BPI - What is it and WTF?
Looking at ESPN this morning, I clicked on the link to "BPI" - a supposed Basketball Power Index. Updated today it shows Michigan at #22, behind Wisconsin, Iowa, OSU, and MSU, good for only the #5 rated Big 10 team and a projected 6 seed!!!
I would assume this is not looked at for NCAA seeding, right? How can this possibly be calculated to come out with this rediculous result?
<edit> Even better, it projects Michigan as an Auto-Bid for the Big Dance - meaning it expects M to win the Big 10 tournament and still only get a 6 seed? It is insulting that they even would post these rankings in a format that would suggest it as data.
I think it's a bit more blueiculous than rediculous.
I had never looked at BPI before. I was amazed a rating system could appear to be more flawed than RPI.
bpi is like rpi, kenpom, sagarin, etc. it's just a ranking system. however unlike the others, it massively discounts victories over teams missing "key" players. it was invented by espn last year.
because it discounts our wins against MSU (missing "key" players, huh, huh), it isn't as favorable to us as it would be in other years. but there's not much we can do about that. it is one of many ranking systems that the tournament committee looks at and it shouldn't hurt us.
Thanks for the explanation. It is distressing that the Selection Commitee would have this "data" in front of them at all.
I could be wrong about this but I don't believe the committee uses this "tool" at all in the selection process. I think this was just something ESPN created so they could use it on their broadcasts instead of RPI or Kenpom.
Officially, they don't. Unofficially, they can take any information into the selection room they see fit. I'm guessing they don't leave their laptops/smartphones at the door; they've got access to every ranking system out there.
But have they seen the size of our balls?
http://mgoblog.com/mgoboard/mbb-i-have-huge-balls-hand-motion
Relax, man. It's one of many alternate-to-RPI ranking systems out there, all of which have flaws one way or another (because there will never be such a thing as a perfect system) and none of which have any bearing on seeding whatsoever.
"Mai favrit team iz 2 low!!!" is widely reviled as the worst possible critique of a ranking system there is.
Does it give us a bonus for playing without McGary?
it doesn't because it basis its assessment of a "key player" on minutes played. mcgary never accumulated enough minutes to be a key player. that's one of the reasons that bpi undervalues our season and overvalues state's. but worry not; the committee looks at many factors and will not give us a lower seed simply because some dude at espn came up with a system that gives us an anomalous ranking.
The BPI also really screws us for playing some terrible non-conference teams. That is the biggest thing holding our rating down
Then should UM be reward for winning most every game while missing a key player? (McGary anyone!)
We have been missing a pre-season All American all year, WTF indeed.
For what it is worth, ESPN explains their system to some extent in this 2012 article here - LINK
I will say, although it is annoying that they do not go into much technical detail on the claim that most would not be interested (obviously, they've not seen us pick apart systems here), they do mention one thing that they have somehow incorporated:
On top of this, we decided to incorporate a little bit more information than the other power ranking systems use. In particular, we added a way of accounting for missing players. If a team or its opponent is missing one of its most important players (determined by minutes per game) for a contest, that game is less important for ranking the teams compared to games in which both teams are at full strength.
So, that might explain MSU to some extent in their system. In any case, there is a chart in the article comparing what they factor into this to Sagarin and Kenpom as well.
ESPN may SAY it's better than a Colley Rating, Pomeroy Rating or the RPI --- but they aren't putting the formula out there for anyone to critically review it.
Each of the three ratings above is replicable --- they may each have their flaws but at least one knows how the rating is calculated.
For BPI, for all we know ESPN has inserted a line of code that says "IF TEAM = 'SYRACUSE' THEN RATING = RATING + 10." Only they would know.
Freaking ESPN Voodoo
doesn't take it into account.
Even if the ranking is sensible, there is far too much room for ESPN to tweak them to fit their personal agenda. And I mean, we could blindly trust that ESPN would not do that...but come on. The whole idea of rankings is to be totally objective. And it's kind of hard to prove that when we have no idea how there rankings were made. It's way too open for abuse.
