^ I get what you are saying (a couple of posts up) but correlating goal differential to probability of winning is not that useful either. You have 100 percent probability of winning on a given night if you outscore your opponent.
Goals are a result. What do good teams do to consistently get more goals, though?
I think people are continuing to develop an appreciation of the value of shot based metrics. In this case, Bingo does a great service by presenting the data for consideration.
Does it sound reasonable that shots are a proxy for possession? Sure
Does it sound reasonable that the team with the most possession controls the game and has a higher likelihood of winning? Plausible.
Can you use shot based metrics on an individual player to compare how many shots and scoring chances he generates? Sure
Can the shot metrics distinguish between a cross crease tap in and a puck stuffed in to the goalie’s pad? No, or not necessarily
Does a shot from the so called home plate area have a better chance of going in than an Ovechkin one timer on the PP? Not all shots are made equal.
Can you correlate the outcome of a game to the shot counts? Sometimes. Maybe 55 % of the time? Not in a statistically significant way.
The problem we have with Calgary is that too many times game results, and the eye test, disagree with the inference that you hope you can make from the counting stats. Going by memory, there was a 10 game streak last year where the Flames outshot their opponents on average something like 40-26, and they were getting nothing dangerous, and put up something like a 3-5-2 record over those 10.
Then you had the Hartley years where they gave up a tin of shots from the outside, collapsing and having quick counter strikes with odd man rushes aplenty. Counting stats over the year, and the hypothesis which many want to make, disagreed with the results.
I think by looking at it over many games, you appreciate shot counting stats as data points.
Where things will get interesting is when they get to the point of puck tracking and player tracking. You can then try to categorize situations based on puck movement and play development, and have expected save percentages corresponding to each situation.
I think it’s great to look at counting stats, see what their strengths and weaknesses are, and contemplate how things can be measured better
|