2021 RPI and NCAA Tournament Bracket

I completely agree: It makes absolutely no sense that Brown would rise to 7 spots in RPI--inside the top 20!--after beating a mediocre Penn team. One can only attribute it to the mathematical weirdness of the RPI--and there is /most definitely/ some mathematical weirdness in the RPI, though don't ask me to explain it. Chris Henderson has projected both Brown and Harvard as 3 seeds, but that seems crazy to me based on the eye test/schedule, every assessment BUT RPI test. He's got NCAA bubble teams who have had very good seasons--VCU, to name one, but there are others--missing the tourney because they have no RPI top 50 wins--yet as you note the only RPI top 50 win Brown has is against Harvard. Harvard, as best I can tell, has ZERO top 5 RPI wins. I invite anybody to scan Harvard's schedule/results and explain why it would deserve a national seed? Harvard hasn't beaten anybody of note, unless I've missed something.

 

Exactly my point. If you’re already in the Top 20, logically it should take a big win to send you up 7 positions. Beating Penn (RPI somewhere around 120) shouldn’t do that. For some reason (opponent winning percentages and opponent’s opponents winning percentages?) the RPI appears to be over valuing the Ivy League. Which is why the committee doesn’t solely rely on RPI for selection and seeding. As you mentioned, VCU was another team in the Top 25 of the RPI most of the season despite zero meaningful wins. You would think an eye test would disqualify Harvard and Brown from being in contention for a seed.

To put it into perspective, Clemson and Wake Forest scored massive road victories this weekend over the #11 and #1 RPI teams, but barely moved up in the RPI any more than Brown did with a win over Penn.

There’s a simple mathematical explanation for all of this I know. I’m aware how the RPI functions. Just pointing out what would appear to be a limitation/blind spot in the RPI. And I’m not saying Harvard and Brown are bad teams by any means. Just that their RPI ranking and their resumes don’t seem to add up.

 

NCAA doesn't publish the raw numbers, just the resulting positions. If there were a situation where the RPI scores were clustered closely together for a large number of teams, it is possible to see a sudden jump as a result of a good result while the others were having bad results over a week.

 

I apologize: Henderson has projected Brown and //Princeton// as 3 seeds. Not Brown and Harvard. There was a 3 between Princeton and Harvard on his list of likely NCAA qualifiers, and I confused Harvard with Princeton. My mistake--dumb.

Princeton does have an impressive victory over Rutgers to its credit, and that's a big credit, and it tied Georgetown. It has two losses--to Hofstra and Brown. For what's it worth, Massey has Brown ranked #30 and Princeton at #39.

 

RPI is a formula, but the way people use it can be misleading. It is thrown out as everything from the best measure of a team to one of many factors in breaking ties when it comes to at large bids.

 

I just learned something new.

And, as I expected, there is a tight cluster around Brown's position.

 

Fantastic. Thank you. Fascinating that Harvard wouldn’t even get into the tournament on one set of criteria bit would be considered for a seed on another. At least I know I’m not the only one wondering what is going on here. Will be interesting to see what the committee decides to do with them. There are lots of bubble teams with resumes that seem at least arguably more deserving.

 

Here's Harvard schedule this year. It's had a good season, no doubt--but where are the victories that merit a No. 12 RPI ranking? I don't want to knock the Ivy League, but I guess I will: what would Harvard's W-L record be in a P5 conference? I just don't get the 12 RPI...at all. It's got a better RPI than UCLA, Santa Clara, other big dogs. I'm not saying the Crimson isn't a good team--it may be very good for all I know, haven't seen the team play--but where are the wins that earn it such a lofty RPI?
https://gocrimson.com/sports/womens-soccer/schedule/2021

 

There is a reason basketball stopped using the RPI as a metric. The RPI is easily manipulated. The Ivy's have been queens of it (there is some natural manipulation that can happen because of their reduced schedule). It has been discussed in the past. Head Coaches now have Assistants do deep dives into RPI manipulation and very carefully schedule to try to "artificially" boost their RPI's or keep them in a proper range for an at-large bid. Wake Forest did this to a certain extent (completely understandable in my view by the way). They played an absolutely abysmal non-conference schedule to make sure they would have a good record knowing they could ride their ACC schedule to a high RPI.

