Calgarypuck Forums - The Unofficial Calgary Flames Fan Community - View Single Post - Not to worry; this won't last says Myrtle

Calgary4LIfe · 11-18-2014, 10:01 AM

I am sure this has already been mentioned, but the sample size is still small.

It was a very good predictive tool for the Leafs and the Avs for one reason - their bad numbers lasted the entire season (yet both teams made the dance, after all). The following respective seasons, they seemed to come back down to earth.

The numbers for the Flames aren't great. There is no denying that. However, those numbers are better I bet if you break it down into 5 game segments and use those numbers to see if there was a trend. Calgary got very lucky against Chicago. Calgary got unlucky against Montreal the 1st game. Stuff happens, and in a small sample-size, this does skew results. If you are going to use a small sample size, I think it is often helpful to break it down further into smaller segments to at least spot trends as they are happening, which I find few (if any) advanced stats media gurus are doing for the Flames.

I think advanced stats are extremely useful and help to describe what is happening. They are SUPPOSED to help remove bias, but they are often used in order to support a bias as well. They are also not perfect. Sure, they accurately predicted the collapse of two teams, but those same two teams made the playoffs (and one made it as a divisional champ in a very difficult division).

I am not an advanced stats guru by any means, but I do enjoy examining them. I don't understand them all - just haven't really bothered all that much to be honest - but I find them more and more adding bias to a discussion these days than removing them. What I would love to see is what the teams are really using. What is the predictive stats that they employ, how are they removing biases, and what do those numbers say about the Flames?

There is always going to be a fair measure inaccuracy with these stats. What constitutes a take-away? What is a scoring chance? What is a hit? Was it really a shot on net, or was that a deliberate play off the boards in the hopes of creating a scoring chance since the lane wasn't open or there wasn't enough traffic in front? What about the stats when a team is up by a goal, down by a goal, or even? What about when they are down by 2? Down by 3? What about when they are missing guys who are injured? Obviously not all injuries are the same, so is there a way to 'rank' the games lost to injury so that the stats don't get skewed? Do we care if it gets skewed?

There are simply too many variables that are too difficult to track within reason I think. I think it is very difficult to be consistent and unbiased. Can you trust advanced stats? Yes, and no. I honestly believe advanced stats helps to paint the picture. Maybe just helps to keep things in focus. I do not think they are the whole picture.

11-18-2014, 10:01 AM	#222
Calgary4LIfe Franchise Player Join Date: Jun 2010 Exp:	I am sure this has already been mentioned, but the sample size is still small. It was a very good predictive tool for the Leafs and the Avs for one reason - their bad numbers lasted the entire season (yet both teams made the dance, after all). The following respective seasons, they seemed to come back down to earth. The numbers for the Flames aren't great. There is no denying that. However, those numbers are better I bet if you break it down into 5 game segments and use those numbers to see if there was a trend. Calgary got very lucky against Chicago. Calgary got unlucky against Montreal the 1st game. Stuff happens, and in a small sample-size, this does skew results. If you are going to use a small sample size, I think it is often helpful to break it down further into smaller segments to at least spot trends as they are happening, which I find few (if any) advanced stats media gurus are doing for the Flames. I think advanced stats are extremely useful and help to describe what is happening. They are SUPPOSED to help remove bias, but they are often used in order to support a bias as well. They are also not perfect. Sure, they accurately predicted the collapse of two teams, but those same two teams made the playoffs (and one made it as a divisional champ in a very difficult division). I am not an advanced stats guru by any means, but I do enjoy examining them. I don't understand them all - just haven't really bothered all that much to be honest - but I find them more and more adding bias to a discussion these days than removing them. What I would love to see is what the teams are really using. What is the predictive stats that they employ, how are they removing biases, and what do those numbers say about the Flames? There is always going to be a fair measure inaccuracy with these stats. What constitutes a take-away? What is a scoring chance? What is a hit? Was it really a shot on net, or was that a deliberate play off the boards in the hopes of creating a scoring chance since the lane wasn't open or there wasn't enough traffic in front? What about the stats when a team is up by a goal, down by a goal, or even? What about when they are down by 2? Down by 3? What about when they are missing guys who are injured? Obviously not all injuries are the same, so is there a way to 'rank' the games lost to injury so that the stats don't get skewed? Do we care if it gets skewed? There are simply too many variables that are too difficult to track within reason I think. I think it is very difficult to be consistent and unbiased. Can you trust advanced stats? Yes, and no. I honestly believe advanced stats helps to paint the picture. Maybe just helps to keep things in focus. I do not think they are the whole picture.