RPI Overrated!?!?

Sivert - thanks for joining the discussion, and providing very insightful comments. Here's what I would add



Bias? Toward things that work better? For shame...

But to your point. Indeed, RPI does only medium-well in terms of picking winners. However, it is deeper than that. The place where it does the best, not surprisingly, is in the early rounds, where matches are bigger mismatches. There, it actually keeps up with Pablo for the most part. Then again, it's easier to pick mismatches (and the wrong answers tend to be very large upsets that everyone gets wrong). Moreover, with regionalized matchups, relative RPI is a lot better in the early rounds.

Where RPI really falls apart is in the regionals, when matches get a) closer, and b) cross-regional. The difference between RPI and Pablo in the regionals is around 2 - 3 matches per year. Considering that there are only 15 matches, and Pablo is getting about 12 right, that means that RPI is only getting 9 - 10 regional matches right a year. That's significantly poorer results, considering that 7 - 8 would be expected based on chance alone.





One of the features of RPI is that it is biased toward the top of the mid-major conferences and against the bottom of the big conferences. Thus, it already naturally favors bringing in teams from weaker conference at the expense of another team from the big guys. So the mid-major schools are absolutely getting a chance.

Unfortunately, they aren't doing much with it. Know how many mid-major non-conference (regular season) winners win matches in the NCAA tournament? Very, very few. Far fewer than the 6th and 7th place teams from the Big Pac Ten 12 win. So if you look at what teams do when they get the chance, it looks as if the NCAA is giving way too many opportunities to the non-championship schools from the mid-majors and not giving near enough chances to the bottom half schools of the big three. Then again, that's because that's what RPI does.




This is great insight, and spot on the mark. In fact, the NCAA makes no bones about it. One of the objectives of using RPI is to get the teams from the top of the conferences to schedule each other. In fact, in the best of RPI world, the preferred scheduling would be that the conference teams all schedule their counterparts in other conferences. That provides the best RPI outcomes.

The problem with that in volleyball is that there aren't enough weak conferences out west for the top teams to schedule against each other. There are 20 conferences on the east coast, and 15 of them are pretty weak. There are 5 conferences out west, and 4 are above average, and the other not far away. In the era of limited travel budgets, it's easy for teams in the east to schedule very effectively within the RPI model. The west coast teams - not so much. This is basically the origin of the regional bias in volleyball RPI.




Well, it isn't explicitly put into RPI, but this is effectively what the committee does. As much as the committee worships RPI, there is one thing that trumps it, and that is bad non-conference scheduling. If you look at the good RPI teams that have been left out of the tournament (Eastern Wash, Wich St, LSU), they have one thing in common: an awful non-conference schedule, lacking non-conference matches against top 50 teams even. This was why Ball St didn't have a shot last year, even if their RPI were in the top 50, and is why I still fear a little for Michigan St this year (although their Ball St win helps, and they really needed that win over Michigan)

So while the RPI does not put more emphasis on non-conf SOS, the committee does that for it.

1 slight adjustment to that thinking is that if a team were looking to boost their RPI, they wouldn't simply look at another team's RPI.

e.g. Opp A goes 25-5, has an OPP win% of 40 and OPPOPP win% of 40

Opp B goes 18-12, has an OPP win % of 60 and OPPOPP win% of 60

Opp B would have a higher RPI than Opp A (.450 to .408), however, Opp A would have a higher contrigution to your Team's RPI (.517 to .450 if these were the only opposition)

There has been much talk about the RPI having an east coast bias, and I think Bofa did a good job of proving this previously (maybe it was just an East of California bias). Just looking at this year's current RPI (a new one should be out shortly), it seems to have a center of the country bias. The top 6 are all "midwest", and 33 of the top 50 are "midwest" - I am defining midwest as a general boundary from Ohio/Alabama to Minnesota/Nebraska/Texas. I don't know the percentage of schools in D1 in this geographical area, so it may be representative of that, and with the Big 10 and Big 12 getting non-conf opponents from the MAC, MVC, Summit it can make it easier to raise your RPI staying in the middle of the country. However, the next 50 (51-100) has 26 of 50 from outside of my made-up boundary, so 24 of 50 from the center.

That's a fallacy. Or, at least, something that requires a very clear distinction to be made.

The RPI does indeed encourage scheduling games against teams with what I'll call "high RPI value" (think of it like high nutritional value) but THAT IS NOT THE SAME AS HAVING A HIGH RPI.

Lonewolf already addressed this but it's worth spelling out more plainly. As a good general rule, you want to play teams with HIGH WON-LOSS RECORDS but not necessarily high RPIs. The best opponents are the ones who are good enough to dominate weak conferences but aren't so strong that they will likely beat you. That way you still get the "W" for your own won/loss record (which is Element 1 in RPI) while getting the benefit of their great won/loss record (which goes into your Element 2).

There is a simple way to see how an opponent is evaluated by RPI in terms of strength-of-schedule. You can see it for women's soccer on the NC-Soccer website. www.nc-soccer.com/wsoccer/2011/

It is the column called "Approx Strength". When you click on the header twice, it will list the teams in order from highest to lowest - the higher teams being the ones that contribute most to your strength-of-schedule if you play them, and the lower teams the ones that contribute less. I don't know if Rich Kern has this feature on his website -- I didn't notice it when I visited a year ago. If you do it for women's volleyball, I guarantee some of you will be shocked to see which teams are considered a tougher "strength-of-schedule" opponent than others.

You may then notice something else -- that where a team ranks in "Approx Strength" is sometimes nowhere near its own calculated RPI. For example in women's soccer recently Ohio State's ranked #42 in the RPI but its "Approx Strength" was ranked #104.

Now I know lonewolf and Bofa can swoop in here and say something about not being precise with language, but for us commonfolks talking plain words, are we really blockheads for thinking that #42 in RPI has something to do with being the 42nd strongest team or something like it? But then how does it make sense when you're evaluating some other team who plays Ohio State that Ohio State ranks 104th in terms of "strength of schedule"? How can a ranking of how strong you are possibly be different than how strong you are as an opponent on someone's schedule? Let alone vastly different?

THIS IS SIMPLY ILLOGICAL ON THE PART OF THE RPI.

(*the formula for the Approx Strength column is simply .5 x won/loss record, ie. a team's Element 1, + .25 x Element 2, i.e the average of the opponents' won/loss records. When you play some team, call it Y, this is what they will contribute to your RPI. Team Y's Element 1 (times 0.5) contributes directly to your Element 2. And Y's Element 2 (times 0.25) contributes directly to your Element 3)

The 14 teams (from Pac 12 and Big 10) that would dominate Florida State and Northern Illinois as well are: Illinois, Nebraska, Purdue, Penn State, Minnesota, Ohio State, Michigan State, USC, UCLA, Cal, Stanford, Washington, Oregon, and Arizona. We will see in the tourny how sad the other conferences (mostly eastern conferences) are outside of the Big 10, Pac 12, and select teams from others.

Based on pablo:
UNI is ahead of Cal, UW, Oregon, Arizona, Minnesota, Michigan, Ohio State, and Michigan State. FSU is also ahead of all of those except Cal and UW.