Quote:
Originally Posted by Resolute 14
The problem with the advanced stat crowd - and in particular the trolls who use them to cause arguments - is that first and foremost, you judge based on what you expect to happen. Then you fit the stats in where they support your position and ignore when they don't. That is why this argument is never "We all knew Toronto, Montreal and Colorado would eventually fall off". Because Montreal completely effs up the narrative you want to create.
|
I actually strongly disagree with this. I think everyone was nearly as surprised that the Habs did what they did (Boston should have beaten them, they outplayed them over the course of the series). The reason Montreal's success was less expected than Colorado's was,
1. They had a goalie who had a track record of being as good as he was (and still is). Varlamov had no such track record. Turns out stats guys were wrong about Varlamov.
2. Montreal's possession numbers were below average (still are), but they weren't ATROCIOUS like the Avalanche. EDIT: after checking, some actually were atrocious. Fenwick close numbers had a 2 point gap, but 5v5 CF% was basically equal.
In the end, I disagree that it's a matter of post facto rationalization. It should be about post facto explanation of why your predictions didn't pan out where they didn't pan out. In the Avs' case, it was their ability last season to sustain unreal goaltending that was totally out of the realm of what you could have reasonably expected from their tandem before the season, and their shooting percentage not falling off. In other words, the exact same formula that led to the Leafs making the playoffs 12/13.
In Montreal's case, as stated, they simply weren't as bad. But they also got a bit of luck vs Tampa that Bishop was out, and in the Boston series, the inherent issues with small sample sizes in a single series allowed them to beat odds that were in Boston's favour. But you have to realize that in no case are those odds insurmountable. If the Flames finish in 3rd in the pacific and play the Kings, it doesn't mean the Flames can't beat the Kings. It just means you'd be ill-advised to bet on it.
This was put really well by Cam Charron (borrowing from Garrett Hohl) when he was writing for Leafs Nation before he was, erm, hired by the Leafs. He said about the value of analytics,
Quote:
|
Those readings led me to accept probability, not destiny, if I have permission to steal a line frequently used by my friend and occasional bartender Garrett Hohl. If you were given the option to pay 45 cents to predict the outcome of a coin flip, and should you be correct in your prediction you win a dollar, you may lose 45 cents on the first try and you may even lose 90 cents after the second try—but if you could take the same bet an infinite number of flips, you could write a script to program millions of flips per second and retire early. There's value there. If you pay 45 cents for an expected return of 50 cents ($1.00 divided by two possible outcomes) you can expect a large return in the long run even if you'll face a loss 50% of the time.
|
Analytics-based predictions, even when flawlessly reasoned (which is hard and depends on data quality), are not prophecies. They only predict the more likely outcome. Variances will occur - some major like the Avalanche; some less major like the Habs, and to some extent these variances will have good explanations underlying them (goaltending quality or good special teams results). Other times they may be random. Usually some combination.
If people could look at it that way and hopefully are able to be rational and not emotional, they'd be better equipped to participate in these discussions without knee-jerk extreme reactions.