All Hokie, All the Time. Period.

Virginia Tech Basketball Board

Gobbler-100

Joined: 9/23/02 Posts: 20146
Likes: 977


It's just an abomination. The easiest way to explain it is...


...the RPI treats the difference between a 5 win team and a 10 win team the same as the difference between a 13 win team and an 18 win team.

And that is obviously crazy. If you are going to be a major conference team that wins about 20 games, you are going to beat a 5 win team. You are going to beat a 10 win team. Those are wins. There might be a 1% difference in the likelihood that you are going to win those games.

But there's obviously a much bigger delta in your likelihood of winning vs. the 13 win team and an 18 win team. There might be like a 20-25% gap in your win likelihood between those two opponents. But to the RPI (assuming those teams themselves played roughly similar schedules), the gaps are the same. If you play the 18 win team and the 5 win team, that's the same SoS as playing the 13 win team and the 10 win team.

This, in a nutshell, is why our OOC SoS always looks so outrageously bad when looking at the RPI or other systems that calculate SoS in similarly idiotic fashion. If you aren't using win likelihood to calculate your SoS, your SoS is a TRASH METRIC.

Is some of this our fault? Yes, undoubtedly. The game is the game. We have known for many, many years that this is how it works. A bunch of people totally unequipped to analyze an array of stats are given team sheets that scream OOC SoS = 315 or something like that (as calculated by the RPI) and they rub their chins and go, "Well, this team certainly hasn't challenged themselves in non-conference play. I'm going to arbitrarily decide that they need to go 12-6 in the toughest conference to sufficiently atone for this scheduling abomination to the extent that we may bestow upon them the honor of participating in a play-in game." And that may be horrifyingly poor analysis and unfair and wrong...but, still, we know that's the risk. You play 5 or 6 bottom 50 teams and you are placing a hard cap on your RPI metrics right out of the gate. And it could cost you.

So, yes, it's our fault and, yes, holy God it's also incredibly stupid. Just remember that there's no reason to worry about it until we actually get to the point of worthiness of being snubbed. That is, none of this matters if we don't win at least 2 more regular season games. Then I'll start freaking out about how we're gonna get jammed.

(In response to this post by hokiepro)

Posted: 02/14/2018 at 12:29AM



+6

Insert a Link

Enter the title of the link here:


Enter the full web address of the link here -- include the "http://" part:


Current Thread:
 
  
RPI need to be scrapped; we’re 59 dropping -- hokiepro 02/13/2018 11:22PM
  It’s extraordinarily flawed -- hokiepro 02/14/2018 10:07AM
  And that proves my point -- hokiepro 02/14/2018 08:53AM

Shop Our TSL Affiliate Links!


Our Sponsors








vm311