2022 NCAA Regional Rankings

[PaulNewman]

FW, excellent as always...

So it sounds like we don't know how the cmte will treat ranked draws, so in turn all of us amateur prognosticators should be cautious in our predictions about who's in and who's out.  I have been operating under the impression of past years when at least one ranked win was virtually a necessity for a team with otherwise very strong credentials and often two ranked wins seemed like a minimum...and then three ranked wins (and four and above) put teams in pretty good shape as long as the record was good to very strong and the SoS wasn't so low as to be disqualifying.  The increased number of draws (in general and ranked) may add a level of complexity that keeps us in the dark (until we learn more about how the draws will be treated).  Will the cmte consider the number of ranked draws in some way, like 3 draws equals 1 or 1.5 wins?  And/or will they look at the quality of draws? 

Strongly agree with your proposal.  Taking Kenyon, for example, I certainly place a fair amount of blame on the program for perennially having lower than expected and sometimes borderline too low SoSs, as it has happened far too often to chalk it up solely to perennial bad luck.  I think the Owls will end the weekend up a little, probably around .545, but they should do better than that.  They obviously can't control for a conference foe like Hiram going 1-15 and Oberlin 5-7-3, but they don't have to schedule Muskingum who is not a surprise to finish at 2-14.  Spalding also is a choice for 3-9-6.  And then the totally unexpected calamity that was Centre at 2-12-2.  Your example of Bowdoin nicely showed how impact a couple of bad records can have on an otherwise strong to very strong schedule.

[Gregory Sager]

Perfect example...Bowdoin.  I still think they are safe but if I'm calculating correctly Bowdoin only has one ranked win.  They could be restored to two if Conn makes the final rankings.  But think about it.  The Polar Bears just won the NESCAC, accepted as the most competitive conference in the country and they are short on ranked wins.

Well, with more ties overall due to no regular season OT, it's logical that RvR's are going to contain more ties and less wins.  It will be interesting to see the RvR's of the teams thought to be the last one's selected for at-large berths versus the RvR's for the first ones left out to see if it sheds any light on how the committee evaluated and rewarded ties versus wins, and how many RvR wins seemed to almost guarantee selection.

You are correct that Bowdoin's RvR for the fourth rankings contains just one win at 1-1-3 with Conn Col. dropping out of the rankings.  But you could look at it this way: they only lost once in five games against ranked opponents.  Middlebury is also down to 1 RvR win at the moment for the same reason. 

What hurts Bowdoin's resume is that they don't leave Maine in their non-conference schedule which ensures their SOS won't be as high as most others in the NESCAC (only Colby--the other Maine school--has a lower SOS).  Playing Univ. of New England (0-16-2) especially killed their SOS this season.  How much it hurts their SOS is, IMO, unfair.  So indulge me as I explain a solution for this perceived unfairness.

It takes too many good opponents to compensate for that one absolutely bad opponent which is why I continue to say that the SOS formula (which is only relevant for teams in the running for a place in the NCAA tournament) should have a bottom threshold for Opponent's Winning Percentage (OWP).  For example, let's make .375 the bottom threshold.  This would mean that any opponent's winning percentage (OWP) less than .375 is taken as .375 for the purposes of calculating a team's SOS.  This would base the SOS much more on how good the good teams you played were, and less on how bad the bad teams you played were.  To illustrate, two borderline Top 25 teams and contenders for a berth in the NCAA Tournament could play nearly identical 18-game schedules except for two games.  If not for those two games, they both would have a very good SOS of .600.  In those two games that differed, however, Team A's opponents had records of 5-9-4 and 3-12-3 while Team B's opponents had records 4-13-1 and 1-15-2.  None of those games should have been challenging for tournament-calibre teams and more importantly none of those games should be considered relevant to whether a team is worthy of an at-large berth.  Under the current SOS calculation, those two games lower Team A's SOS from .600 to .577 while Team B's SOS decreases from .600 to .566, giving Team A a not-insignificant 11 point SOS advantage all because they played a 5-9-4 team instead of a 1-15-2 team.  The difference between Team A and Team B garnering one of the last at-large berths could be a result of Team A getting lucky that their worst opponent wasn't as bad as Team B's worst opponent.  With a .375 bottom threshold on OWP, the SOS's would come out .581 and .580, respectively, making how bad their worst opponents were virtually irrelevant as it should be, IMHO.

Returning from the hypothetical to Bowdoin's reality, if you remove their game against Univ. of New England, their SOS would be a much more competitive .583 instead of the vastly lower .560 with that game included.  That one game cost them 23 pct. points!!!  Remove the Maine Maritime (5-10-1) game as well and their SOS would be an very good .596, meaning their two weakest opponents cost them 36 pct. points!  But you can't completely ignore that they played weak opponents giving them easy wins to boost their winning pct.  With a .375 bottom threshold on OWP applied to those games, Bowdoin's SOS would be .573 which to me feels much fairer.

