Analytics for Player Development

[hkstats]

Using stats and analytics, how would you identify what a player needs to work on to improve their game, in a way that can be applied to any player in a meaningful way?

I have a tracking tag system that auto-tracks players stats during games (minor leagues) and I'm looking for the best way to use the data to help each player individually understand where they should focus their development efforts to be most effective.

My take is to give players 'playing style' ratings that essentially have positive and negative contributing factors, then outline for that player the types of things that can improve and detract from the individual ratings, and tips on how to improve those.

Click to read more...

 

[Canadiens1958]

Define minor leagues clearly. Focus on age and skating.

 

[hkstats]

I look at it that competition is all relative in the minor leagues and the data is too, but I can appreciate that as you move into higher competitive situations it maybe becomes more about how you use the tools that you developed before.

To narrow down, lets say ages 12 - 19 upper tiers (ie. Academy, AAA, Major Jr, Jr. A, B). These kids would be good enough skaters to make those levels at their age, but would be a mixed bag, and all can improve something. Our tracking tags can capture almost any stat and where / when it took place.

 

[Filthy Dangles]

Using stats and analytics, how would you identify what a player needs to work on to improve their game, in a way that can be applied to any player in a meaningful way?

I have a tracking tag system that auto-tracks players stats during games (minor leagues) and I'm looking for the best way to use the data to help each player individually understand where they should focus their development efforts to be most effective.

My take is to give players 'playing style' ratings that essentially have positive and negative contributing factors, then outline for that player the types of things that can improve and detract from the individual ratings, and tips on how to improve those.

Don't mean to ruin the premise of your thread but honestly I just don't think the question you are asking can really be answered statistically or analytically. There are analytics in those leagues that help you identify good or undervalued players but if you are talking about identifying intricate and finer points of a players game, i don't think stats or analytics really help you there. A deep dive on the player's film would be a better solution imo.

 

[hkstats]

@Filthy Dangles, thanks for your feedback about using tape, much appreciated. My concern about analyzing players film to develop them is 1) it takes a lot of time, 2) everyone will interpret it in their own way. I think can be really good to supplement collected data though, or to review a good/bad play in detail rather than looking at dots moving around on an ipad.

How I'm looking at it:
If you analytics works to rate a 'good or undervalued player' than really a player/trainer can focus development around improving those exact stats and data points that make up the rating and become a better rated player. Maybe one of the stats contributing to their lower rating is a poor shooting percentage, or maybe its shots against from a zone they defend; if that's the case, the player can work through that in development and improve their ratings.

Maybe the answer is more geared towards tactical type stats rather than measuring skill or technique?

 

[Filthy Dangles]

@Filthy Dangles, thanks for your feedback about using tape, much appreciated. My concern about analyzing players film to develop them is 1) it takes a lot of time, 2) everyone will interpret it in their own way. I think can be really good to supplement collected data though, or to review a good/bad play in detail rather than looking at dots moving around on an ipad.

How I'm looking at it:
If you analytics works to rate a 'good or undervalued player' than really a player/trainer can focus development around improving those exact stats and data points that make up the rating and become a better rated player. Maybe one of the stats contributing to their lower rating is a poor shooting percentage, or maybe its shots against from a zone they defend; if that's the case, the player can work through that in development and improve their ratings.

Maybe the answer is more geared towards tactical type stats rather than measuring skill or technique?

it's a really complicated question, but i will try and break down the problems i have with it. in my opinion, analytics moreso tell you if a player is good or undervalued and if there's a problem but perhaps not necessarily why.

forget about puck tracking for a sec, and just go to the more mainstream shot share and other event share metrics like Corsi and Goal share stats. They can identify potentially undervalued players but they don't really telly you why or what player might do to make them undervalued.

puck tracking might tell you how fast a player skates....but it can't tell you what might be wrong or inefficient about a younger player's stride and skating mechanics and how to rectify it.

puck tracking might tell you a player shoots a poor % from high-ish danger areas....but it can't tell you anything about shooting mechanics, deception etc.

puck tracking might tell you a player struggles in transition....but again it can't tell you why or what specific traits or aspects are the problem.


that's my biggest issue with trying to use analytics in terms of refining aspects of players game. sure it might help tell you there's a problem but you're still gonna need to dig deeper. so I guess it could be used as a tool to help identify but it only leads to more questions and you're gonna need to go 'under the hood' in scouting and looking at film and working with players on the ice to really solve the bigger problems.

