Discussing advanced statistics

Happy Friday, BSB community! As we all know, our beloved Rangers missed the playoffs this season.  Suffice it to say, this has created a much longer off season than we have all recently grown accustomed to.  While a huge bummer, I prefer to look at it as an opportunity.  For one, I personally have enjoyed watching stress free playoff hockey for the first time in forever.  Well, that’s not entirely true.  When the Pens, Devils, Bruins and Flyers were alive there was a little stress, but thankfully they didn’t last long.

More importantly, it gives us an opportunity to discuss the game, the team and other hockey related issues without the break-neck pace of a typical off season.  This is exactly what I would like to do this morning.  But, I need your help.  We have an incredibly diverse readership here.  People coming from all walks of life to enjoy content and discussion about our favorite team.  Different ages, professions, education, geography, gender, etc., creates a multitude of different perspectives and ideas, which is a great thing. We don’t always get along, but the variety of opinions is what makes this site so great.

So, here is what I ask of you.  I am going to tackle a difficult subject in the hockey world: the application of advanced statistics to the casual fan.  This is obviously ground that has been covered before.  However, I want to take a different bent on it.  I was discussing this issue with a couple of gentlemen on Twitter yesterday and would like to get some additional feedback. So, as I go through this discussion, I would ask two things: 1) be honest with yourself about how you feel about these issues and if you choose to share your thoughts on this post, be honest in your comments on it, and 2) be willing, if just for one day, to cast aside the stone throwing and animosity you may have to the other side of this debate. Everyone good with that?  Good.

The big question raised was what barriers exist to helping the casual fan get on board with a deeper mathematical analysis of the game? This seems like a simple question, but it is actually quite a lot to unpack.  Let’s start with the current state of advanced hockey statistics.  At this juncture, we have correlations.  We have enough data to suggest that certain characteristics positively correlate to certain results.  That is not to say we have proof of such things, but we believe we can use their predictive value to make better decisions.

This theory has been applied through numerous, and relatively recently created statistical benchmarks.  I think the creators of these metrics and even their most devout believers will concede that they are not perfect, and they cannot be used as the end-all, be-all reference point for either constructing a roster or debating a co-worker at the water cooler.  To be used correctly, they must be looked at in context, in conjunction with other markers (some more developed than others) to arrive at a defensible conclusion.

From a cultural standpoint, sports are incredibly slow to adapt these types of advances. Historically, the sports industry has been administered with a jockish perspective that doesn’t need any fancy numbers to know the game.  Baseball is a great example.  We now know that baseball is fairly easy to analyze.  Each play happens in a vacuum and can be physically measured in a number of ways.  It’s all very isolatable and over the past 10-20 years, these stats have become more and more reliable and applied.  That’s not to say their inception was universally beloved.  Hell, they made an entire movie (from a best-selling book) about that very struggle.  People who believe they have a history and a deep understanding of the game based on their experiences in the trenches do not particularly enjoy being told everything they believe is wrong by some Ivy-Leaguer with a spreadsheet.

As we know, hockey is much more fluid.  Literally thousands of tiny plays happen every single shift, by a varying number of players who all may be involved, or not.  Events are chain reactions, which could be set off by a small mistake 20-60 seconds prior.  It could have been with half a different line on the ice when the determining play was actually made.  There is far more frequency to players being in a different area of the ice than they lined up at a faceoff.  Goaltenders influence their defensemen and vice versa.  Forwards either do or don’t back check, which influences goaltender performance.  Everything is dependent on everything else.

And sometimes you catch lightning in a bottle.  Stats guys and gals like to minimize the impact of the emotional component of the game. I get it, it’s harder to quantify than the already hard to quantify events on the ice.  However, this is still a game played by people and they have human responses, both positive and negative.  Historians will still probably be trying to explain the Golden Knights fifty years from now.  Sometimes a group of players just has “it”, whatever “it” may be. This is why I personally believe that a mix of observational analysis will always be a companion to statistical analyses, especially in hockey.

Obviously, the point of advanced statistics is to be able to create a better decision-making process for organizations to build their rosters to contention.  Without the framework they provide for a universal definition of what makes a team successful, it is basically a free for all with all different executives using their own personal compass for what they value in a player. Trends and market inefficiencies are ripe during the early days of this type of implementation and I believe it makes watching offseason maneuvering all the more enjoyable.

So, why are people adverse to accepting the role that this information plays in the game moving forward? That is ultimately what I would like to have all of you help me understand.  One theory is that stats people are pompous, condescending and arrogant. They treat your more casual fan as uneducated and ignorant.  There is probably some truth to that, but I don’t necessarily believe that is the missing link between hockey ideologies. There is obviously just the simple theory that they aren’t accurate/effective.  Sure, there are some holes in the story that these stats tell, but I think we are past the point where we can say they have no merit.

Well, what about general hockey philosophy? I think that falls into how much you value intangibles. If you feel that someone’s ability to deliver a big hit, a timely fight or win a big faceoff is something to be valued, even in the face of evidence to the contrary, you might be willing to outwardly dismiss the stats saying so.  I think less superficially, the philosophy that you would rather sacrifice a little skill for a player with a certain work ethic or willingness to do those unpleasant things on the ice (battle in corners, agitate in front of the net, play with an edge, etc.), can be a driving force in the rejection of the statistical conclusion, as well.

Barring some dedicated ideology that draws a line in the sand about how you want your hockey team to look, the rest of it is emotional.  I honestly don’t think there is any getting around that.  This goes for acceptance of the stats, as well. Some people want to be part of a certain perspective, something they can help to identify their fandom with.  Some people are resisting a change of something that has been a major part of their life since childhood.