The NCAA's SOS will always be flawed because of it's simplicity that results in a game against Husson helping your SOS almost as much as playing Middlebury and more than playing Cortland State.  Not sure how you can remedy that (high win pcts. that disguise a mediocre to weak opponent) with a simple tweak to the formula, but my proposal would at least solve the problem at the other end of the win pct. scale in which the formula makes a irrelevant distinction between how bad teams' very worst opponents are and allows that to impact a team's chances of being selected.

The NCAA's own Regional Rankings point to how flawed their SOS formula is.  The SOS formula says that Husson's contribution to Bowdoin's SOS is .697 while Middlebury's is barely higher at .699 while Tufts' contribution is .644 and Williams' is .617.  If Husson is considered a tougher opponent then Tufts and Williams and just a hair less tough an opponent as Middlebury, they why weren't they ranked 3rd or 4th in Region I. Instead they correctly probably didn't even get a sniff of the regional rankings. The rankings tell us that Husson isn't as good (probably not even close) as most NESCAC teams while the SOS formula that is used in the ranking criteria says that Hussin is a tougher opponent than most of the NESCAC.

Other D3 sports compensate for this, as well as make their SOS ratings more granular, by creating SOS out of a formula consisting of two-thirds opponents' winning percentages and one-third opponents' opponents' winning percentages (excluding outcomes against the team in question).

[PaulNewman]

And then there's the irony of opponents markedly benefitting off a team's stellar record but not the team with the stellar record.

[Flying Weasel]

Perfect example...Bowdoin.  I still think they are safe but if I'm calculating correctly Bowdoin only has one ranked win.  They could be restored to two if Conn makes the final rankings.  But think about it.  The Polar Bears just won the NESCAC, accepted as the most competitive conference in the country and they are short on ranked wins.

Well, with more ties overall due to no regular season OT, it's logical that RvR's are going to contain more ties and less wins.  It will be interesting to see the RvR's of the teams thought to be the last one's selected for at-large berths versus the RvR's for the first ones left out to see if it sheds any light on how the committee evaluated and rewarded ties versus wins, and how many RvR wins seemed to almost guarantee selection.

You are correct that Bowdoin's RvR for the fourth rankings contains just one win at 1-1-3 with Conn Col. dropping out of the rankings.  But you could look at it this way: they only lost once in five games against ranked opponents.  Middlebury is also down to 1 RvR win at the moment for the same reason. 

What hurts Bowdoin's resume is that they don't leave Maine in their non-conference schedule which ensures their SOS won't be as high as most others in the NESCAC (only Colby--the other Maine school--has a lower SOS).  Playing Univ. of New England (0-16-2) especially killed their SOS this season.  How much it hurts their SOS is, IMO, unfair.  So indulge me as I explain a solution for this perceived unfairness.

It takes too many good opponents to compensate for that one absolutely bad opponent which is why I continue to say that the SOS formula (which is only relevant for teams in the running for a place in the NCAA tournament) should have a bottom threshold for Opponent's Winning Percentage (OWP).  For example, let's make .375 the bottom threshold.  This would mean that any opponent's winning percentage (OWP) less than .375 is taken as .375 for the purposes of calculating a team's SOS.  This would base the SOS much more on how good the good teams you played were, and less on how bad the bad teams you played were.  To illustrate, two borderline Top 25 teams and contenders for a berth in the NCAA Tournament could play nearly identical 18-game schedules except for two games.  If not for those two games, they both would have a very good SOS of .600.  In those two games that differed, however, Team A's opponents had records of 5-9-4 and 3-12-3 while Team B's opponents had records 4-13-1 and 1-15-2.  None of those games should have been challenging for tournament-calibre teams and more importantly none of those games should be considered relevant to whether a team is worthy of an at-large berth.  Under the current SOS calculation, those two games lower Team A's SOS from .600 to .577 while Team B's SOS decreases from .600 to .566, giving Team A a not-insignificant 11 point SOS advantage all because they played a 5-9-4 team instead of a 1-15-2 team.  The difference between Team A and Team B garnering one of the last at-large berths could be a result of Team A getting lucky that their worst opponent wasn't as bad as Team B's worst opponent.  With a .375 bottom threshold on OWP, the SOS's would come out .581 and .580, respectively, making how bad their worst opponents were virtually irrelevant as it should be, IMHO.

Returning from the hypothetical to Bowdoin's reality, if you remove their game against Univ. of New England, their SOS would be a much more competitive .583 instead of the vastly lower .560 with that game included.  That one game cost them 23 pct. points!!!  Remove the Maine Maritime (5-10-1) game as well and their SOS would be an very good .596, meaning their two weakest opponents cost them 36 pct. points!  But you can't completely ignore that they played weak opponents giving them easy wins to boost their winning pct.  With a .375 bottom threshold on OWP applied to those games, Bowdoin's SOS would be .573 which to me feels much fairer.

