[Itse]

Originally Posted by DeluxeMoustache

I’m certainly not making a claim that the Flames have made good use of any particular stats on the ice. That is obvious. Also, fyi, the conversation I had on this topic was in the second half of last season.

My only point was that people who actually do this for a living (meaning on or on behalf of NHL teams) aren’t relying entirely on simple counting stats.

Many people on CP and the Fan keep talking about them but the people who do are far behind. Take a guy like steinberg - he is paid to talk for hours so he will do his best to bring whatever data is available to the table. He works hard, and relies on public domain info. Good on him, but his analysis is often skewed.

I am pretty sure a guy like Chris Snow isn’t plunking down a folder on the desk of Gulutzan and Treliving showing which teams have out scored their opponents and how they rank in the standings.

As for what it has to do with your argument, I am not sure what you are looking for.

We both agree that simple counting stats like Corsi fall short.

You seem to think that goal differential is a better predictor I believe.
I, like many, believe that goals for and against are more of an outcome based statistic.

Maybe another way to say it is this. The use of goal differential as a stat as predictor of team success is about as useful as a coach telling his team that their strategy is to outscore the other team.

Sorry for coming across as grumpy, wasn't actually intended. In any case, I guess I could elaborate a bit on my thoughts on hockey statistics, since this is what the thread is so much about (again).

If the question you want to ask is something like "assuming no big chances, how is the team likely to do for the rest of the season" or "will we make the playoffs playing like we are", I truly believe that you're best off not looking much further than standings and goals. That information on it's own is already highly statistically predictive. You don't need to look into the secondary stats for predictions.

Remember that what ever secondary stats you have, they are not from the future. We already know what results you got with those secondary stats, their effect into performance has already happened in the games they were tracked in, and the sum result of those stats is visible in the goal differential and the standings.

In other words, you really can just look at the standings and goal differential and get a good idea on how things are likely to play out ... IF NOTHING CHANGES.

However, if you're trying to manage or coach a team, you're always looking to improve the team, and that's where secondary stats become relevant. They're not very good at predicting stuff, but they do tell you a lot more about how you got to where you are now than primary stats.

If you're not doing well, what's the problem? Are we not getting enough shots? Are we shooting from bad positions? Not enough odd-man rushes? Are we getting in on the rebounds? How fast do we get shots on net after a turnover? I can think of about a million questions that could potentially give you further insight into the way you have been playing, and that insight can, at least theoretically, be used to improve the team and thus get better results in the future.

So what I'm saying is this:
- Primary stats are a predictive tool. They tell you not just how you did in the previous games, but are also quite reliably pointing to where the team is heading, unless nothing chances. (This is a statistically proven fact. If you choose to not believe, you're just being obtuse.)
- Secondary stats are a post-factual analytics tool. They tell you what happened in the games so far, and can be used to make changes so you play better in future games. Or a coach might better identify strentghs and weaknesses of the team and change strategies accordingly. Or maybe the GM can see something that could be helped with a player trade.

So you should use shot-based metrics (for example) exactly opposite to how they're commonly used by fans and media right now.

You can't say "we will do better than we are now because we have good corsi", because you already had that corsi in the games you played and it didn't help. But maybe you can improve your corsi if it's bad and see if it helps in the future Or get more shots from high-danger areas or something. But if you already have high corsi and you're doing poorly that's worrisome because it means there's no room for improvement in that area.

If you don't understand the above, think of the team as a race car.
Primary stats are your lap times and standings in the positions. This is the information on how fast you're going and how well you are doing relative to the competition. Secondary stats are the performance measurements of the car. Horsepowers, wind resistance, grip, brakes, what have you.

Once you've had a few races you already get a pretty good idea on how things are going to go. The teams at the top of the first races are likely to be at the top at the end of the seasons unless others improve their cars midseason. You can use technical measurements to see where you're lagging behind and could possibly improve, but you can't claim that "we have the most horsepowers so we're going to rise up the standings, just you watch". That's not how it works. The results you're getting have been with those high horsepowers. In fact, if you have the most horsepowers it's more likely that others will catch up to you and you'll start doing even worse.

If all your technical measurements point to a competitive car but you're not doing that well in the standings, you should probably try to find a better driver.

(Obviously secondary stats and team performance have a much more complicated relationship than mechanics have on racing car performance, and I'm not saying that good secondary stats are actually bad and bad secondary stats are actually good. I'm just saying that when there's something wrong with the car, you want to find what's wrong so you can fix it, and if there's something wrong with the team secondary stats might point to what you need to fix. And maybe if you can't find anything wrong with the team you should at least consider the possibility that your coach isn't that great.)