Quote:
Originally Posted by Freeway
True. But at the same time, based on the (admittedly small) historical sample size, you can understand the advanced stats community's hesitance in proclaiming the greatness of the Flames. If history suggested that I had a medical condition that had a 95% chance of impending death, no matter how much I felt I could be the exception or thought the historical sample was flawed or skewed, I'd still hedge my bets a bit.
|
Sure I can see why the skepticism, but your analogy to a terminal illness is exactly what (IMO) is wrong with a lot of the "advanced stats community". In your situation ALL you can do is hope. There is generally nothing you can do to battle you're way into that 5%. You can't look back at your own case to see where you went wrong and what you can do to improve your chances of being in the 5%.
In sports, you can. You can improve your chances of not regressing by working on the things that had you in those spots in the first place. To be under the assumption that all these statistics will revert back to the mean has you under the assumption that none of the other (pretty much infinite) other factors will change. Only their shooting percentage and, thus, they will score less. A teams combined shooting percentage is pretty useless, because certain players will affect it differently either positively or negatively. This again gets lumped into an average that is just not indicative of the play of the team. There's nothing average about anyone in the NHL. All of them have skills that defy the average, hence why they are top-level athletes. To assume someone like Gaudreau's or Monahan's shooting percentage will regress to an average assumes that they are average shooters, which they are not. It also assumes that they will take no strides to improve this part (and others) of their game, which they do. And then, as mentioned, it doesn't take into account the affect that new team members (positively or negatively) may have on those events. You can't do a bunch of exercises that will potentially improve your terminal illness, but if you spend 2 hours a day taking faceoffs, your faceoff percentage is likely to improve.
The stats are an interesting way to look back at what happened and to tell you what needs to be improved on for sustained success, and that's exactly why they can't predict the future. Because their existence means they can be reacted to and thus the possible outcomes are constantly changing.
Think about the article by Quick about top shooters. Some of them are because they literally power the puck past the goaltender. Some of them are because they can shoot at full (or close to) velocity from different stick positions. Some of them are because they have deceptive body movements and so it makes it difficult to predict where they shoot. We don't even have individual players average shot velocity, and you'd have to combine that with their average shooting percentage, what goalie their facing (and all of their individual averages on how good they are on glove/blocker/low shots, moving sided-to-side), what defenseman (again with their individual stats) they have to shoot through, how much of a threat their teammate is (which at every individual shot will be different depending on who that player is, where he is on the ice, how long theyve been out there, etc) how often they have to face Jonathan Quick vs Ben Scrivens, etc.. And even then, you'd only get an average of what they MIGHT do.