For me, I’m of the age when Richter and Leech and Messier were roaming The Garden ice.  They were larger than life.  They made hockey mystic.  In many ways, I wasn’t being naïve, as their legacies have stood the test of time in the hockey world and their feats were just as impressive to everyone else as they were to me.  It is part of the fabric of who I am and what makes me love the game.  I understand people who want the game to always be the way it was when they fell in love with it.

Is that it then? I don’t necessarily think so, I think it varies, probably greatly.  There are definitely people out there who simply do not have the intellectual capacity to digest these stats and understand their significance, so they lash out against them.  There are people for whom, hockey is a simple manifestation of their aggression, and they use that as an outlet to assuage their need to see someone get hit with something.  Call it the gladiator mentality.  “These guys are here for our entertainment and I want to see hits and fights.  If the game loses that, it isn’t worth it”.

There are those who believe that the indicators of success are leading to softer skills being valued and the type of person who plays the game is changing.  Those players with aggressive, violent tendencies are being marginalized for players whose focus is on specific skills, rather than a specific mentality.  They are making the game soft.

There is also an intrusion component. The hockey world tends to be pretty closed off.  For you to be accepted in the world of Hockey Men, you need to have grown up around the game, played your whole life, etc., etc., and this new wave of people who have taken interest in the game, despite having never played or shared their experience, is also resisted by those at the top level, and down to your youth hockey organizations.  “What could you know about it if you haven’t played it”?

Additionally, there are political undertones. In my experience, those with a more progressive political bent are generally more willing to embrace this type of change, as they do with most societal issues.  People that are more conservative, tend to resist the need to fix what isn’t broken.  You can debate all day whether the thing actually is broken, but the type of mindset that guides your political leanings tends to do the same with sports.

With all that said, the question I pose to you, our community, is this: if you don’t believe these stats have value, what type of conversation/perspective/message would make you more willing to have an honest discussion about acceptance of, if not these current set of stats, but overall mathematical application of information to hockey?

On the other side of the coin, for those of you who believe in the merits of advanced statistics, how do you think the message could be better delivered? Short of dragging non-believers by the scruffs of their necks to enlightenment, how can we help those who reject their merits feel more comfortable being open minded?

Now, there is a high probability that this little experiment here will be a complete disaster, given the subject matter.  But, I believe in you.  I think we can have an adult conversation about the path forward to a better understood game.  If not, well, enjoy the weekend, everyone!

"Discussing advanced statistics", 3 out of 5 based on 5 ratings.

45 thoughts on “Discussing advanced statistics

  • Jun 1, 2018 at 8:44 am

    First off, this is an interesting question. I’ve been following hockey and the rangers since ’88, reading this blog for a while now, but never commenting.

    Simple answer, how is this improving my enjoyment and appreciation of the game I love? For all of its depth, your post doesn’t even ask that basic question.

    Furthermore, you want to consider cultural and societal factors. Dude, this is a game. About those factors… you might not realize it, but you are treating these advanced stats as some kind of truth that certain less progressive people are too backward or slow to recognize. How pompous. The language you are using to discuss advanced stats is eerily similar to a cnbc discussion on how the democrats can win back Trump supporters. I’m not trying to be hostile, just calling it how I see it.

    Speaking of what I see, I see highly skilled, fast, big, rough hockey players succeeding in the playoffs, and putting on a very entertaining show. This year’s rangers looked like they weren’t playing the same sport as this.

    Do advanced stats support this, refute it, predict it… I don’t know or care that much and that is ok. I have no problem with them. I enjoy using them in analysis and reading articles that incorporate them. But please, don’t treat stats this way, as tenets to be believed, or socio-cultural tribal signals. They are stats. This is a game.

    • Jun 1, 2018 at 9:51 am

      Thanks for your thoughts, Benny. To your first point, whether or not the stats add to your enjoyment of the game isn’t really the issue. People who don’t feel that they add anything to their enjoyment tend to simply ignore them, at least in my experience. I’m trying to get at people who either believe very strongly in them or outwardly reject them.

      While I do understand that it is just a game and ultimately where you fall on this issue doesn’t matter in the grand scheme of things, that’s kind of everything in sports. People spend time during their day to call in and yell at radio personalities, spend millions of dollars a year on merchandise and tickets, engender relationships with their children and friends, all dependent the game. It’s an important societal construct, whether or not the individual results have an impact on anything outside the sport.

      I think your own perception may be clouding your reading of the article. I believe these stats have academic and intellectual integrity. I said multiple times that they are not “truths” as you put it, but information to be analyzed and assessed. There is a large group of fans that want to actively bury this type of analysis and seek to discredit it for the sake of continuing a traditional model of looking at the game. This is something worth discussing.

      Additionally, my observation of this phenomena has tended to split along socio-political lines. Of course there are exceptions to every rule and I’m sure there are hard left liberals who think these stats are trash and hard right conservatives who believe they are the way forward. My observation could also be biased by small samples sizes and the natural inclination of the individuals to make their voices heard on the issue. The point is that these are factors worth discussing if we want to have productive, civil conversations about relatively mundane sports concepts without tearing each others heads off over basic point of view.

    • Jun 1, 2018 at 1:25 pm

      This was exactly how I felt after just always reading through here but never commenting in previous yeara. I use to go in on the articles when I first started to comment on here. I also thought that the writers were usually pretty bias on who’s going and staying etc but always a bias point of view.

