Scoring Chances – A Brief History and Background

 

 

I won’t be quoting any numbers or arcane stats in the post today. I won’t even be contributing much original material. Mostly I just wanted to call attention to this interesting article by David Staples of the Edmonton Journal on the history of scoring chances and "non-traditional" stats in the NHL.

 

The secret statistics – most notably the individual scoring chances stat – are tracked by coaches on each team. They’re used in games to make between period tactical adjustments and in video training sessions to help players understand what they’re doing right and what they’re doing wrong. They’re also used by NHL general managers to evaluate players and make trades. It’s not so simple to know exactly when these secret stats — or alternative stats as I call them — became widely used in pro hockey, but there’s little doubt that Hall-of-Fame coach Roger Neilson was a key innovator.

Neilson developed a systematic way to keep track of scoring chances and also to keep track of which players helped create those chances and which players made mistakes on chances against.

Neilson created a scoring chances sheet, which was photocopied and passed on for decades to various coaches. It had a line down the middle. On one side, he or an assistant would record each scoring chance for, with a plus mark for only those players involved in the chance. On the other side, they would record chances against, with each player who made a mistake getting a minus mark.

 

One thing I get asked a lot when it comes to "new stats" (misnomer!) is whether NHL teams actually use any of them or not. The truth is, much of this stuff is old hat to many NHL coaches and teams. Corsi was named after the NHL coach that came up with it, for example. Neilson has been tracked scoring chances and grading players according to their chance differentials since the late-70’s. We’re just kind of stumbling on stuff that smart men in the NHL have known about for about three decades.

 

Neilson defined a scoring chance as a dangerous shot from a zone in front of the net. For his boundary, he drew a line back from the side of the goalie crease to the faceoff dot, then straight back to the blueline. But any shot that came from the point in this scoring chance zone either had to be screened or tipped to count as a scoring chance.

 

This is fairly in tune with the chance definition we use here at FlamesNation use when counting chances. I was never overly anxious about it, but it’s nice to know we’re in line with Neilson here. Staples notes that Neilson sometimes included "chances at chances" (for example – a two-on-one that doesn’t result in a shot on goal), but I’m comfortable in excluding them nonetheless.

 

“You see value in players other people don’t see. Because some people only see goals and assists and flashy plays. They don’t understand what a guy can do for you in a certain role. So it does give you a little clearer picture, for the coaches and the manager of the team to know what you’ve got.

 

That’s Staples quoting Dave King, former Flames coach and a Neilson disciple. The rationale for tracking scoring chances noted by King here meshes with a lot of what we talk about in terms of SC’s and new metrics: finding the value (or fault) in players that can be missed by perception and observation alone.

 

Things are so advanced now, King says, that all NHL teams have a video coordinator and coaches assigned to video analysis. The chances stat is used in making major decisions. “Sometimes these stats factor in for decisions on players. There may be a player who has reasonably good numbers in terms of goals and assists, but when you really look at his game and you can see he’s a player who is a real negative chances guy, that sometimes weighs into a decision on making a trade with another team.”

 

Emphasis added by me. King doesn’t use the terms like variance, luck or the percentages here, but it’s clear that’s what he’s talking about (even if he doesn’t use the mathy words). If anyone wonders we bother with PDO (SH% + SV%) or why I looked at goals% versus scoring chance% for the Flames recently…this is it.

NHL teams keep this data secret, King says, because they believe they have something of value that their competitors do not have. “We guard the information carefully.”

Too bad we’re giving it away for free around here*…

*(if we don’t do scoring chances next year, you can be reasonably sure we’ve been bribed by the Flames to shut-up. Or threatened with physical violence).

  • The key distinction between the scoring chance methodology in widespread use around the blogosphere and what David Staples is doing at Cult of Hockey is that the one is an on-ice metric (the qualivalent of +/-) and the other attempts to attribute individual involvement. The Neilson method seems to favour this latter approach:

    “On one side, he or an assistant would record each scoring chance for, with a plus mark for only those players involved in the chance. On the other side, they would record chances against, with each player who made a mistake getting a minus mark.”

    Or so I read it.

  • Derek Zoma’s earlier post is, indeed, a good one. That said, it’s only part of the story of Neilson’s central insight into rating players.

    Zona focuses on just one aspect of the Neilson method when it comes to scoring chances, the recording of the actual change itself, and what defines a scoring chance.

    Neilson did all that but he took his system one huge and crucial step further.

    According to Dave King who learned the system directly from Neilson — Neilson only gave plus marks to those players who contributed to the scoring chance, and to no one else. And Neilson only gave minus marks to those players made some mistake on the change against, and no one else.

    Of course, it’s one thing for an NHL coach or asst. coach to make this kind of assessment of who isn’t and is involved in chances, and it’s another for fans to try to do it.

