Friday, February 24, 2012

rpi-fixing

Good old RPI.  It's a constant source of headaches, hate, and discontent.  It seems like a simple enough formula, easy to calculate if you've got enough time or a program to do it for you.  And yet the Internet can't even agree on what it looks like.  ESPN thinks UVA is 32nd in it; RealTimeRPI thinks we're 37th.

Either way, that's not really a bubbly RPI, but it's closeish.  It doesn't leave a lot of room for error before you start sweating.  Theoretically, though, if you want to be a 6 seed or better - which I consider the bare minimum cutoff point for being a real, actual contender in the tournament - you would have to be no lower than 24th.  Here's the thing: we easily could be, and it doesn't have anything to do with playing any better.

I took the liberty of calculating our RPI myself, so I could then futz with it some.  The other thing I took was some shortcuts, otherwise this post wouldn't be here til 7 AM tomorrow.  It's late as it is.  But I got a workable estimate, and the conclusions remain valid, as the estimate is actually somewhat conservative.  For brevity's sake I'll sometimes refer to RPI, which is actual, and eRPI, which is my estimate.  ESPN's RPI calculation for UVA is .5967; my estimate is .5908, so it's workable.  (For the record, I know exactly why my estimate is different, too, which lets me at least draw better conclusions.)

OK, so.  You look at our strength of schedule and it's not so hot.  112th to ESPN, 98th to RTR; either way, if you took the top 50 RPI teams, it's well in the bottom quartile.  This is due to two things: a general lack of marquee nonconference teams (outside of Michigan which is really doing us proud) and the presence of some really assy teams at the bottom.  It would look even worse if Drexel hadn't gone on a friggin' tear since New Year's.  Seriously: you rock, Drexel.  Big help.  But I have two major recommendations here, that would set this team up for much better success without actually having to be any better.

1) Don't schedule multiple teams from the MEAC.

Or any other ridiculously bad conference, but really, the MEAC is the main culprit.  Here's the thing: opponents' opponents is one-fourth of the RPI calculation.  I actually went through and tallied them all: there are 726.  Yes, many, many teams are counted twice or more, and that's the thing: of all the conferences in the country, the MEAC has the third-heaviest weight in that section of our RPI, behind the ACC (duh) and the CAA (we played three CAA teams.)

This is because we played both UMES and South Carolina State.  So every team they've played, their record counts twice.  If they both play a team twice, it shows up four times.  Some teams have even more of an impact than that, because other ACC teams play them too.  Of the 726 elements of the opponents' opponents SOS average, Norfolk State is eight of them.  (Thank God for small miracles; Norfolk State has a biggish winning record.)

But MEAC teams, on balance, are lousy.  Really lousy.  If you could remove all MEAC teams from the opponents' opponents SOS portion of the eRPI, it jumps from .5908 to .5920.  Is that a lot?  In real life it's two spots.

Of course, you can never excise the MEAC completely.  It's a near-perfect geographical overlap with the ACC, so ACC teams will always use that conference as a win farm.  As will we.  And if we dropped them from the schedule they'd just be replaced with another team from a lousy conference like the Big South.

Still: I estimate - conservatively, mind you - that if we'd played any other team but a MEAC one, in place of one of the ones we did play, we'd be at least one spot higher in the RPI.  Just by virtue of scheduling, like, Radford instead of UMES.  Schedule, say, Georgia instead, and now you're really talking.  I know there are other scheduling considerations besides wanting an extra spot in the RPI, but keep in mind that the conference's share of tournament dough (and thus, the team's) depends on how far the conference's teams advance.  Is it worth that extra spot if it means a difference in seeding?  Possibly.

2) Don't schedule the complete disasters.

Here's where the money is made, so to speak.  It's a common lament among those paying attention that there are three teams on the schedule (Towson, SC State, UMES) whose combined record is 12-71.  Towson is freakin' one and twenty-nine!  It's even worse than it looks, because the bottom-of-the-barrel teams like to puff up their record by playing D-II squads, and those games don't count in the RPI.  So those three teams are really 11-71 for RPI purposes.  Those games against Longwood we like so much?  Longwood is 10-18, but for RPI purposes, they're 6-18, with four wins against non-D-I schools.