      Caps finally get to a Cup final let alone a ECF and now the narrative is back that the Rangers lack in size toughness. I think that’s a lame approach to take a dig at a team that has killed some pretty physical past Caps teams and skilled Pitts teams. Jets finally made it to the playoffs let alone a WCF and I’m still not impressed by them. When the NYR, Pitts, and MTL are ready to square up and contend than a big bad bruising team like the Caps won’t see pass a 1st or 2nd round and that’s fact. LA might be a bad example because they were truly a skilled, big and fast team to go against but any prime Hawks or Detroit team would have dropped Las Vegas or Winnipeg. Bottom line is that the teams that may not win all but tend to normaly go deep into the playoffs weren’t ready to get in on the fun this year. The top winning teams had either a cup hang over or weren’t ready to compete.

  • Jun 1, 2018 at 8:56 am

    IMO – many of the hockey stats cannot tell the whole story because the game is so fluid. A goalie’s save percentage for example. Depending on the defensive style, it could be high with low percentage shots, but that does not make him a great goalie.

    Stats like +/- for defenseman is another useless stat, as it does not take into account the competition. You also cannot just supplement a mediocre team with a forward with a high Corsi on a low scoring team to make them better. That player needs to blend well with his teammates or get tagged as selfish.

    I do agree there is a place for advanced stats, but IMO, I would like to see forward line stats, defensive pair stats, with some way of taking into consideration the level of competition. For example, Forward Line 1 needs to excel with good offensive stats, but are they great against Defensive Pair 3 and subpar against Defensive Pair 1?

    In other words, because hockey is a team game, and I want to see grouped stats that better reflect the numbers of the “teams within the team”.

    • Jun 1, 2018 at 9:59 am

      I could not disagree more.

      • Jun 1, 2018 at 2:26 pm

        No shock there

        • Jun 1, 2018 at 4:48 pm

          I think your foundation of thought is flawed. Fancy stats, corsi, +/-…. none of it means a thing if you can’t play a man’s game.

    • Jun 1, 2018 at 10:07 pm

      Well put, Sal. Another fun example is a team on the PP that records no “shots on goal”, yet spend the entire time on the O zone peppering the other teams shin pads, ringing shots of the post and deflections off their own that just go wide and just have the fans at the edge of their seat versus a team with just one SOG coming from a missed pass (dump in), and spend less than 20 seconds in the O zone. Theres no stat for PP puck possession but it sure would tell the story better than SOG. Would drive me nuts when the announcers would claim ” and the Rangers had no shots on goal in that PP” meanwhile Devs players would be sprawled on the ice and on the bench exhausted and us fans applauding the NYR effort.

  • Jun 1, 2018 at 9:10 am

    Is there a point in there somewhere?

    Corsi (or whatever they want to call it now) isn’t predictive of anything beyond shooting, let alone winning(a .3-.35 correlation will do that)

    I’m familiar with statistics and politically to the left of Marx, but most stats packages are just reactionary constructs. Like Burke said, it’s support, not illumination.
    This is a function of how the stats are compiled. GIGO. Until visual tracking becomes commonplace and automated, it’s picking up pennies in front of a steamroller.

    All this gets you what? No edge in the draft, not much of an edge in advance scouting.

    The new WAR star that Manny Ek rolled out is another great example. He’s a monster in that stat, yet every time I see him he’s a train wreck in his own end. Who’s right, the guy with a bunch of tenuous stats or my lyin eyes?

    • Jun 1, 2018 at 9:30 am

      The point was to have the conversation in a civilized manner. Twitter isn’t really conducive for it.

  • Jun 1, 2018 at 9:28 am

    I went thru 3 phases of stats: 1) What is this? 2) This is great, I’m going to use them all the time. 3) Hmm, there’s more to it, I’m backing off a bit.

    I think we have a pretty good grasp on forward analysis. We get about 80% of the analysis right. There are always outliers, like stats darlings that for some reason can’t stick on a club. Getting that extra 20% is incredibly difficult and we lack the tools to get there, IMO.

    Defense is a whole other story. We, as a whole, won’t be able to get a true picture of what’s going on until we get puck tracking and, this is important, tie the play to the system being deployed. Dan Girardi is a great example, and sucks far less (he’s still not that good, but he’s better than he was with AV) in TB because it’s a less aggressive system. Less skating intensive, more zone than man.

    Until we fully incorporate systems analysis into our stats, we will have a gap. There are very few people who are well versed in systems, and even fewer that have the wherewithal to combine the two into real time analysis. That is the major gap we see today, and why we see stats darlings like Clendening, Hunwick, Franson, etc never sticking on a team. There is something we are missing (and to be frank, from what I understand a lot of it is pivoting and skating edge work).

    As for goaltending – well that’s just voodoo. I don’t go near it, other than making subjective analysis about the number of high quality chances a team gives up.

  • Jun 1, 2018 at 9:36 am

    IMO, advanced stats obviously serve a purpose and have utility in the current NHL. To dismiss them completely would be foolish. However, the push back I think you’re referring to comes from the perception, that I share as well, of advanced stat supporters relying solely on their lab tops and shunning any form of emotional elements. As you said, hockey is more of a fluid, “emotional” game. Numbers don’t tell the whole story. Many bloggers have the perception of skeptics as dinosaurs or typical “old school” hockey guys who refuse to change. It’s the rigidness of the analytical population that angers me. For instance, Ryan Reaves. If he were a Ranger, every blogger would be screaming from the highest mountain that he would be a waste of a roster spot. His possession numbers aren’t good, he’s not skilled and doesn’t fit in today’s game. However, the success of Vegas’ fourth line has proved otherwise and he’s been a main component. I think again, the conflict arises when the analytical community completely disregards or undervalues a player that plays with emotion, grit, determination or God forbid…fights. Many are already worried that Quinn wants to be tougher to play against. He talks of being physical and the knee jerk reaction are the articles about this philosophy. But, look at the teams in the Finals. Big, Heavy, Physical, Fast, Skilled teams. Winnipeg as well. To get to the point, advances stats must be married to hockey knowledge and the eye test. One without the other won’t work. Unfortunately many feel the “eye test” is outdated.

