MICRO – A community tracking project that needs your help

Mmmm, stats.
Mmmm, stats.

Throughout this painful season, one of the comments that has stuck with me is when Alain Vigneault referenced that the Rangers look good when you see the advanced metrics. That was a false statement based on what’s publicly available –teams track their own stats, but it’s proprietary and we have no idea what they track or how effective it is– and all of the analytics community was pretty confused by this statement. The Rangers felt the pain of being a bad possession team once the playoffs rolled around as they were absolutely trumped by a possession goliath in Pittsburgh.

Even though the season has felt like a fluke in the standings, there must be someway to explain their stellar season heading into the January, PDO is certainly one of the prevalent reasons. The current measurements of shot quality can’t currently explain the season or where AV’s stats were coming from (Jim Sullivann, head of the analytics department, maybe?). It is almost impossible for a team to win time of possession and be so inefficient to not out shoot the other team on a regular basis.

This post isn’t entirely about the Rangers, in fact they are merely a case study. The goal of this post is to make you think about the fundamental thinking of hockey analytics. If you are like me, you don’t settle for Rolle’s Theorem. You want to connect the points to get to the Mean Value Theorem.

There is so much in the game of hockey that is not quantified. We put it off and focus on our shot statistics because it is easier to scrape data automatically than to track data manually. With your help MICRO will help bridge that gap. Trackers will record numerous events that are not supplied by the NHL, including shot attempts, in order to help with venue adjustments and ghost shots. If you help track these events, you and many others can use these numbers to create better models and understand the game better.

Here is a list of what will be tracked:

Shot Attempts:

  • Who shot the puck?
  • From where?
  • What type of shot?
  • What happened?
  • Was there a pass proceeding it?

Zone Entries/Exits:

  • Where did it happen?
  • Who performed it?
  • Was there a pass before it?
  • What type was it?

Challenges (Hits/Takeaways/Recoveries)

  • What type was it?
  • Where did it happen?
  • Who did it?

All events will answer the time, man strength, and the new zone strength. These events should seem familiar to you if you have followed hockey analytics, but is not usually available. What was Sidney Crosby’s controlled entry percentage this year? MICRO is going to try to supply these stats, but we need your help. Whether you can only track a game a month, a game a week, or multiple games a week, we need your help. If you are interested in tracking email:

Show More
  • From what I saw we won the possession battle with Pitt. I get that going in we were bad possession team, but all possession stats went Rangers way.

    Bad possession had zero to do with being trumped by Pens at least in the matrix used to determine possession.

  • Wasn’t there an article on this site titled “the Rangers trolled us during the regular season”. Didn’t that article pretty much say there were a good possession team in the playoffs?

  • OK guys. I am the co leader in the project. We were specifically talking about how they got there in the first place. The AV comment came before the all star break when the Rangers were second in the division. Yes they were like 51% possession in the playoffs which is +3 relative to regular season.

  • These stats don’t work in Hockey there are two many variables. Only sport that stats work in is baseball.

      • Stats don’t even always work in baseball, either – because it often comes down to HOW they are used. For example, a baseball pitcher can have their WHIP be sub-1 but their ERA over 5 (usually in short samples of a couple months or less, but still). Those who just look at ERA, well, they call that a bad pitcher…when the WHIP tells us, no, he’s just an unlucky pitcher and an in-proportionate amount of batters that he allows to reach base are somehow scoring at an unsustainable rate. Hockey right now has a bunch of mainstream stats that are more like ERA…we need more hockey stats that are more like WHIP. We can do the same sort of supplemental things in hockey, if only more things were being tracked. It may not lead to some sort of Holy Grail single-number statistic in hockey the way baseball has WAR for example…but it still can’t be dismissed as entirely unhelpful, either. To use a cliche…”The More You Know!”, haha.

    • exactly. It works so much better in baseball because individual players are in much more control of the outcome of single plays. A batter has what other factors besides his ability to hit the balls. The pitcher throwing against him, the defensive scheme of the defending team, and maybe the weather?

      How many variables does a hockey player encounter on a given play?

      • In baseball a picther throws to a hitter and we get a result that’s base on their interaction. Now defensive alignment is based on stats. If a player make a error no hit is given to the batter. Now in Hockey a player can jump over the boards never touch the puck and be given positive results. He may have enough thing to do with anything happening on the ice. There isn’t any errors noted in Hockey like bad changes or who is on the ice. If they are doing their job. Like what does a player do when he’s on the ice certain players and against certain players.what happens when the ice is better a the beginning of the period as opposed to the end of the period..Then there’s the goalie and how does a player with with certain players on the ice perform against better goalies. To many variables.

        • Here is a list of what will be tracked:

          Shot Attempts:

          Who shot the puck?
          From where?
          What type of shot?
          What happened?
          Was there a pass proceeding it?

          Zone Entries/Exits:

          Where did it happen?
          Who performed it?
          Was there a pass before it?
          What type was it?

          Challenges (Hits/Takeaways/Recoveries)

          What type was it?
          Where did it happen?
          Who did it?