Let's say instead of Towson and SC State, we had Northeastern and Coppin State.  Both were sub-mediocre instead of completely rotten last year; both are sub-mediocre instead of completely rotten this year.  In other words, it's possible to foresee who will suck and who won't.  Replace the two worst teams on the schedule with teams we're not really that much less likely to beat, and the result is: eRPI jumps from .5908 to .6052.  That's a real-life change of fifteen spots.  Anyone who thinks we wouldn't beat Coppin State and Northeastern instead: please.

Fifteen spots!  I know there's a lot more to tournament seeding than just the RPI, but still: do you realize that's pretty much the difference between a 50-50 shot to advance to the second round (and then be slaughtered by a #1 seed) and being seeded for the Sweet 16?  How could anyone not want to do that?

I also realize that setting up a basketball schedule is the art of keeping 16 eggs in the air at once.  But dang, man.  I'm not even calling for a home-and-home with Kentucky here.  This is just scheduling for success.

********************************************

The recruiting board has gone a little while without an update, so here's one for you:

-- Added LB Dorian O'Daniel to yellow.  I didn't add him at first because he could really just pick any school he wants.  But at least he's giving UVA a sniff.  We'll see if there's any staying power here.

-- Added LB Doug Randolph and OL Parker Osterloh to green.  Randolph is technically committed to Stanford, so why green?  Because he gave that commitment to Jim Harbaugh, who happens to coach the 49ers, not Stanford.

-- Moved WR Andre Levrone and LB Buddy Brown from yellow to green.

-- Removed LB Holland Fisher from red (VT commit.)

There was a junior day last weekend and there'll be another one this weekend.  Junior-day season is the time to find out who's really serious about whom.  However, because the class size is limited as compared to the last two, don't expect things to move fast and furious this spring.  It's not necessarily worth it any more to pick the low-hanging fruit unless that's someone you'd take regardless of anything.  We're now recruiting from a position of strength - the "up-and-coming" line is coming true before everyone's eyes - so it doesn't make sense to fill up on the mid-three-stars in March.

3 comments:

Deane said...

It would be nice to get a post about basketball recruiting. I'm not in the know, but I haven't heard a thing about the class of '13 except that we're pursuing Sina (who looks like someone who would be a great addition, especially considering how thin we are at PG). Who else are we looking at? Given our abundance of scholarships we'd better be in the mix for several players.

Anonymous said...

I appreciate your thoughts on RPI. It seems the rule of thumb when putting cupcakes on the schedule is "schedule the best teams that you will nevertheless beat". (Any worse than that is just a waste.)

One other thing that has long bothered me about RPI is that beating a bad team can actually be worse for your RPI than *not playing anyone at all*. How can that possibly make any sense whatsoever?

Team A went out and beat Little Sisters of the Poor. Team B stayed home and played Xbox. Therefore Team B is better than Team A. Say what??

Brendan said...

I think the problem with that view is the conclusion that Team B is better than Team A. It leaves out too much information. Team B will have to play another game eventually, and if they don't play Little Sisters of the Poor, then they'll be deservingly better. (Assuming they win.)

Don't think of RPI as a daily power ranking, that's what the polls are for. It wasn't designed to do that. Think of it more like an unfinished painting. People who track the daily ups and downs are barking up the wrong tree. It's best viewed from the standpoint of the whole, not the daily fluctuations.

Re: hoops recruiting - I'll see if there's enough free info floating around for a post, but no promises. There are a lot of reasons I don't really cover it and that there's no hoops recruiting board (and never will be.) It's very fluid; it's often skeezy (not with Tony Bennett, but still); a guy could be the next Ralph Sampson one day and get a Duke offer the next and then it's over. And unlike football where it's very easy to separate the free info from the pay info, most hoops stuff is proprietary, because of how fluid it can be. In football, if info is a month old, it's probably still good. Not so in basketball. Here are a couple names you can familiarize yourself with for starters:

- Jaren Sina (2013)
- Stanford Robinson (2013)
- Matt Thomas (2013)
- Anthony Barber (2013)
- Brandan Stith (2013/14)
- Devon Hall (2014)