 

[DatsyukToZetterberg]

it's a really complicated question, but i will try and break down the problems i have with it. in my opinion, analytics moreso tell you if a player is good or undervalued and if there's a problem but perhaps not necessarily why.

forget about puck tracking for a sec, and just go to the more mainstream shot share and other event share metrics like Corsi and Goal share stats. They can identify potentially undervalued players but they don't really telly you why or what player might do to make them undervalued.

puck tracking might tell you how fast a player skates....but it can't tell you what might be wrong or inefficient about a younger player's stride and skating mechanics and how to rectify it.

puck tracking might tell you a player shoots a poor % from high-ish danger areas....but it can't tell you anything about shooting mechanics, deception etc.

puck tracking might tell you a player struggles in transition....but again it can't tell you why or what specific traits or aspects are the problem.


that's my biggest issue with trying to use analytics in terms of refining aspects of players game. sure it might help tell you there's a problem but you're still gonna need to dig deeper. so I guess it could be used as a tool to help identify but it only leads to more questions and you're gonna need to go 'under the hood' in scouting and looking at film and working with players on the ice to really solve the bigger problems.

I don't think they're two separate issues, I kind of think they are more of a step 1 and step 2. I think to properly use data you need to have a development approach that can make changes based on what the data is suggesting. You will need some sort of video, but I think good data should be able to easily and simply show what is happening as well.

You can use the data and analytics gathered to identify certain traits that may be leading to a players performance. Right now we can loosely tell what a players contributions are on the ice by looking at their shot share, xGF/xGA, GF/GA, and on/off ice numbers. While these numbers can't tell us exactly what they're doing it lets us know what happens when they're on the ice and if they're getting lucky/unlucky. For example, if a player is constantly able to get to high danger areas and is unable to score the data will reflect that, but it won't tell us why they aren't successful, at least not yet. Perhaps they're constantly shooting the puck into the "cold" areas of the net and a new curve or stick will help them alter their shot just enough that they can begin to capitalize on more of their chances. When a player makes an effort to change something in their game the data should reflect those changes, assuming that it is able to capture those changes.

I think we will eventually see is something similar to Driveline develop, but with a hockey focus. Driveline is an analytics focused baseball training service that is offered to MLB players, minor league players, etc. and they've been part of the reason why there have been so many player adopting new swings and modern approaches. They helped show the value in elevating the baseball, the importance of launch angle, and really helped change how players and teams approach hitting. You see it quite a bit in baseball when a player makes a swing change they have lackluster MLB numbers and get sent down to AAA. They alter their swing and get recalled months later and they look like an entirely different player.

Smart teams will begin to look at every aspect of a player, things such as a players curves, shooting motion, and other areas that they may not have paid attention to previously. If you have a player with excellent skating, but a lackluster shot what can a team do to unlock the most of their skillset. A book was recently released called "The MVP Machine" and it had an exert talking about how the Houston Astros completed overhauled their player development and began incorporating data into their players development. Since they've made those changes they have been one of the most consistent player development teams as they just seem to make every player more effective upon joining their team.

It may be a little harder to make noticeable changes to a hockey players game due to the nature of the sport, but there most be ways for teams to find some player inefficiencies similar to what was found in the MLB.

 

[StreetHawk]

Is there a link or video of any sorts that shows plays that generate positive and or negative analytic stat?

I hear numbers and stuff but it would be nice to see plays to demonstrate how it gets recorded.

 

[Filthy Dangles]

Is there a link or video of any sorts that shows plays that generate positive and or negative analytic stat?

I hear numbers and stuff but it would be nice to see plays to demonstrate how it gets recorded.

This explains it a bit, but it's still somewhat obscure as the actual data is black box.

But essentially, it can be distilled down to transitional play.If we talk about defenseman for example, we often think of their offense (how many points they put up) and their defense or play without the puck. A big component that is often overlooked or not understood is transitional play. The diamond in the rough defensemen are the ones who don't put up big points but make a lot of those 'possession driving plays' and what not.