    P.S. I apologize to Reenavipul for throwing God in there. No such thing in a good Communal Society.

    • Jun 1, 2018 at 10:00 am

      Thanks for your thoughts, VLo. I think you make a good point that both sides of the discussion have to contend with. History played out this way in baseball, as well. Stats were introduced, they were rejected by those with a reason or predisposition to reject them, war was waged, and ultimately the stats guys came out ahead. This final act may not play out the same in hockey, as they are very different games, but I don’t think either side really had a desire to understand each other.

      At the end of the day, baseball still has scouts that are important to the operations of professional franchises and stats did not completely take over. Their more wide spread adoption has taken the fundamental debate away of whether they are useful, and allowed a more productive dialog of how to use them. That’s where hockey, in my opinion, needs to get to.

      • Jun 1, 2018 at 11:33 am


      • Jun 1, 2018 at 12:02 pm

        Baseball is much easier to analyze given baseball can be broken down to for the most part a one on one battle between the hitter and pitcher. Yes the ability of the fielder comes in to play to a small degree, but the the play and performance can be isolated. In Hockey the stats are interesting but there are too many other factors going on all over the ice to enable us to put real stock in these stats. For example a player like Names(or JT Miller for that matter) with TB may have great looking numbers but what does that really tell you about his play.
        How much of his stats is he generating ? or are they just a product of world class line mates?

  • Jun 1, 2018 at 10:01 am

    The application of advanced stats is nice, but at the end of the day, this is a man’s game. If you can’t play like a man, you have no business being on the ice. Corsi is fun and all, but that stuff goes out the window as soon as the first punch is thrown.

    • Jun 1, 2018 at 11:45 am

      Yes, this is why teams like the Flyers are so good in today’s game. They can send out Gudas and Simmonds at the SAME TIME!!! Who could possibly counter that level of toughness?!?

      Wait. That’s not actually the case, at all. Hmm. Odd.

      Guys need to play tough, because it’s a tough game. Nothing involving skill goes out the window as soon as the first punch is thrown, however. Nobody makes it to the NHL and/or lasts for any respectable amount of time if they don’t have the requisite level of toughness. Willing pugilism and a tough style of play are not at all mutually exclusive. Stats help you gauge skill, and skill wins hockey games.

      • Jun 1, 2018 at 1:43 pm

        That can be said for every contact sport at any level. As a senior who had the badges for all team defense in football, hockey and lacrosse I looked liked some beast to the crowd from the outside looking in. Those freshman and sophmore boys on my hockey and lacrosse teams knew the real me once they share the same bench. I spot picked everyting, don’t hit just to hit and don’t waste energy on someone who is 6× bigger than me. Be a hybrid of everything and make sure you have moves and skills do every situation that there is. That’s when you can look a lot tougher or skilled than you really are. Sean Avery was key at doing this in hockey imo. He’ll throw down with medium size guys and if they’re that big than he does two things. Throw a clean blind side hit or made sure that he got the first 2 punches off quickly. Cally is almost like the opposite, he’ll put his body on the line to throw a check just like a idiot goon who only knows how to fight. Eventually you’ll break down alot quicker oppose to the guy who knows when and where to perform that kind of gameplan. This is why someone like Glass isn’t in the league anymore and Simmonds still is. He is a natural skill set to gamble on if he wants to continue to take punches to the head. Dorsett will never play hockey again while D Moore could probably stick around for another 2-3 seasons because Dorsett was reckless type of player who didn’t choose wisely on his physical contact.

        • Jun 1, 2018 at 4:46 pm

          Hey Mint… not one single person here cares at all about how many badges for all team defense you had. Hardooooooooooo

  • Jun 1, 2018 at 11:05 am

    Where stats could be helpful is say in faceoffs, was it a key win? defensive zone draw with the goalie pulled, those kind of situations. Other than that it’s just a tool that answers only a part of a story and never the whole story. Those who punch numbers into the computer and then think everyone who doesn’t agree with them, I just laugh off and move on from their thoughts.

    To be reliant on statistics to determine a system of play is also foolish. Tops you gain 30% of information from analytics.

    Minimize emotion? Seriously? that’s idiotic as motivation and emotion is a huge factor in sports.

    Every goal has it’s own individual story so analytics can never give you what the story was. What it can give you is situation, position and player from where the goal was scored thus you have vital information on mechanics only. So patterns are what it offers

    Yup analytics is part of the solution not even achieving 50% of the information in a team sport but still vital to use as a tool in your arsenal for individual aspects of the game.

  • Jun 1, 2018 at 11:29 am

    I’ve used statistics pretty much throughout my working life, so I understand them pretty well. In my experience, they’re extremely useful for pointing out potential problems or opportunities, but they’re not the be-all and end-all of any discussion. First, most of them can show a correlation between factors, but finding causation is much more difficult. Example, high possession statistics correlate with winning records, but does possession cause winning, or do winning teams just possess the puck more? At the end of the day, the game is about two things, in my mind. First, is it entertaining? If both teams compete hard, play well, and the officiating is consistent, the game is enjoyable to watch. Statistics may give you an idea of whether or not this is the type of game you’re going to see. Second, of course, is who won? the better team, hopefully, but, even though hockey is the epitome of a team sport, there are those games where one player just dominates, and tilts the ice in his team’s direction, and in this case, the stats going into the game don’t mean squat. The same thing happens on those occasions where the players of one team seem to be reading each other’s minds, and their plays just click, or, sadly, when they appear to be totally oblivious to one another. Then of course, there’s the completely unpredictable way in which teams mesh, either into a unified group that can outperform anything the stats tell you, or into an unruly crowd of guys who just happen to be wearing the same uniforms. Stats can’t help predict how that’s going to work out, and, yet, as this year’s finals show, it may be the most important factor of all. In hockey, every player’s performance is influenced by every other player on the ice, on both teams. Individual stats are much less important than the overall team stats.

