There's two issues with Chayka imo.
First of, credit where it is due, he's a very bright and impressive guy. The way he approaches solving a problem, is very similar to the way many in my business do, that's empirically, and with data.
The first issue is that the data he's using. Hockey, unlike baseball is a fluid sport. There are no stoppages between plays where you can measure and outcome from start to stop.
The second issue is, hockey unlike basketball, another fluid sport, does not have it's situational passing/shooting/scoring mechanism result in a goal more than 5-6 times a game, as opposed to 160 times a game in basketball.
Both of these phenomena present an issue when it comes to data integrity, and order of magnitude. With baseball for example, you have a much higher level of data integrity, lefty v righty, pitch type. Start. stop. Basketball: 3-pt off of screens/drive and dish etc.
So those situational measurements are far more quantifiable. With Corsi for example, it's the same metric of a soft 60 wide attempt on net, as it is a slot shot. Hockey has gotten much better about situational metrics, and group metrics in particular, but the reality of it is that these stats and metrics require a long tail view... which many teams do pay for... but it's a recent trend. Also, different players will perform differently in the same metrics on different systems so there's no normalization curve.
So what's the point of this: Chayka and co might be too early for what they really need. Which are a universal set of metrics that have a high differential value, and can be replicated over and over again.
Does the game of hockey have some of these? Yes. But sparse, and circumstantial at best.
Here's one:
3 RH shots on the PP. LH half walk, RH Point man. Net front is a preferred RH shot. Sniper from the off-wing key. LH playmaker below the circle. PP is a 1-3-1. Teams that run this consistently-> have a 18-26% conversion on the PP. It's very hard to stop. Additionally, the teams that have mastered this, have either won the cup, or been to the conference finals the last few years. They are contenders.
Why is this set up to replicable: because there is less fluidity on the PP, there is a clear and distinct differential value based on the situational setup and outcome.
Here's another: players are tracked on a controlled zone entry, but not just any kind. There are two players who are UFA this year who are going to get wtf contracts according to HFNYR standards. They are checkers. But checkers who are very good and grabbing the puck in the d-zone, and skating into the offensive zone, often times solo, and allowing their team to get a fresh line change while establishing a forecheck.
One plays on a very good team, one on a crappy team, but he can score. HFNYR value for one of those guys would be your typical $1.5M 4th liner, and the other as a stretch for $3.5M. The "4th liner" is going to get 4 years at $2.5M minimum. The $3.5M player is going to get 4 years at a minimum of $4M, maybe higher.
Why is this set up to replicable: because there is a clear and distinct differential value based on the situational setup and outcome, and it can be stack ranked against their peers.
Just some food for thought.
Bloggers: Feel free to run with this. Your content needs some refreshing.