Especially seeing how the supposedly are taking into account "missing players". How? Are you doing it consistently? Does it take into account missing a guy like McGary? It seems like a very very subjective thing, and I wouldn't be surprised if there were a variety of "fudge factors" in their formula.
On your last point, pretty much every bracket projection uses the current conference leader as the filler for the autobid until they're knoced out of the conference tournament, at which point they're replaced by the highest remaining seed. That doesn't mean they've already plugged wins from the conference tournament into their computers.
It still is absurd to have a fill-in slot that intuitively contradicts other data presented.
players (determined by minutes per game) for a contest, that game is less important for ranking the teams compared to games in which both teams are at full strength.
Given that UM is ranked behind a team they swept—Michigan State—the translation of this metric is thusly:
If Michigan State is missing one of its most important players (determined by minutes per game) for a contest, that game is less important for ranking the Spartans compared to games in which Michigan doesn't have a consensus lottery pick available the entire fucking season.
If the purpose is to project results for the remainder of the season, and especially if the purpose is to project results for the tournament, it makes sense to account for missing players that will return and not for missing players that are done for the year.
What about a team like Arizona that loses a guy for the remainder of the season? Do their losses mean less when predicting their future success? Unless, of course, they are manipulating data by hand, which renders the whole ranking as potentially biased.
Also, at some point, you have to just discount the effects of the injuries because of the potential of another injury happening in the future.
I don't think it's really a problem to manipulate it by hand. It makes sense to treat the injuries to Brandon Ashley and Mitch McGary differently than those of Dawson / Payne / Appling. If the entire goal is to predict, you'd be silly not to use that information. Not accounting for injuries (assuming you do it in an intelligent way) is going to make your metric worse.
Apparently there is no consideration given to a coaching staff's ability to compensate for a missing player.
ESPN's BPI geniuses to Tom Izzo: "There, there, Tommy, let's dry your eyes and blow your nose...that's good, now blow. OK, we promise to be nice to you since two of your players have boo-boos but are coming back."
To John Beilein: "Mitch who?"
and do not watch on TV or visit their web page. Their magazine is just awful.
ESS EEE SEE / NY Yankees / NY Jets network. Screw 'em.
This ranking system had Florida at/near #1 almost the whole year. And then we smoked them by 20 in the tourny.
with their "regulars" back, perhaps we should get bonus points for playing Staee when they were injured since they appeared to be better then.
Here's a table from ESPN comparing some of the popular rankings systems.
System comparison
How is the College Basketball Power Index by the ESPN Stats & Information group) different than RPI or other advanced rating systems like Kenpom.com and Sagarin? Here is how the included elements compare to other systems.
| Includes | RPI | BPI | Sagarin | Kenpom |
|---|---|---|---|---|
| Scoring margin | No | Yes | Yes | Yes |
| Diminishing returns for blowouts | No | Yes | Yes | No |
| Pace of game matters | No | Yes | No | Yes |
| Home/Neutral/Road | Yes | Yes | Yes | Yes |
| SOS beyond Opponent's opponents' W-L | No | Yes | Yes | Yes |
| All wins are better than losses (before Opp Adj) | Yes | Yes | No | No |
| De-weighting games with missing key players | No | Yes | No | No |
They are all best guess ways of predicting. I don't think you can necessarily say the system is crap because it doesn't like Michigan.
ESPN's insistence upon creating their own statistics (BPI, QBR) drives me nuts.
This seems to be their approach to it:
If anybody's interested in seeing how the various systems stack up historically, Massey has a comparison page:
http://masseyratings.com/cb/compare.htm
Top left is a link to "archived", where he's got this more or less weekly going back to 2003.
I think it is a bit down on Michigan, but it's a system ESPN is trying to make relevant. Kind of like their QBR stat. I think there is merit, but like all predictive models it can be skewed one direction or other because of the input set, and I think the one issue is that they take in so many different factors (cascading SOS, missing players, etc.) that it may be introducing so much noise that is is causing issues.
BPI = MSU friendly rankings. BPI actually discount losses if the team wasn't playing at full strength.
/not being sarcastic