The RPI also doesn't account for performance. Last night, Santa Clara demolished league foe and winless LMU 6-0. Outshot them 22-2. Did everything that you would expect a top team to do against a winless foe. The RPI didn't care as, according to a tweet by Chris Henderson, their RPI tanked from 28 to 37. They had to play them. They beat them 6-0. The RPI isn't interested. This maybe more than anything else is it's biggest flaw.

A lower ACC team could play a FSU, UNC, UVA, or Duke lose 20-0 and their RPI would still increase.

 

Yes, the RPI is most definitely weird--and the Santa Clara game/RPI drop is a perfect example. Tennessee earlier in the week beat a weak Florida team (RPI of 114) and UT's RPI fell from 11 to 12. Fine. Pepperdine beats an equally weak Pacific team (RPI of 124--10 spots lower than Florida) and Pepperdine's RPI improved significanly--/rising/ from 17 to 12 or 13 (NCAA). Curious.

 

Lot of teams didn’t like seeing St. John’s beat Xavier last night. St. John’s RPI jumped 15 levels from 58 to 43. They are firmly in the at-large conversation even if they don’t take the auto bid.

 

No

And yet Xavier's RPI dropped by only 1. We have teams that win and yet drop multiple spots because the opponent was weak and yet a No 10 RPI team loses to the No. 58 RPI team and essentially stays the same. Wha??? Compare Georgetown's record/schedule to Harvard's and explain how Harvard has a higher ranking?

Who is on the selection committee--anybody know? I wonder how many games each committee member has watched, on average--and how many different teams. I think it would be incumbent on every member to watch a lot of games for some eye-test comparison to RPI.

Despite the anomalies, the current RPI ranking does seem to fairly accurately reflect the strongest and weakest teams, except that the 3 Ivies in the top 15 might be getting generous mathematical treatment. But we'll see--they could prove me wrong!

 

I'll agree that Wake had an abysmal non-conference schedule but to be fair, they were respectable in the ACC and did not "ride."

They were respectable against Virginia, beat Virginia Tech, lost to Duke in the regular season but beat them in the first round of the ACC, and gave Notre Dame a good game outshooting them 22-11 in South Bend. The Irish needed a late goal to win the game. That said, Wake had a bad loss to a bad Louisville team and it took them 2 OT periods to beat a bad Miami side. Tonight's ACC semifinal vs FSU will give a good indication of how good Wake is.

Many expected Wake to get taken to the woodshed when the conference schedule started and that simply didn't happen. Da Luz deserves some credit for getting the team to a respectable level this season after some pretty mediocre seasons starting with the 2018 season Wake also benefitted from Covid by having some 5th-year players and a grad transfer that played significant minutes.

 

UVA finished top of the ACC regular season and on RPI, they are the 4th team in the ACC, below the team that finished 7th. Lost one game all year to the #20 RPI team Penn State.

 

I wasn’t disparaging Wake Forest’s performance in the ACC at all. I was discussing that because of the way the RPI works it is not necessary for ACC teams to schedule tough non-conference schedules to get a good RPI. They did ride their ACC schedule to a good RPI, and, by that, I mean you won’t find a situation like Santa Clara where a league win tanks their RPI.

By the way, my statement was “they knew they could ride” which implies I was discussing when they set up their non-conference schedule not based on looking at results. I also said I understood their scheduling choices. I think Wake Forest is a really good program who coaching staff do a good job.

 

The eye test definitely should not be part of the criteria. It would be far too subjective.

I prefer to watch matches that feature a lot of passing, are more defensive in nature, take place primarily in the midfield, and are low scoring and close. Someone else might prefer a more direct match that primarily takes place up and down the flanks and is high scoring. Each evaluator would have a tendency to reward teams that play their preferred style well. As much as they should keep bias out of the evaluation, it would be nearly impossible to do so.

 

How does UVA not even enter the discussion for a#1 seed if they win the ACC tournament?

 

Forget winning it. They should be a LOCK now from just making the final. Sometimes, it yields anomalies. It is what it is

 

Define weaker? Im sure there is a mathematical formula that spits it out. Dont think it has anything to do with how the game projects to play out based on who is on the field etc. Also head to head has no impact im sure.