Thursday, September 5, 2013

Coach Stats vs. GM Stats

Last week the National Post published a story centered on the Leafs Jay McClement and in it McClement referenced some of the stats that the coaching staffs tracks and utilizes. After pointing it out as an interesting tool I received some replies that said it was basically a useless waste of time. I couldn’t disagree more. As someone who coaches hockey I consider some of the things the Leafs coaching staff tracks to be useful as a teaching mechanism for my players, and if I had access to those stats for my team I would use them all the time to breakdown plays, consistencies, weaknesses, strengths, etc.  The thing is, most of these stats are not great indicators of long-term sustainability with the ability to project the future well. Herein lays the key fact of the matter that there are differences between coaching stats and GM stats and those different stats have different values depending on your position and what you’re trying to achieve.

My goal here isn’t to breakdown all the stats and assign them labels as a “coach stat” or “GM stat;” what I’m really trying to do is discuss how stats have various strengths and weaknesses, and how they can help or hurt our judgement depending on how we are viewing the game.

A good place to start would be discussing the stats in question that the Leafs used. There were only two named and they were tracking turnovers-takeaway ratios, and tracking hitting location, both of which I’d say are fairly peripheral for a GM but can be important for a coach to use.

We already know how the Leafs breakdown turnovers, because they’ve told us. Carlyle records turnovers in three categories: 1- Guy is playing as an individual, 2- Offensive player takes chance 3- 'Brain-dead.’ What that really means when it comes to the Leafs tracking turnovers is that they want to erase “3.” You can show a player his CORSI and it will mean absolutely nothing and have no effect on his game whatsoever, but if you’re breaking down his turnovers with him maybe that leads to a swing of shots on net against, to a few more shots for.

A player such as Lupul, for example, is encouraged to try and create offense so while you of course never want to see him turn the puck over, it will obviously happen. As an offensive player you can live with him taking a chance, it not working, and losing the puck. If you don’t encourage him to try things he isn’t going to produce to his full capabilities. However, a player like Lupul has also been known to turn the puck over in his own zone and that’s the stuff you need to work with him on. If you tell a player he turns the puck over too much that’s not going to do anything, but if you sit with him and breakdown where he is making his mistakes specifically and what the problems are, you can use that to teach and instruct.

Furthermore, that turnover description can also be broken down through player roles. Yes the Leafs let Phil Kessel take chances (and again, I’m just using the Leafs as an example here but this applies to every team really), because he’s in a scoring role and that’s what they ask him to do. However, a player like Jay McClement is in a grinding role so the Leafs aren’t as comfortable with him falling under the “1” category. If McClement loses the puck once or twice a game because he’s tried to beat a defenseman one-on-one, that’s probably not acceptable because that’s not his role. I’d wager a guess that they would ask him to chip and chase, or pass the puck to the trailer, instead of deking.

Showing players the type of turnovers they make and what they can and can’t do is how you preach puck management. You can’t just show a player his possession stats and think that’s going to change anything; you need to look into what’s causing that and how you change that. Specifically breaking down turnovers is one way that can be done.

That takes us to tracking hits. Hits have some, little, or no value at all depending on your beliefs, but knowing where a guy is making his hits can be valuable in maximizing a player’s efficiency. Regardless of where you stand on the value of a hit, the ability to hit a guy, separate him from the puck, and retrieve it for possession is important and valuable for any player to have. If player X and player Y both throw 100 hits, and X has 25 hits that change possession while Y has 35, player Y is obviously more valuable physically and I’d like to see the breakdown of X’s hits to see why he’s being physical yet not able to change possession as much.  Although Dustin Brown is a much better player than Cal Clutterbuck, I would compare their hitting styles (as their hit counts are usually similar) and guess that Brown is much more effective at hitting on the forecheck and getting the puck versus Cal Clutterbuck who more so finishes a lot of his checks. As a coach you can’t just shrug and say “well I have Cal Clutterbuck who hits a lot but isn’t very effective at doing so to turn the puck over,” you actually need to try and find ways to maximize his skillset. That’s what good coaches do.

Basically, if I’m a GM am I taking a player who throws a lot of hits that don’t really do anything and hope he changes? No. But if I’m coaching said player because he’s already on my roster, it might (hopefully) be beneficial to track his hit location and try to coach him on how to better use his physicality. Because the Leafs are the team in question who use this stat, I would look at Nik Kulemin and suggest he’s excellent at hitting to get the puck back. The Leafs just brought David Clarkson in and he is a physical player, but if his hit totals aren’t leading to anything that changes puck possession, Nikolai Kulemin is a player that I would use as a model for him to try and use his physicality as (note: this is an example, not a fact; from what I’ve seen Clarkson is great at dumping the puck in and retrieving it physically).

A more appropriate example would be using hit location to track defensive positioning. If the opponent has the puck in the offensive zone corner, passes it off, and then the defenseman on my team goes out of his way to finish his check, he isn’t helping my team and more often than not he’s just putting himself out of position. It would be more appropriate (at least in my view), for the defenseman to locate where the puck is going and get into proper position rather than taking a few additional strides and seconds to finish that type of minimal hit.

There’s no doubt that this is a little thing, but if a coach get a forward who hits a lot to throw 20 more hits over the course of the year that cause a change in possession and that leads to say, two extra goals, while also showing his defensemen when to finish hits in the D-zone and when not to, and that leads to getting into position better and preventing three goals against that otherwise probably would have happened, then that’s a win. A coach can only use what he has –something that is too often forgotten online—so if he’s getting players to be just a little more effective than usual that’s a win.

A lot of the stats used on the internet now are rather ineffective for a coach to use. Are you going to tell a player he has a high or low PDO? What’s that going to change? If a player isn’t scoring but has a lot of chances, he knows to keep going because they will eventually go in. We hear players say that all the time, they don’t need to know their PDO. But should a GM understand that stat to help decide if a player had a career year? Yes. If you’re telling a player to cross the blueline and throw the puck on net to help his CORSI, that’s probably going to make him even worse and make his shooting percentage a wasteland. Yes these stats are sometimes able to help us predict the future, but in terms of using them to teach a player how to make adjustments and correct these stats in and of themselves, they really do just about nothing.

Even in Moneyball, we see the staff talking to players about things such as “if you take a first pitch strike, your batting average drops ___ for the rest of the at bat” versus telling a player “your OBP is too low, now you know, so change that.” Is OBP useful in baseball? Of course. Is it useful in terms of teaching a player how to self-improve though? Not really.

What it really boils down to is coaches use certain stats to cover the nuances of the game, and the GM uses overarching stats that look at the big picture to ask “what’s all this work really producing?” There are many ways to use analytics to help a hockey team and just because something doesn’t directly incorporate shot-counts or goal counts doesn’t mean it’s useless. Getting more shots and more goals is always the goal of anyone working in hockey, but part of the process is breaking down actual gameplay into singular events and seeing where improvements can be made.

This is why, I believe, many coaches are terrible GMs. Mike Keenan (Luongo trade) and Darryl Sutter (Phaneuf trade) immediately come to mind. Being a GM takes a certain mind frame where you always project the future, work within the parameters of the cap, juggle expiring veteran contracts with the rookies in your organization and so on. The best GMs are ones that can properly analyze and predict the future and when to buy low and sell high.

Whereas coaches look at players and see “I like this size in my line-up” or “I want that guy because he wins a lot of battles.” It’s a completely different thought process.


So next time you see an organization discuss how a stat they use, really take a second to consider how they might be using that stat before you instantly criticize it simply because you don’t agree with it. 

1 comment:

  1. Great read! Thoroughly enjoyed that. Very informative.

    ReplyDelete