PumpkinBombX
Registered User
- Jan 29, 2009
- 961
- 52
I've been thinking about some of the stats that use time on ice as a metric.
In thinking about it, i believe they are pretty flawed.
Here's my train of thought on it:
I've been looking at coaching strategies and how they've changed in the past 10 years, as well as set plays. What i keep coming to is how coaching is making the game more football-esque.
With the coaching strategies, i feel like Hockey is becoming a weird type of football game just played very quickly and with many possession changes, but the coverages can be oddly similar.
With football both teams most often get the same amount of possessions per game. (coin tosses can vary this).
Time of possession becomes very important then since both teams will possess the ball pretty much the same number of times. (since they go back and forth either half-time or end of game can really change this, or pick-sixes)
possession time doesn't really then matter with score-effects too much, which hockey would have similar problems.
Hockey doesn't track possessions or possession time but i too think it's pretty similar, i think both teams get pretty much the same amount of touches per game.
If you assume possession time is equal (which we don't know, and i bet the teams all pretty much fall into 1 std dev either side similar to face-offs anyways) then i think ES/60 stats have to be divided by 2 to account for possession time.
or make it like finance and divide by the square root of time.
My problem is, using per 60 makes players look closer than the eye test always does.
Accounting for saying that if they play more minutes their time of possession will only increase by 1/2 that time i think paints a better picture and can be more predictive in what it would actually mean for a players increase or decrease in ice time.
For example, last year McDavid and Brad Marchand were close in ES/60, but McDavid played about 2 more minutes per game ES and had a lot more points. If Marchand played more ES minutes i don't think his points would be close to McDavid's at ES, but i think they would go up by a certain fraction.
With some of the new tracking tech i think they'll get these stats licked, tracking possession time and adjusting for situations etc
In thinking about it, i believe they are pretty flawed.
Here's my train of thought on it:
I've been looking at coaching strategies and how they've changed in the past 10 years, as well as set plays. What i keep coming to is how coaching is making the game more football-esque.
With the coaching strategies, i feel like Hockey is becoming a weird type of football game just played very quickly and with many possession changes, but the coverages can be oddly similar.
With football both teams most often get the same amount of possessions per game. (coin tosses can vary this).
Time of possession becomes very important then since both teams will possess the ball pretty much the same number of times. (since they go back and forth either half-time or end of game can really change this, or pick-sixes)
possession time doesn't really then matter with score-effects too much, which hockey would have similar problems.
Hockey doesn't track possessions or possession time but i too think it's pretty similar, i think both teams get pretty much the same amount of touches per game.
If you assume possession time is equal (which we don't know, and i bet the teams all pretty much fall into 1 std dev either side similar to face-offs anyways) then i think ES/60 stats have to be divided by 2 to account for possession time.
or make it like finance and divide by the square root of time.
My problem is, using per 60 makes players look closer than the eye test always does.
Accounting for saying that if they play more minutes their time of possession will only increase by 1/2 that time i think paints a better picture and can be more predictive in what it would actually mean for a players increase or decrease in ice time.
For example, last year McDavid and Brad Marchand were close in ES/60, but McDavid played about 2 more minutes per game ES and had a lot more points. If Marchand played more ES minutes i don't think his points would be close to McDavid's at ES, but i think they would go up by a certain fraction.
With some of the new tracking tech i think they'll get these stats licked, tracking possession time and adjusting for situations etc