  • Jun 1, 2018 at 11:36 am

    My take is really quite simple: more information is better, therefore I support the inclusion of advanced stats in evaluating NHL hockey.

    Now, if you cherry pick any one stat – traditional counting stats or advanced stats alike – as the king of all stats that determine anything/everything on paper, IMO that’s a terrible way to approach it. Obviously there is no such infallible stat. It’s a complicated game, and even if you played at top levels and have an intimate understanding of systems, it is impossible to capture and process all the variables in real time. Even the more complicated stat packages are not guaranteed to paint any perfect picture. So, stats of all manner help us bridge those gaps, but even then, there is still so much going on in a hockey game that we haven’t figured out a way to express on paper. Emotional/momentum effects, as one big example, simply cannot be measured that way. Did the fight help or hurt? We can have an opinion on that, but there’s no stat that will ever inconclusively decide it for us. If a team is getting beat and there’s a fight their guy wins, and they continue to get beat…well, that doesn’t technically mean they didn’t get an adrenaline boost. It means they didn’t have the skill on that night to match, and/or had some bad luck, regardless.

    On the flip side, certain stats/stat packages are clearly better than others. It doesn’t take much of a study to find that individual +/- is pretty bad for gauging a player’s actual worth – bad players can easily be + and good players can easily have a – as well – so if someone wants to die on that hill, not gonna lie…I probably won’t have much respect for a case based on that one stat alone. You look at HERO charts however, again just an example, well, those don’t generally lie a whole awful lot, and they pull in various factors. You don’t often see players who look terrible with glowing HERO charts, and you don’t often see players who look great with ugly HERO charts. Perfect? Nope. Better? IMO, very clearly so.

    So, for me…it’s fairly straightforward. The eye test alone is subject to being tainted by individual bias, and the pace of the game itself. Eye test plus traditional counting stats is a step up from there. Eye test plus traditional counting stats plus advanced stats is another step up. Keep the information coming, far as I’m concerned – can’t hurt.

    • Jun 1, 2018 at 12:47 pm

      Comparing HERO charts and +/- is unfair. The former is very sophisticated, the best advanced metrics can offer. The latter is sloppy, battered by empty net goals and I think more subject to strength of opposition.

      • Jun 1, 2018 at 1:28 pm

        I concur. Nonetheless, there are folks who would/do toss out a HERO chart in favor of +/-.

        • Jun 1, 2018 at 5:40 pm

          I will just speak for myself +/- is my go to, but I have no idea what a HERO chart is nor will I ever unless it is posted on a article I read. I just don’t care that much to go that deep into it.

          But I will explain why my go to is +/-….cause you can take a player you have watched for a long time like Stepan….with the Rangers…. he was always a plus 15 for his career here…and then he goes to Arizona …the average every game player is putting up -20 something….he posts a -7. To me that makes him alot more valuable. Shows me night in and night out over the course of 80 plus games his line wins more battles then loses in comparison to the rest of the team. A guy who can put up a -7 on a team surrounded with -30s is a pretty good player. This is counting in the systems …the line matchups… all the power play goals guys like ovi get.

          Ovi doesn’t get a plus for scoring on the powerplay. But Stepan does get a plus for scoring shorthanded which is alot harder to do. That is how a guy scores 51 goals but is a negative -35. You are all offense and absolutely no defense and I feel for judging a forward….. defense is first and foremost. And I know everyone wants the Ovi’s but do you really want them?

          This year he plays defense more then he had and his plus minus is +3 with 87 points for the regular season and a plus 5 for the postseason with 24 points. 31 points from the regular season came on the power play …56 even strength points. and out of that number he is a plus 3? Pretty pathetic.

          Maybe I got this wrong but to me that tells me out of the 56 times he got a point 53 times he gave it right back at even strength. We all know he is not playing shorthanded and we all know he is never out on the ice when the team is up a goal in the final minutes. Obviously i am generalizing but he is no pk specialist and he is the last guy you want out there trying to hold onto a lead.

          And to me those are the guys that win cups. It is the guy that battles in the final minutes …it is the guy you throw out there when the other team is dominating your team and you need to stop the momentum.

          Do certain players exploit that stat? Of course but in my eyes it is all relative to the team. If everyone is putting up plus 15’s for the season and player X did the same …not really telling me much about player X. He is just another guy on the line. But I guarantee there is a player on that team that is driving the offense and consistently putting up a plus much higher then the rest. To me those are the value finds with the stat.

          You can look at Stralman for this exact reason. Or Giradi for that matter. Tampa is a perfect example …the middle of the pack players are there but Gourdes, Hedman, Stralman are way above the rest.

          And the stat is only good to me at the end of the season. Cause like any other sport guys can get hot ….exploit a weakness or be on a hot line….but over the long haul that all goes back to normal. It is similar in my mind to shooting percentage. Guys get hot …but over the length of a 82 game season it comes down to normal.

          The great one was a career +/- of +520 and bobby orr was a career +582. Gretzky scored 2000 more points then Orr did. And as someone else noted the other day….How many Cups did the Great one win without Messier? 0 …

          Scoring is great but IMO defense wins championships.

          • Jun 1, 2018 at 6:15 pm

            I actually like +/-, though it needs some correcting.

            BUT there is large sample size error because there are just too few goals scored. Over a consecutive six year period, Ovechkin went +45, +24, -8, +2, -35, +10. The -35 is simply an outlier.

