work to produce any counterargument, is that it’s not particularly remarkable that when one team is much better than the other, it will win most of the games. 25-7 in games where there is a 5% difference in corsi?

    Questions:

    1. How does that compare to any other way of analyzing the results?

    2. 32 games in 7 years? Why use a criteria which yields such a small sample size?

    3. A whole bunch of other stuff for anyone who had time to analyze it properly.

    So I tend to agree with WW, that’s a pretty meaningless stat on it’s own.

    • It shows that, in that time, the fenwick close numbers had 78% successful prediction rate.

      Walter White’s goal differential, on the other hand, can only be used in hindsight.

      The argument has never been “Only look at shots! Goals? Bah! Useless!” The argument is, if you try to predict using goals, you’ll be far, far less successful than someone making predictions using corsi/fenwick data.

      • seve927

        I’m not disagreeing. I’m just saying it’s an assumption. Show me a complete analysis. 78% in a small sample size – I wouldn’t be at all surprised if I could do the same for a whole lot of different meaningless stats.

  • Jeff Lebowski

    The issue I have with the current state of affairs with the stats people are:

    -Corsi is a team stat. It doesn’t tell you what a specific player’s impact was in generating a shot. What kind of shot? (a legit scoring chance or a feeble last resort to throw puck at net).

    But on a shift to shift basis just looking at corsi – what do you ‘know’? Not much without eyes. What happens is a lot of non scientific assumptions somehow get labelled scientific because someone is using ‘numbers’.

    You ASSUME possession and zone time and passes because you had to have the puck on your stick before you shoot it?

    But there are a lot of different shots out there AND some guys are just along for the ride. Meaning they do squat with the puck but they may get a good rating. Over the course of a season – sure – it tells you something.

    In essence there is unquestionably value in the meta of corsi. However in a granular view there is just as much subjective guess work and confirmation bias as advanced stats mortal enemy: eyeballs.

    It’s getting better all the time and to dismiss it outright is foolish. It is not an elegant stat on its own.

    • seve927

      advanced stats mortal enemy: eyeballs.

      Where does this perception come from? It has no basis whatsoever in fact. No one–anywhere–dismisses the eye test.

      • Jeff Lebowski

        But stats are being used to say: this is what’s really happening. What your eyes can’t see.

        Stats for the ‘thinking’ fan. This makes me wanna puke because of how much guess work still is required.

        The point is, it is a cool tool. But the arrogance of stats people to imply they are thinking while others aren’t is a function of low self esteem and ego. Or that they have uncovered some secret sauce that makes their opinions somehow transformed into empiricial truths are laughable.

        The common thread being this:

        Of course corsi is only relevant over large sample sizes and there is a lot of noise but having just said that, lets ignore this and start making some conjectures and assertions about things happening in game. I’ll back it up with these stats that I just stated are only relevant over a number of seasons – but forget i said that – errrr, ummmmm NUMBERS!!! So I’m right and you’re wrong. Show me your numbers!

        Corsi as a proxy for shot differential are a large piece of what happens very quickly on the ice. No doubt.

        The notion that it is the only significant thing – that the shot is the only necessary and sufficient condition for winning hockey is hard for me to accept (and I could be wrong).

        What about the guy who digs the puck out of the corner – who battles in front to screen – or the guy who knows how to space himself out on the ice to open ice for a teammate. Don’t they have impact on generating wins or shots?

        There is nothing in corsi that can quantify this. So amongst proponents it doesn’t exist!

        That is the other thing the smugness that because your tools can’t quantify it – it therefore doesn’t exist. Get better math!! The dude who found the higgs boson did.

        • Parallex

          “Don’t they have impact on generating wins or shots?”

          Sure they do… to use your example of the guy who digs the puck out of the corner, if he does it successfully he’s taking possession of the puck which will over the long-term lead to shot attempts and on balance positive Corsi/Fenwick events (which lead to shots, and shots lead to scoring chances, and scoring chances to goals, and goals to wins). If you do something that positively contributes to possession it’ll show up in the possession stats.