    Not all fans can do this kind of work, but I’d argue that hardcore fans can do it. I’d add that from reading the work of many of the stats guys, such as Derek Zona, Dennis King and you, Kent, my bet is that you do have the hockey knowledge to fairly and accurately do this kind of work in assessing scoring chances the way that Roger Neilson did it.

    I’d love to see numbers on the Flames, and it would be useful to have more bloggers using Neilson numbers to rate players, both to find out your results and to see if our results are similar or different.

  • According to Dave King who learned the system directly from Neilson — Neilson only gave plus marks to those players who contributed to the scoring chance, and to no one else. And Neilson only gave minus marks to those players made some mistake on the change against, and no one else.

    I think there’s value there for sure David, but I think there’s also value to capturing total on-ice events because some actions may improve general possession without directly effecting a chance for or against.

    There’s the opposite problem of capturing “false positives” with the wider net, of course, but with higher total events you can winnow out aberrations with a large enough sample size.

    My ideal scoring chance project would have the total events counted as they are by the generalists now as well as the attributional counts (“The Neilson numbers”) as a supplement. Of course, I’d also like multiple, independent observer correlations for chance counts and the ability to “grade” chances by how close they are to the net, but…we only have so much time, hey?

  • King said that when there’s an argument over what is or isn’t a scoring chance, all the Phoenix coaches will weigh in.

    This kind of discussions is very useful when you’re weighing such things, and also assigning blame and credit on each chance.

    On goals for and against the Oilers this year, Bruce McCurdy and I would often discuss/debate each goal, and that was very helpful in making better decisions. . . .

  • I think there’s value there for sure David, but I think there’s also value to capturing total on-ice events because some actions may improve general possession without directly effecting a chance for or against.

    I agree with this. I think there is value to both methods, and that they provide complementary, rather than conflicting, information for the most part.

  • There is a huge problem to this system:

    How does the ‘tough, gritty veteran’ get any credit for his ‘heart’, ‘leadership’ and ‘character’ directly attributing to other players goals when he is sitting on the bench?

  • Well, I really don’t understand why we measure any of this nonsense at all. I have been watching Flames’ broadcasts on TSN, Sportsnet and CBC all season, and from their expert advice it is my understanding we should only track the following statistics:

    1. The number of times a player moves his feet in a shift. Apparently no matter what else he does on the ice, as long as the feet are moving good things will happen. I have yet to hear a coherent explanation of how this happens but I understand it has something to do with friction on the ice and magical fairy dust.

    2. Surveys of the players’ confidence. This particular facet of the game appears so crucial to athletic performance that we should actually stop play every few seconds to ask players how confident the feel and track the results in real time.Could also be replaced in a pinch by measures of how much a player “wants” it.

    • I think you’ve missed a couple of crucial elements, Tach:

      3. Players have to keep it simple. If a guy is thinking too much or trying to get creative, bad things are probably going to happen. Or at least, less good things.

      4. Pucks to the net is the primary PP strategy. Formations and personnel are far less important than getting pucks to the net.

  • Well, I really don’t understand why we measure any of this nonsense at all. I have been watching Flames’ broadcasts on TSN, Sportsnet and CBC all season, and from their expert advice it is my understanding we should only track the following statistics:

    1. The number of times a player moves his feet in a shift. Apparently no matter what else he does on the ice, as long as the feet are moving good things will happen. I have yet to hear a coherent explanation of how this happens but I understand it has something to do with friction on the ice and magical fairy dust.

    2. Surveys of the players’ confidence. This particular facet of the game appears so crucial to athletic performance that we should actually stop play every few seconds to ask players how confident the feel and track the results in real time.Could also be replaced in a pinch by measures of how much a player “wants” it.

  • If you need to justify scoring chances to me, so you can trade a player that produces lots of points. Then by all means take a grunt player from my roster. I hope all the coaches believe in-reference to what Dave King said “You see value in players other people don’t see. Because some people only see goals and assists and flashy plays. They don’t understand what a guy can do for you in a certain role. So it does give you a little clearer picture, for the coaches and the manager of the team to know what you’ve got.”
    Yah, you’ve got nothing!!!

  • How does the ‘tough, gritty veteran’ get any credit for his ‘heart’, ‘leadership’ and ‘character’ directly attributing to other players goals when he is sitting on the bench?

    Domebeers, my good man, you are missing what is right in front of your eyes. *This* is how we assess their value.

    Take for example a veteran who seems to bleed a lot of scoring chances and take bad penalties — let’s call him “E.M.”.

    If EM’s scoring chance differential is 50 worse than what’s expected from a player in his position, then his intangibles must be worth at least +50 to the rest of the players on his team. Otherwise, EM would not have an NHL job. QED.