  • Jun 1, 2018 at 12:07 pm

    Agree stats can’t hurt but the eye test is more important. Stats are impacted by factors largely outside of any one players control. Yes an elite player will likely have great stats, but average players playing with that player also would have great stats.

    • Jun 1, 2018 at 2:20 pm

      The eye test, more often than not, is filled with personal bias. We are all guilty of it, it’s human nature. We nitpick the issues if we don’t like a player, the same way we ignore those issues with players we like. We then make excuses one way or another.

      • Jun 2, 2018 at 2:27 am

        Only has bias if you let it.

        I thought Gilmour sucked his 1st season, when he looked better this past season I said so.

        I’m all over Boqvist, but when I 1st saw him he was only a name. Every time I saw him, it’s not like he changed except when he played against men, when it quickly fell apart.

        • Jun 2, 2018 at 10:32 am

          A word of warning here. Years ago, USC had an All-American running back named O.J. Simpson. He was incredible, but as a senior he faced a Notre Dame that pretty much stopped him cold. They had linemen that could move and he couldn’t just run around them. The guy couldn’t break a tackle to save his life. I predicted his failure as a pro and gloated for the first year when “proved” right.

          Sometimes the big fish in a small pond doesn’t have skills because he doesn’t need them and when he needs to develop them, he does.

  • Jun 1, 2018 at 12:08 pm

    Hey it’s finally JUNE big the month is here 3 weeks to the draft!

  • Jun 1, 2018 at 1:48 pm

    It is ironic that this discussion does actually approximate societal discussions. Like in those discussions, the best outcome is not to swing too far to either side. Reliance solely on the eye test is truly to be a Neanderthal (and other contributors who are familiar with my comments on this site know how I generally hate that characterization) but reliance solely on analytics is equally foolish. I like the comment above of the fellow who works with statistics: they can often provide good bases for correlation, but often fall short in terms of causation. In essence, useful as one of several tools. Because hockey is a physical game and because hockey allows for few genuine one on one competitions, I believe stats are particularly marginal. Not useless, not to be ignored, but only one piece of a puzzle. I do think the Ryan Reeves example is a good one (there goes the Neanderthal in me)…..

  • Jun 1, 2018 at 2:08 pm

    Of course, there is always a resistance to mathematical concepts and to the idea that others know things. That is sad, but won’t change.

    The real problem though Justin is that advanced stats are not there yet. You say baseball stats have come a long way in the last 10-20 years. Well, guess what? Hockey is nowhere near where baseball was in 1985. And if you look how long it took for Billy Beane and Theo Epstein to appear after stats were spot on, you can see how long it will take.

    So much for what you are up against when you get it right.

    Now, I am an opponent of advanced stats as they currently exist. I am not the opponent you typically imagine. I computed baseball batting averages before I was ten, was and am a strong supporter of Bill James and Sabrmetrics, have a Ph D in mathematics, and have taught statistics.

    Baseball is easy. A game has 300-400 events, all of which can be easily recorded and analyzed. A hockey game has either about five events (if you just count the goals which is all that really matters) or tens of thousands (if you count everything). Using the first creates huge sample size error even for something as simple as GAA. If you try to adjust for teammates and opponents, the error is outrageous. On the other hand, counting everything is impossible.

    A reasonable approach would seem to be to try to find a middle ground and that’s what advanced stats are designed to do. The result of course is going to be more accurate measures, but measures of what exactly. How much can you learn by counting things that don’t affect the game directly and ignoring those things that don’t?

    First of all, I ask what is the evidence that advanced stats are valuable. Yes, puck possession correlates with talent and talent correlates with winning, but that doesn’t mean that you measure talent by puck possession. There are other aspects of the game and is there even a shred of evidence that an organization which is guided by advanced stats is more successful than one which is not?

    I have actually done a few limited studies. I analyzed goals scored by team in the 2016-2017 season. Of course, scoring correlates with winning and so this is an important stat. Also obviously, goals scored = number of shots x shooting percentage. I don’t have the numbers in front of me, but the number of shots per team fell into a very narrow range and the shooting percentage varied much more wildly. So, while shots correlated with scoring, it was not that strong a correlation.

    Another study a few years back, I looked at 5 v 5 Corsi over a whole season and compared to WL records. According to Corsi, Toronto, the worst team in the league, was better than Washington, the best. Toronto was among the top sixteen teams and the Caps were not. In fact, picking the playoff teams by Corsi was only a shade more accurate than flipping a coin.

    On goalie stats, using a measure of high difficulty shots during the 2015-2016 season IIRC, Henrik Lundqvist had a phenomenal year. He saved so many goals that I think it was fair to say that he was a better tender than Wayne Gretzky was a center. This despite the fact that he (A) got no Vezina consideration and (B) failed to outperform Raanta, who played behind the same defense. Another star by that metric that year was Steve Mason, whom you placed in the top ten partly as a result (the only inclusion in your top ten that I criticized). The glory of advanced stats is supposed to be that they are a better predictor of next year’s traditional stats than traditional stats are – and so these numbers should presage an uptick in traditional stats. That was true in neither case.

    Another point that is important, which I owe to Bill James. All studies are based on the way the game is played. For example, the idea that the number of runs a team will score is roughly number of baserunners times slugging percentage assumes that you bat Mickey Mantle in the 3 or 4 slot. You can improve the number of “expected runs” by batting him leadoff, but there is no evidence that such behavior will actually increase the number of runs. There are no studies for scoring efficiency with convoluted lineups.