          I’m not sure where you’re getting the impression that it doesn’t.

          • Jeff Lebowski

            True enough. But compared to his linemate who does nothing that shift but gets a positive rating just like the players who actually made something happen,- or the flip side, the guy who does everything right but either a bad bounce or a missed read by a team mate results in a negative rating – how is they guy who did everything right to be cast in a negative light – how do you differentiate between the two based on corsi/fenwick? The comparison of players is what I meant.

            What’s built in these measures is the shot. That’s it.

            All the other stuff is guess work.

            But this is the type of argument I can get with. Absent of ego and arrogance (which I mention above) and with the requisite intellectual humility I would expect in an empirical discipline lets get some answers.

            Not to pump tires but Parallex and beloch are the kind of stats guy pov that people who are trying to understand can relate to. Compared to others who belittle.

            I deeply, deeply respect the intention to understand things better via analysis. It’s in my own DNA. I can do without the cockiness and smugness.

            I don’t hate advanced stats or the people who write about it per se, If I did I wouldn’t return here. I do return because I learn things. And for that I am extremely grateful.

            Intellectual humility!! Necessary and sufficient for critical analysis.

            And that’s all I gotta say about that.

          • BurningSensation

            I think part of the problem is that not all of the new fancy stats are equally useful, and both critics and proponents often look at them in the wrong way.

            Take Corsi as an example.

            Yes, it is a shot based metric.

            Yes, it requires extended analysis and context to do be explanatory or predicitive.

            No, it doesn’t capture the difference between a failed breakaway or a muffin through traiffic that harmlessly bounces off a pad.

            No, it isn’t directly predictive of a players future performance.

            However,

            It will tell you if your team, or, line, or even a player, is sucessful at driving possession (depending on your sample size of course). Since possession stats do strongly correlate with winning games over the long haul, you can make informed decisions about what is occurring by using stats.

            It does have predicitive and explanatory power. Why were stat-heads all expecting the Leafs to fail as the season wore on? Because they could understand the stats of PDO, Corsi, Fenwick Close, etc. which all suggested that the Leafs were unusually lucky in producing as many points over the half way mark as they did, and that when their luck ran out they were going to crash. And lo, that is exactly what they did.

            As far as I know nobody who is serious about understanding the stats thinks they can replace anything (like scouting), but they aree correct in insisting that these stats do have legitimate value, and that as the data collection improves (SportsVu!) the stats predictive and explanatory powers will also improve.

            To borrow from the (criminally underrated) movie ‘The Crow’;

            “I am the new way to go. I am the way of the future”

          • xploD

            Regarding the Leafs comment: you didn’t need to look at advanced stats to predict their decline; the shots for and against told that story…….you could tell they were being outplayed most nights by watching the game, without having to depend on advanced stats.

            WW

          • BurningSensation

            Sure and some people thought the Leafs would eventually fail because they are cursed by Harold Ballard.

            That doesn’t mean they were right when the Leafs eventually failed.

            Weren’t you the guy who suggested correlation isn’t the same as causation?

  • As a reader at FN over the last number of years, it’s neat to see something that’s been on the radar here burst into the spotlight.

    Still sounds funny when announcers try to use some of the stats during the game though.

  • TheoForever

    The article was an interesting read, but also a lot of back slapping. I’m not surprised advanced stats have advanced. If they work, people will adopt them. That simple. If they didn’t they’d have been buried a long time ago, probably by the same people who were interested in them in the 1st place.

    IMO, the entire argument is over-argued. You use whatever stats you can get your hands on and then make sure it correlates with the eye test and anything you may know from 1st hand, behind-the-scenes knowledge. That’s why the debate about Burke’s love/hate of stats is so irrelevant to me. It’s clear he’ll use both. His entire MO is to marry old and new school together.

  • Bean-counting cowboy

    Good grief. . .this sounds like a convention of climate-change enthusiasts.

    The “science” is settled and all that nonsense.

