Originally Posted by squiggs96
I bet there are some stats and graphs for it.
*suit, not suite
You can do whatever you want with your free time. You obviously don't need my permission for it. If you want to look at data, that don't correlate to wins or losses, then go ahead and do that if it floats your boat. What I'm getting at is that sometimes the CA60Vs16 will be above/below the red line when the Flames win/lose. There is no correlation between winning/losing and the position of the CA60Vs16 relative to the red line, and there are not indicative. You want them to be, but they aren't. When they don't go the way you expect them to, they are always explanations. That's not how statistics work. You aren't just counting and presenting. You are counting, presenting, and creating a narrative based on what you want them to show.
Here's how I'd work backwards to find stats that are useful. My end goal is to win the Stanley Cup. To give me the best possible road to achieve this objective I'd like to have the most points in the regular season. This will give me home ice advantage throughout the playoffs. In order to get the highest see I'll need the most points. In order to get the most points I'll (usually) need the most wins. In order to get the most wins I'll need X. It's debatable what X is. If I believe X is special team goal differential, I'll look at stats that will help me draft and trade for players that will help me achieve this metric. X isn't one thing, so X is likely X, Y, Z, etc. I rank these in order of importance to me, and start trying to build a roster.
The stats I'd find useful are ones that correlate to my end objective. If my stats nerds on my team kept bringing me stats that said I need to find players that did Y because it helped with a metric, but that metric didn't correlate to wins or losses over the course of a season, I'd tell them to stop wasting their, and my, time. Why tell me we need to be better in some area if it doesn't lead to wins, which is the point of hockey.
Let's suppose you think having high danger chances is a key component to winning. It might be. It sure sounds like it should. However, if you have a bunch of plugs, relative to other NHL players, getting high danger chances, what good is it doing? Sure you are getting some fancy stats, but it isn't achieving anything. To simply say the output is unlucky is not how stats work.
Now back to the charts. You think it is better to be above the red line, but the chart shows it's the opposite. That's not me creating advanced metrics, that's just me reading the outputs from the data that is counted. In the 17 games that have bar charts, I've been told it's better to be above the line. In post #31 of the thread in the first bar chart there are nine games where at least one of the bars is below the red line. The Flames are 6-2-1 in those games. If there truly was an indication or there was a correlation between this stat and winning and losing, then you'd have to think that you'd need to be above the line in order to win the majority of your games. In fact, the Flames captured 72% of the points with this happening. To me, that says there isn't a point in needing to be above the red line. In the same graph the two biggest grey bars are on games they got soundly beaten.
If the metric was truly a great metric is would be more closely aligned with wins and losses. You are free to think that you need to have more chances, possessions, shots, shots from certain angles, etc., is better, but when the data keeps being presented over and over, and it doesn't tie to results, I'm not sure what the point of it is. I'm also not sure how much more you can say. If you keep telling us that a team needs to rank favorably in a certain metric to be successful, but the data shows that's simply not true, it's time to see if the output is relevant.
|