    Why is this relevant? Someone pointed me to a study years ago that said that the team Corsi numbers over a 20 game period were a better predictor over the next 20 games of team +/- than using the +/- numbers over that stretch (an effect which diminished with sample size). However, this study was done at a time when no one really cared about Corsi. People didn’t care about possession stats and they were a coincidental indicator of success. When people start focusing on a coincidental indicator and trying to maximize it, it stops being a coincidental indicator.

    This was certainly long. I really don’t know if really good metrics are possible. I don’t know if you can really measure shot quality objectively for example. But I think I know with a fair amount of certainty that Dan Girardi has put together a better career than Keith Yandle and metrics which tell us the reverse is obvious are missing something major.

    • Jun 1, 2018 at 2:24 pm

      One of my sentences is unintelligible (only one you think, some laugh).

      “How much can you learn by counting things that don’t affect the game directly and ignoring those things that don’t?”

      should finish

      and ignoring those things which cannot be counted.

  • Jun 1, 2018 at 2:32 pm

    FYI – Chatter and rumors that Detroit is willing to part with their early first round pick, as they feel they can get a good valued defenseman down some spots. Do you give them your 9th and one of your late first round picks if Wallstrom is still there when Detroit picks?

    • Jun 1, 2018 at 4:00 pm

      Crazy talk

    • Jun 1, 2018 at 4:50 pm

      Nevermind, I was just making that up to see if anyone would bite. Pretty sure they have their eyes on a higher end forward.

    • Jun 1, 2018 at 4:52 pm

      I would be open to that trade. Wahlstrom could be a potential elite guy for us in a couple of years.

  • Jun 1, 2018 at 6:10 pm

    I personally just don’t care that much to get all involved in deep math. If I have to start doing 3 formulas to figure out something I not that committed.

    Are advance stats useful? I personally don’t think so …cause I discredit the source. How many times have I seen bloated checking stats or bloated blocked shot stats…endless. Watch a game and post your personal stats that you see throughout the game and then put them up against the official scorer. I would be utterly humbled if they were even in the same ball park. There is just too much going on for one or 2 guys to write it all down as it happens.

    It is easy to say this team had the puck for more time then the other team but like in soccer that team might not be necessarily be controlling the play or working easier then the opposition. Is it easier to stay and play defense if it is zoned and all the shots are coming from the outside and then counter attack quickly with speed and efficiency? You can’t show hard a team had to work to get a high quality scoring chance. How do you measure that with math? One thing with Torts teams were… we always had energy in the tank for a comeback whether it was the conditioning his team had or whether we wore them out with the rope a dope and used lundquist as ali in this happenstance. Which one is right? How does one show that in mathematical equation? Puck possession is puck possession.

    You can have all the skill in the world but have absolutely no compete level and never make it in the nhl. Like I posted above look at Gretzky and how he did without messier and kurri. I mean I sure he was amazing to watch and could single handedly take over a game but when things got tough which they always do come playoffs he couldn’t win it all. That is where the claude Lemuix and jarri kurris came in…they got those tough goals. Those real nasty nothing pretty smash it down the goalies throat goals. That is where guys like Reaves and Tom wilson come in. The unsung heros of cup teams. Personally believe against LA that was the exact type of player we missed.

    SOrry about the side tangent but stats are relative to the user. I agree with the statistician. It is good for the individual user and if it makes them happier so be it but hockey is way too fast and complex to apply it ..One point I see is does anyone follow european football/soccer….not likely to be one on here but maybe. That is there largest sport by a landslide…. do they have advanced metrics? Cause IMO it is the closest game to hockey just with alot less games a season. Has advanced metrics fully taken hold with that sport or is it still an eye test?

  • Jun 1, 2018 at 7:57 pm

    My take: Yes, we should use the analytics. But Corsi, Fenwick, etc seems very much on the surface to me. Yes, it’s telling us something. But it’s not some secret sauce that is the underlying metric by which all is measured. That may not be what the intent is, but that’s exactly how it comes across on many forums IMO. I’m always left wanting something deeper when I see a bunch of charts on the basic analytics. Ok, there’s his Corsi and there’s the Goals Per 60, but then what? It can seem superficial to me, and then presented like it’s gospel and blasphemous to disagree and it’s the end of the debate on any particular player. There’s just so much more that contributes to hockey and what causes a goal and how wins are generated besides possession stats. I mean there’s a correlation between winning and have a great power play too. And you’re not going to go home and work on your Corsi to end a losing streak.

    I honestly feel like Corsi and other stats more show that a player fits well in a particular system or team. I think it can be relevant when looking at marginal players in particular. The NYR d-men for example is interesting to see when they all had a decent sample size at the end of the year and they were all playing their first NHL games. But at the same time, Neil Pionk stood out and really impressed regardless of his Fenwick. And if the fancy stats don’t prove that out, then I’ll continue to disagree with that argument no matter what the numbers show. Likewise, I figure JT Miller’s Corsi improved when he was traded and Namestnikov’s slipped when he came here (just a hunch).

    Show me some stuff that’s maybe more situational or blends some sort of context into it. Which penalty killers keep the puck out of their zone the most… who has the highest avg in the league of time outside their zone while they’re on the ice? Which teams are giving up the most goals after a shift runs longer than 45 seconds? What is a goalies percentage of goals let in from above the circles? Which d-men are on the ice the most when goals are being scored from 5ft out? That would tell me way more about a defenseman than corsi. The kind of stuff that Steven Valiquette has. He does not dabble in the online analytics much, but he has some real insight in the kind of stats he presents. I find his data way more insightful.

  • Jun 2, 2018 at 12:56 am

    Below will be the possession stats from the last world championships, using their next level player tracking along with the final rankings. What would be great is getting the CF% done the traditional way.

