Post by eric on Oct 18, 2015 18:20:56 GMT -6
There is such a thing as luck. If you flip a coin seven times, then flip it another seven times, you will probably get different numbers of heads and tails. This makes empirical measurement kind of hard, but if we measure something enough times we can find the truth.
This is the truth: there is no such thing as matchups.
.
It is very tempting to look at a playoff series and say "aha! Such-and-such cannot guard so-and-so, thus the team has no chance!"... especially after the fact. A common avenue for these proclamations is to cite head-to-head record during the regular season, which makes sense. If Team A beat Team B four times out of four, why shouldn't we expect Team A to beat Team B four times out of seven? After all, teams with a superior head-to-head record have won 69% of their matchups in the post-season!
And there's the lie. The trick is that teams with a superior overall record usually have a superior head-to-head record too, so what we're actually seeing is that teams with a superior overall record usually win. Groundbreaking stuff, eh? Since 1984 there have been 61 cases in best of seven series where the records differed, and that is how we will learn if matchups matter. As it turns out the team with the superior head-to-head record won 25 of those 61 cases for 41%.
There is no such thing as matchups.
.
And that's just with crappy old overall record. If we do something clever by ignoring regular season games decided by 5 points or fewer, we can improve our series predictions from 76.8% to 77.7%. That's not that great an improvement. But putting non-close record against head-to-head record returns 73 disagreements, of which head-to-head wins only 26, or 36%.
There is no such thing as matchups.
This is the truth: there is no such thing as matchups.
.
It is very tempting to look at a playoff series and say "aha! Such-and-such cannot guard so-and-so, thus the team has no chance!"... especially after the fact. A common avenue for these proclamations is to cite head-to-head record during the regular season, which makes sense. If Team A beat Team B four times out of four, why shouldn't we expect Team A to beat Team B four times out of seven? After all, teams with a superior head-to-head record have won 69% of their matchups in the post-season!
And there's the lie. The trick is that teams with a superior overall record usually have a superior head-to-head record too, so what we're actually seeing is that teams with a superior overall record usually win. Groundbreaking stuff, eh? Since 1984 there have been 61 cases in best of seven series where the records differed, and that is how we will learn if matchups matter. As it turns out the team with the superior head-to-head record won 25 of those 61 cases for 41%.
There is no such thing as matchups.
.
And that's just with crappy old overall record. If we do something clever by ignoring regular season games decided by 5 points or fewer, we can improve our series predictions from 76.8% to 77.7%. That's not that great an improvement. But putting non-close record against head-to-head record returns 73 disagreements, of which head-to-head wins only 26, or 36%.
There is no such thing as matchups.
