Post Regional Observations on Selection Sunday

[VObserver]

Just musing about the absolutely fair and impartial process used by the NCAA Softball Selection Committee in assigning at large bids.

15 of the 16 Super Regional participants come from the top 6 RPI Conferences. The 16th is from the 10th rated RPI Conference. So pretty much no one advanced that didn't play in a relatively good conference. So far so good.

Of the 16 Regional winners, Seven [Oklahoma, FSU, Tennessee, Northwestern, Utah, Louisiana and San Diego State] got in on Automatic bids as conference champions. All of them likely would have received at large bids had they not won their respective conferences.

The remaining Regional winners all got in as at large selections, and all except Oregon were host teams. Committee still looking pretty good.

The lowest RPI to advance was San Diego State at 33; the rest of the field at the Super Regionals Ranged from RPI 1 to RPI 20.

All of this makes the Committee look pretty good. [Of course, ESPN and others conveniently overlook home field advantage when they talk about how the committee got it right.]

Conference affiliation of advancing teams is as follows:

PAC 12: 4 teams; Washington, Oregon, Utah and Stanford
Big 12: 3 teams; Oklahoma, Oklahoma State and Texas
SEC: 3 teams; Alabama, Georgia and Tennessee
ACC: 3 teams; Clemson, FSU and Duke
Big 10: 1 team; Northwestern
SBC: 1 team; Louisiana
MWC: 1 team; San Diego State

Honestly, in looking at the total field of 64, especially as regards at large bids, the committee pretty much did get it right. Every team in the field either was an automatic bid winner or had an RPI of 45 or better. The best RPI teams which did NOT receive an at large bid were: South Alabama from the SBC [RPI 42], Penn State from the Big 10 [RPI 44], Ohio State from the Big 10 [RPI 46], Arizona from the PAC 12 [RPI 47]. To get the selections exactly in line with RPI, one would only have to award a bid to South Alabama instead of Notre Dame.

[R1Letterman]

interesting. see also;
percent of teams that advanced per conference

MWC: 100% (1/1)
Big 12: 75% (3/4)
Pac12: 67% (4/6)
ACC: 50% (3/6)
Belt: 50% (1/2)
Big10: 25% (1/4)
S-E-C: 25% (3/12)

[VObserver]

interesting. see also;
percent of teams that advanced per conference

MWC: 100% (1/1)
Big 12: 75% (3/4)
Pac12: 67% (4/6)
ACC: 50% (3/6)
Belt: 50% (1/2)
Big10: 25% (1/4)
S-E-C: 25% (3/12)

Yes, I thought about that as well, but decided not to add it in, as by and large the teams that advanced had finished high in their respective conferences, and the ones who failed to advance generally did not [UCLA notwithstanding].

Also, all those non-advancing conference members met the criteria that people in general [not just the committee] consider as at least minimal for selection to the tournament field: a high RPI and a winning record against good or very good competition.

It is hard to argue that South Alabama with an RPI of 42 and a record of 39-14 should be put into the tournament while Missouri with an RPI of 38 and a record of 34-24 should be left out, especially when you compare SOS.

[CajunVic]

Hey, the SEC finished with par. 25% of the field left and 25% of the SEC teams left.

[SlickRick]

I thought the SEC had 13 teams in the tourney???

[JMV JustMyView]

I thought the SEC had 13 teams in the tourney???

I think it was 1 AQ and 11 At Large.

[CajunVic]

I thought the SEC had 13 teams in the tourney???

Well, if they did, they missed the mark . . .

[VObserver]

I thought the SEC had 13 teams in the tourney???

13 teams play softball in the SEC. Vanderbilt does not field a team. Mississippi State failed to make the tournament.

[R1Letterman]

R. Hosts that have failed to advance last four seasons

Ten total:

By conf:
G5 zero
B12 zero
Big ten 1
ACC 2
Pac 2
Sec 5

[VObserver]

R. Hosts that have failed to advance last four seasons

Ten total:

By conf:
G5 zero
B12 zero
Big ten 1
ACC 2
Pac 2
Sec 5

This is a different, if related, subject. 84.375% of hosts have advanced; not bad. What the elevated numbers for the SEC tell me is that either there is a committee bias toward selecting SEC schools to host, or that the basic criteria the committee has to work from is biased toward large conferences who all generally play reasonably tough non-conference schedules. The RPI is definitely biased in that direction. If every school in the SBC played a top 75 or so non conference SOS, we would "suddenly" become a conference with over half our members in the discussion for at large bids, and perhaps as many as 3 being considered for hosting. [At least until the money folks saw what we are doing and changed the criteria.]

[R1Letterman]

It could also mean that when the others host and lose, they are more stringently looked at for future hosting opportunities, and often denied, whereas the leader in this category is not.

[GoneGolfin]

This is a different, if related, subject. 84.375% of hosts have advanced; not bad. What the elevated numbers for the SEC tell me is that either there is a committee bias toward selecting SEC schools to host, or that the basic criteria the committee has to work from is biased toward large conferences who all generally play reasonably tough non-conference schedules. The RPI is definitely biased in that direction. If every school in the SBC played a top 75 or so non conference SOS, we would "suddenly" become a conference with over half our members in the discussion for at large bids, and perhaps as many as 3 being considered for hosting. [At least until the money folks saw what we are doing and changed the criteria.]

Actually, no ... and this is the paradox of the RPI that folks struggle with. If every team in the Sun Belt played a Top 75 schedule (a proxy for a more difficult schedule), you would see higher individual RPI rankings for the respective teams (due to higher SOS NC schedules) ... most evident immediately before conference play starts ... but this would damage teams' RPIs during conference play because each team's OWP contribution would be lower than it otherwise would be (vs. if the schedules were weaker).

A perfect example of this is Marshall. What Marshall did to themselves by assembling a weak NC schedule helped their conference mates immensely (that played them). In this case, it helped conference opponents considerably more than Marshall playing a Louisiana type non-conference schedule and losing a bunch of games (even though their RPI ranking would be higher by scheduling that way). Similar can be said for Coastal Carolina.

When I was coaching the coaches on how to optimize their RPI, I would also instruct them to look for teams (historically) that schedule weak NC and win ... as opposed to a team that schedules tough (hurting their W/L % accordingly but has a higher RPI) and are better than their record because of it. Obviously your entire NC schedule is not constructed this way ... but a part of it should be (complemented by some number of higher ranked RPI teams) ... while avoiding the teams that are "better" than their record due to a difficult schedule ... or are simply bad period. The mathematical modeling comes down to optimizing risk/reward.

Brian