    I think I will just continue to enjoy the ebb and flow of the playoff games, and the beauty of the game itself.

  • ChinookArchYYC

    Back to the same old arguments for and against Fancy Stats.

    This article is simply showing that Advanced Stats are becoming a more accepted method for the hockey world to watch and understand the game.

    On a personal note, once I buried my bias against Advanced Stats and accepted them as at least interesting, my personal enjoyment and understanding of hockey improved.

    I’ve just never understood why everyone feels the need to fight in favour of or against all stats. Why can’t you like some and not others?

  • mk

    Interesting post. As many people have said in their own words: I find the main reason people don’t accept ‘advanced stats’ is because they miss part of the discussion (or ignore it). As per BurningSensation: Corsi (and shot-related metrics) are a tool that uses shots as a proxy for possession.

    It does not factor in shot quality (which is a much more difficult topic to analyze given unreliable data and seemingly random variations). If you had reliable data on possession time, shot location and other shot circumstances (type of shot, speed of shot, screens, deflections) then you could come up with a much better stat that mere shot differential. But this data isn’t available or is very difficult to obtain.

    Another way of looking at this is to say that ‘goals determine who wins’, therefore players with more goals help to win more. The origin of the assist was an attempt to analyze not only who put the puck in the net, but also what other players contribute to scoring. Goals would say Thornton (342) and Doan (354) are similar in ability, but adding assists gives you additional information about Thornton’s contributions (852 assists vs Doan’s 508).

    Stats are simply a method of evaluating a player: no 1 stat is the end-all be-all of evaluation. Each should be used in context to contribute to the analysis.

    ————————–

    On a not-so-related note: are the prospect-ranking articles going to be weekly? I’m eagerly awaiting #14! 🙂

    • BurningSensation

      Totally agree with your last point, when will we see number 14? Does it mean we have to wait past draft day to get them all? If the draft is at the end of June we not get into this rather than pound advanced stats, some award show, some guys trip etc. I would much rather watch paint dry than some of the stuff posted lately, much of which i read the headline and pass on. There have been a number with 0 responses.

  • ChinookArchYYC

    Read what Scotty Bowman says. They have their own things they look for based on their own philosophy. It may include some of the stuff talked about here or it may not contain any of the stuff like Corsi, Fenwick, etc. And, perhaps the stuff they measure isn’t even all that new or fancy?

  • xploD

    Eyes = Backlund is a good possession player
    / Stats = Backlund is a good possession player

    Eyes = Grats is not a good possession player
    / Stats = Grats is not a good possession player

    Corsi stats and other stats are suggesting possession > dump and chase

    What I see: Possession plays > Dump and Chase (most of the time)

    Point is, advanced stats do make a difference and they DO matter, but that doesn’t take ANYTHING away from having to scout and see these players, but again if they have horrible stats I doubt you’re going to go into a game and see them defy their stats completely or consistently, Unless they’re developing fast, in which stats would also show.

    Correlation doesn’t mean causation, but sometimes it does. Or sometimes it’s just an indicating factor.

    In the end we need stats and eyes. Together they are better than one, and that’s the WHOLE POINT.

    • Jeff Lebowski

      Backlund is viewed even better with this zone entry data.

      His ability to walk the puck anywhere against the toughest guys is going to be that much more valuable as more skill is added to the roster (hopefully).

      I wonder if a guy like Hudler had good possession numbers for his ability to handle the puck?

  • BurningSensation

    @Jeff Lebowski

    I actually agree with much of what you wrote, but there are some items that I wanted to address;

    “-Corsi is a team stat. It doesn’t tell you what a specific player’s impact was in generating a shot. What kind of shot? (a legit scoring chance or a feeble last resort to throw puck at net).”

    Yes, Corsi is a ‘team’ stat, but there are things you can do to extend the stat down to the individual level. CorsiRel, Quality of Competition (which is absolutely necessary for getting a more complete picture, it does no good to say that Brodie is ‘holding his own’ if you don’t know whether he is being deployed against the other teams top line, or just against their thugs and rooks on the 4th), etc. ‘Fancy stats’ have gotten a lot more fancy of late.