    Puck possession and final ranking.
    1: CZE. 7th
    2:USA 3rd
    3: FIN. 5th
    4: RUS. 6th
    5: SWE. 1st
    6: CAN. 4th
    7: DEN. 10th
    8: NOR. 13th
    9: SVK. 9th
    10: BLR. 15th
    11: GER. 11th
    12: SUI. 2nd
    13: LAT. 8th
    14: AUT. 14th
    15: FRA. 12th
    16: KOR. 16th

    That’s probably an improvement in correlation, but even then there’s wild swings.

  • Jun 2, 2018 at 4:15 am

    SInce my first post got deleted since I edited it for spelling …I will summarize my points ….

    1.) Stats are useless IMO in predicting compete level. Just look at Vegas. Ryan Reeaves, Tom Wilson.
    2.) Advance stats are even more useless cause the source is not credible. They get compromised do to human error more often then not. Just too many things happening all at once for any one human to gather. So unless there are game scorers for every player on the ice it is just to inconsistent to sample. I mean the camera man can’t even follow the puck in half the games.
    3.) IMO stats are great to enhance the game for the fan and if something works for that individual then good for them! But the game is won and loss in the trenches and that Matteau Matteau chant most of us know would never have been predicted.
    4.) The closest game to hockey in my opinion is soccer and I wonder if anyone on here follows it. And if so do or does advanced metrics get applied to it. As americans well atleast most of us we focus on the big 4 but if someone knows please entertain my question.
    5.) Simple point about possession stats I made is just like soccer …if a team has to work harder to control the ball/puck but never can get a solid scoring chance how is that shown in a stat? I mean the Kings were the corsi darlings of the regular season but didn’t even make the playoffs.

  • Jun 2, 2018 at 4:55 am

    I also think most on here think tough to play against translates into having a guy like Colton Orr. Sorry for this but it has been bothering me lately. I think when people say tough to play against I think more towards the lines of ryan callahan in his prime not today’s version. Not a tom wilson. When I hear that phrase it translates to me into compete level. Just how hard is someone willing to play.

    I think last year’s team had absolutely 0 compete level for large vats of time. And it showed. Guys just were not willing to play hard. They showed up but that is about all they did. But how does one measure that in the draft? How does one predict a brendan smith? Or Mika Zibenajad? I mean mika has all the tools to be an amazing center in the NHL but where is his compete level? Where is rick nash’s compete level after his 3rd or 4th concussion? How badly does an individual want to win? Say what you will about McSorely and that horrible outcome but you know what no one will ever question?

    But how does a scout recognize compete level? Is it even a thing they look for? Crosby for lack of a better example is a superstar not just because he has all the talent in the world but because he is willing to compete as hard as his body will physically let him. Mike Sauer could most likely have come back from his concussions if you are seeing what Crosby is doing now. The hits that crosby has taken are not any worse/easier then the one that sauer took to end his career. But some how, some way, some people just find a way to beat the odds.

    There is absolutely no way crosby should be playing but there he is and succeeding no less on the highest level. Winning 2 stanley cups in a row.

    So show me toughness and a willingness to compete. I don’t need you to fight every player in the league to show me you are tough to play against. I need you to compete hard every single shift like that little hob goblin we all call Zuc who nearly lost his brain but came back and stood right in the same place he always did. Dodging shots from MAck who kept trying to kill him! To me he is the player that represents the best of “tough to play against. ” And I don’t care about size or skill or speed…I would take a whole team of players like Zuc and I guarantee we would have a cup hoisted here in no time.

    So IMO trade everyone who isn’t willing to compete…no matter who it is. And get guys that are willing to battle with that guy next to them. And if the coach benches them for sticking up for a teammate well fire his ass that second. No waiting til the end of the season. Kick him to the curb for taking away a player’s heart. Otherwise you get the 2017 New York Rangers!

    • Jun 2, 2018 at 8:32 am

      For this post I gave a thumbs up, right on the mark!!!!!!!!

  • Jun 4, 2018 at 4:42 pm

    Great post – really interesting read. I think you hit the nail on the head with the “condescending” point. To touch on a few different points you made, I tend to lean right politically, have played hockey since I was 3 years old, and have a BS in Accounting and Economics and a Masters Degree in Statistics. So I definitely qualify as being able to understand the statistical aspect of the argument.

    I don’t say these things to (not so) humble brag, just to give perspective on my background which I think is very important to my overall point. My intro to the statistical world of hockey was Rob Vollman’s Statshot (which Rob if you read this ever, was freakin awesome…) but that’s besides the point. As a “old school hockey guy”, I was admittedly intrigued by one of my favorite subjects and hockey being intertwined. I even have my own metrics that I don’t share, and you’ll know why in a minute.

    Despite my love for stats and hockey, I find myself wanting the blog-o-sphere and twitter to stay away. I find some (by no means all) of those in the Rangers internet world who are almost always right, statistically/mathematically speaking, are extremely condescending. Two major exceptions. HockeyStatMiner and Shayna Goldman. They both put it in context that’s easy to digest, for any fan (as long as they actually read what they write). But those who are condescending, unfortunately, are also the loudest to scream “I’M SMART AND I’M RIGHT AND YOU’RE WRONG”… moreso on twitter than their actual blog posts. This is the largest turn off for me, so much to a point I wish it was never discussed sometimes.

    It’s okay to be the smartest person in the room. But you should never say it out loud.

    To be fair, I’d like to say that the other end of the spectrum is just as loud and obnoxious. But I think that the statistics crew should think, as I do, and say, “the math works, and it’s incredibly useful, let them babble.”

    To sum it all up – I think the solution here is to be the bigger person. Let those who don’t want to/cannot understand the math hope for their goons and fights to live forever, and don’t antagonize, just ignore.

Comments are closed.