          Yes, there are a lot of variables in hockey. This study looks to isolate those variables. This is the future. Once all the players and the puck are chipped with tracking software and cameras are following each individual player and the puck, all those variables will be easily isolated and we can see how effective each player is each shift.

          • A system similar to what baseball has in Pitch F/X would be excellent for the NHL. It wouldn’t be cheap, but the technology does exist. While it might certainly be true that technology cannot track all of the above reliably, it could track a lot of the variables surrounding puck movement especially. For example, they could almost certainly design something to take care of the whole shot attempts section, if they really wanted to. Puck tracking and number recognition software aspects combined could take care of all of that section.

          • I heard that the NBA has tech that uses 11 different cameras, one for each player and one for the ball. I think the biggest issue hockey has is that changing on the fly will completely confuse the cameras.

            I still think chips on everything that communicate with tracking software is the future.

          • That’s another option – chips embedded in the jerseys. I work in the retail industry, on the landlord side of things, and there is already tech being developed that can recognize a face and track their movements throughout a shopping center. Not really for some kind of Big Brother purpose so much as simply analyzing foot traffic trends. While not practical and out there on a large scale yet, it’s coming.

          • Chris A,

            To further add to your point. Formula 1 has chips in their cars whereas, when a car enters a particular zone on the track (usually on the straights), the trailing car has an option of pressing a button on their wheel to mechanically drop their rear wing hence making that trailing car faster for a few seconds until the car reaches the end of the straight where there is another electronic gate that has the wing mechanically go back up again. This is called “DRS” Drag Reduction System.

            The technology is certainly there and out of all the major 4 sports, you would think the NHL would lead the way when it comes to technological innovation an advancements to further the game.

            I believe there was data recently to suggest that our fans are by far the most tech savy as well amongst the big 4.

    • The whole point of this study is to obtain higher quality data than what has been compiled in the past (simple shot attempts) and to actually see if there are real correlations between advanced stats and team success. To just dismiss it out of hand is dumb. You can be skeptical, but at least wait and see what the results look like before saying it doesn’t work.

      • Ok what does X player do went he takes a pass from X player that’s in his snakes on his backhand and not on his forehand. Which would lead to a higher quality scoring chance. Charting shots and where they come for on the ice doesn’t tell if it’s a high quality shot. It could have fluttered on goal. How is that figured in? Was a teamate busting down the other side and the goalie had to cheat a little. How is the figured in. Was a player being checked while trying to get the shot off. Are these things factored in? Or are all shots viewed as just the same. Who said anything a out dismissing. That was dumb by your dumb ass.

        • Charting shots/locations does give some indication of quality. It’s not perfect, we all know that and readily admit that. But it gives us some indication.

          I think the disconnect is that folks, perhaps yourself too, are expecting a “catch-all” stat that tells the whole story. That’s pretty much impossible. It’s about using what we have to analyze play.

  • What does AV consider an “advanced stat”? Plus/minus? LOL. Girardi was excellent, if that’s the case. +18 baby!

    No, but seriously…I think he may have been considering plus/minus an advanced stat, haha. It’s the only possible statistical justification I can think of to say this team was successful; we only had 7 guys in minus territory, and with a max of -4.

  • I want to scream! Stats and numbers are the end results of actions or non actions. They are indicators of areas that may need to be addressed. When I hear a coach like AV spew some crap regarding a stat I am not sure if he is just giving a sound bite, or wants to seem heady or maybe he is like the manager of a large operation who never leaves the office to touch and feel what is going on in his or her operation. All the bean counters and analyst want to tell you that your problem is this due to a stat. A stat can’t tell you about gap control effectiveness or any fluid aspect of a this really fast game. Its fun for some to have these deep conversations about stats but at the end of the day all you have done is drain your brain in theory and have taken no action to fix what is obvious to see. The decision making of a player, the actions of a player create the stat in hindsight. I guess you can track decision making but thats like second guessing. He should have done this. Sometimes its an obvious bad decision so you don’t need a stat for that. The rest is very subjective and we tend to only question a decision when it doesn’t work out.
    All these guys behind the bench don’t need stats to identify a problem. They are the leaders who drive, fix or create a system. They are constantly looking to tweak it. This stat stuff is great for showtime at these post game press shows. the reporter can seem smart asking dump questions and the coach and player gives a stock answer. This may be a great marketing tool to create buzz but in my humble opinion its a bit overdone.

  • Hi, this sounds exciting, I’ll be trying to participate since I think that underlying numbers will tell us things about certain players that we just don’t notice or can’t, either due to the speed of the game, or due to our own internal selection bias. It will significantly further our understanding of the game, and help up understand what plays and actions on the ice are helpful to scoring goals and winning the game versus those that have been traditionally favored (looking at you, +/- and hits).

    One questions, will you be tracking any passing statistics? I’ve always been curious as to whether passing rates are indicative of possession and winning, and who both attempts and completes the most passes. I know there are a bunch of stats you are taking already, just wondering if this was considered.



    • Co leader of the project here. We will be tracking passes leading up to shots, entries, and exits. They increase shooting percentages is what we know. The whole goal is to be able to apply a good expected goals model in order to replace corsi and Fenwick which are good, but don’t come close to telling the whole story.

  • Back to top button