    “But on a shift to shift basis just looking at corsi – what do you ‘know’? Not much without eyes. What happens is a lot of non scientific assumptions somehow get labelled scientific because someone is using ‘numbers’.”

    Shift by shift is not the best way to look at Corsi. A single game is a tiny sample size subject to a wide range of effects like luck, etc. Even a quarter of a season isn’t really ideal, but once you have half a season or more Corsi’s value increases in its predictability as the luck effects wash out.

    That said, we now can look at a batch of stats at a time to see what is really going on, even on a shift by shift basis. Consider a guy like Couturier. His ‘counting’ or ‘traditional’ stats are borderline mediocre, but when you dig into the fancy stats you can see that; he gets ridiculously bad zone starts (2/3 of his starts in his own end of the rink), he faces top quality of competition, and he manages to ‘push the river’ possession wise despite these handicaps. At a global level these stats show that Couturier is an excellent defensive pivot who can match up against top lines in bad areas of the ice, and more than hold his own.

    Digging even deeper you’ll find that when you compare head to head matchups of Couturier vs Malkin, you’ll see that Malkin’s #’s are destroyed when he faces Couturier.

    So, on the one hand ‘yes’ you need to have the ‘eye test’, but if you only looked at Couturier’s counting stats (mediocre to poor), and didn’t look at the context those numbers were created in, you wouldn’t have as full an understanding of the value Couturier brings to the Flyers. (incidentally, this is why I don’t see Philly ever trading him to Edmonton for the stinking pile of garbage that is what remains of Sam Gagner’s career, Couturier is a player much better than his counting stats show, Gagner is a terrible player who is actually much, much, worse than his counting stats would indicate. Sorry Rex.).

    “But there are a lot of different shots out there AND some guys are just along for the ride. Meaning they do squat with the puck but they may get a good rating. Over the course of a season – sure – it tells you something.”

    This is pretty much the first thing people complain about when they talk about how all these fancy stats are just ‘counting shots’. Trillions of pixels have been killed discussing how ‘shot quality’ isn’t captured by fancy stats.

    Except….after all is said and done, ‘shot quality’ doesn’t actually seem to be a thing that is repeatable. For example, the stats crowd were down on the Leafs 2nd half of the season because the fancy stats were terrible even though the team had a winning record. It was suggested that the reason why the Leafs were defying statistical gravity was because while they were being grossly outshot each night, the shots they were taking were ‘higher quality shots’. Unfortunately for Leaf fans, this turned out to be wishful thinking. There was no evidence that the type of shots the Leafs were generating were somehow better than other teams, and lots of evidence that some hot goaltending, all round luck (PDO), and unsustainable shooting % were responsible for the difference between actual and expected results. As the season wore on the luck, hot goaltending, and shooting %s returned to statistical norms, and, not surprisingly, the Leafs were taken down to Zed’s basement for a meeting with the gimp.

    Lastly, and for what it is worth, Vegas oddsmakers have long used ‘fancy stats’ to assist them in setting betting lines. The biggest one they use? Fenwick Close.

    If it’s good enough for Vegas….

  • Truculence

    I don’t want to hijack a thread on fancy stats, but since the title reads “Random Thoughts”, I thought I’d throw out this piece from Brad Wilson, one of the writers/scouts that follows College Hockey at Dobber Prospects, on Mark Jankowski:

    “May 2014 – Jankowski took another step forward in 2013-14 while playing for Providence College. He posted 25 points in 39 games and is showing signs of elite level puck control and passing. He desperately needs to put on about 30 pounds but at only 19 years of age and with at least one more if not two more college seasons ahead of him, there is time. The skill is there, the body needs to catch up. He is really slick with the puck and knows exactly how to use his big wing-span to move in and out of traffic. Keep a close eye on his development, this kid could be a lock for the Flames top two lines in a few years.” Brad Wilson

    • BurningSensation

      If Jankowksi ends up being a stud #1 C I may just go through all the comment threads to source his (and Feaster’s) biggest critics so I can name names for a giant serving of crow pie.

      • ChinookArchYYC

        Let me help you out. I was and continue to be a Feaster critic. Not that he was entirely bad (that post is reserved for Garth Snow), but on balance I didn’t like him as Calgary’s GM.

        As far as Jankowski goes, I wasn’t happy about the picky for the simple reason that it was too risky. That said, if Mark becomes a useful NHL player, I’ll happily eat crow.

        • BurningSensation

          I already had you on the list. 🙂

          We’ve had our argument about how ‘risky’ it is to trade down from where they were (and I’ve already eaten my hat with regards to my initial thoughts on Tervainen and Maata), but Jankowski represented Feaster taking a home run swing to find exactly the kind of player (big, skilled, #1 C) the team has been desperate to develop for more than a decade.

          If it works out?

          So. Sweet.

  • BitGeek

    I think stats (of any kind – fancy or otherwise) offer more texture to watching the game. Its still fun to watch without the information that stats provide but more color is added when you look at the numbers.

    Most of the time stats tell us what we already know (Crosby is an elite level player), and sometimes they tell us what most people denied even when using the eye test (yes Backlund was a pretty good player and was due to break out like he did this year).

    They can be predictive over longer periods of time and that helps managers and decision makers with their jobs.

    Unfortunately till this point they’re not very “corrective”. Such is the reason whey players don’t put much value in them. ‘ok so I’m an average puck possession player – what do I do to get better?’. It’s hard to tell a player how to be better at take-aways, and how to be less prolific at give-aways. It’s hard to use stats to create a clinic for a player on how to improve his numbers. That’s what coaches are for.

    Who knows maybe someday stats will be able to help correct problems for individual players and thus be truly useful to them. Until then, they are just numbers that I like to use to enjoy the game more.

    The problem with enjoying stats, is that you can no longer rest your hopes on intangibles (it’s our year I can feel it) and enjoy the hockey game for what it is. You’ll know better and have to look for other things to be excited about even when your team really shouldn’t have a chance – because the numbers just don’t support a winning season.

  • Skuehler

    It’s amazing how absolutely enthralled we men are with chasing balls. We love to chase balls a lot, and during those times where we cannot ourselves chase balls we will passionately watch other men chase balls. When they chase them on behalf of a nation or in the pursuit of winning a cup for one year, we go nuts.
    And when we neither can chase nor watch others chase, we discuss and dissect ad nauseam. There is no meaning to this search for corswick beyond life avoidance and money making.

    Maybe we are all taking this a bit too far? Humans love to make things complex when they could be simple and to try to find meaning in the wrong places. Hockey is not baseball. That’s why I like it. All this stats stuff is a bit of a bore.

  • v4ance

    Advanced Stats proponents can be arrogant and condescending just like people who don’t believe in the advanced stats. The difference is that people like Wilson, Mirtle, Charron, mc79 and others can show that they’ve figured out some predictive factors, but they recognize how much more they don’t understand.

    The stat based analysis that is going on is just scratching the surface. We use shots as a proxy for possession but that is just showing the results of possession.

    Taylor Hall mentioned that he has looked at the stats stuff and hasn’t made up his mind yet. The reason is that the current metrics don’t show HOW to get better, or as Wilson stated, “actionable insights born from a marriage of quantitative analysis and scouting/observation”. Every player wants to put more shots on net but the way forward is how can we use current analyses or new metrics that haven’t been developed yet to find out ways to increase that shot volume, individually, or as a team?

    To compare it to the Greenhouse gas debates, we can measure the CO2 (the results) but we don’t know how to reduce them yet or make it better (actionable). BUT at least we can see the icebergs on the horizon (Toronto Maple Leafs 2013-14, even if the captain of the ship is asleep at the wheel (Carlyle)