The NCAA's SOS will always be flawed because of it's simplicity that results in a game against Husson helping your SOS almost as much as playing Middlebury and more than playing Cortland State.  Not sure how you can remedy that (high win pcts. that disguise a mediocre to weak opponent) with a simple tweak to the formula, but my proposal would at least solve the problem at the other end of the win pct. scale in which the formula makes a irrelevant distinction between how bad teams' very worst opponents are and allows that to impact a team's chances of being selected.

The NCAA's own Regional Rankings point to how flawed their SOS formula is.  The SOS formula says that Husson's contribution to Bowdoin's SOS is .697 while Middlebury's is barely higher at .699 while Tufts' contribution is .644 and Williams' is .617.  If Husson is considered a tougher opponent then Tufts and Williams and just a hair less tough an opponent as Middlebury, they why weren't they ranked 3rd or 4th in Region I. Instead they correctly probably didn't even get a sniff of the regional rankings. The rankings tell us that Husson isn't as good (probably not even close) as most NESCAC teams while the SOS formula that is used in the ranking criteria says that Hussin is a tougher opponent than most of the NESCAC.

Other D3 sports compensate for this, as well as make their SOS ratings more granular, by creating SOS out of a formula consisting of two-thirds opponents' winning percentages and one-third opponents' opponents' winning percentages (excluding outcomes against the team in question).

It's the same for soccer.  2/3 OWP + 1/3 OOWP = SOS

My hypothetical illustration assumed a realistic OOWP for weak teams from bottom half conferences in deriving my numbers.  And the Bowdoin examples use the actual OWP with head-to-head removed and the actual OOWP but without any games being removed (the manual no longer mentions that for OOWP like it used to IIRC).  Either way, my numbers should be within a couple points.  Now, Bowdoin's SOS did (unfairly?) benefit from playing Husson, so maybe sometimes things even out.

[PaulNewman]

Brockport....darkhorse Pool C?  9-3-7...RvR of 2-1-1 (beat Oneonta AND Cortland)...and SoS is probably gonna jump from .546 to .565-.570 range after playing Cortland and Oneonta back to back.

[4samuy]

Not a bad call. Could get them above skidmore and below Vassar in regional rankings, which would put them third in pool c consideration in the region.  I noticed they were not regionally ranked in the most recent rankings, but it's not unprecedented, at this time of year for teams to make a jump based on results coming down the stretch in their conference tournaments. If they can get to third in pool c rankings they will get on the board and it will be up to the committee on what to do with vassar imo.

[Yankeesoccerdad]

I'm not complaining and arguing that it should be different but the criteria and selection process is challenging.

Perfect example...Bowdoin.  I still think they are safe but if I'm calculating correctly Bowdoin only has one ranked win.  They could be restored to two if Conn makes the final rankings.  But think about it.  The Polar Bears just won the NESCAC, accepted as the most competitive conference in the country and they are short on ranked wins.

And looking at Hamilton, Vassar, and Skidmore...Hamilton now down to two wins (unless Conn comes back in), Vassar has only one ranked win and it's New Paltz down near bottom of Reg 3, and then Skidmore only has one which is Hobart clinging to the very last Reg 3 spot.  Hamilton's two wins appear far better (Midd and Tufts).  Vassar and Skidmore both needed a win today.  SLU I think is safe, but Vassar and Skidmore are on very shaky ground (unless of course Skid gets the AQ).

Does this mean a Conn win Sunday is better for Bowdoin?  I thought Bowdoin was certain to get a bid no matter what but this discussion has made me wonder if I am overconfident.

[College Soccer Observer]

So here are my unscientific thoughts based on last week's rankings. 
Region 1--None of top 7 have auto bids yet. Amherst can get one tomorrow. 
Region 2--Babson has auto bid, Endicott very slim chance at Pool C, everyone else is done. 
Region 3--Oneonta has auto bid, either St. Lawrence or Skidmore will get an auto bid, Hobart and New Paltz done, Vassar and Cortland St along with St. Lawrence/Skidmore loser in the conversation.
Region 4--Stevens and Rowan have auto bids, Montclair St. and NYU in the mix for Pool C, Lancaster Bible and Misericordia done
Region 5--Messiah and Johns Hopkins have auto bids, F&M and Catholic in the mix, Drew, Alvernia, Gettysburg done.
Region 6--W&L auto bid, either CNU or Mary Washington with auto bid, the other and Lynchburg in the mix, VA Wesleyan, Sewanee, Covenant done
Region 7--John Carroll, Ohio Wesleyan, Rose Hulman have auto bids, Kenyon, Ohio Northern, Case Western in the mix, Carnegie Mellon done
Region 8--Chicago, North Central, Calvin auto bids.  North Park Pool C.  MSOE, Wheaton, Carthage all done.
Region 9--UW Eau Claire, St. Olaf, Luther auto bids.  Gustavus Adolphus Pool C.  UW Platteville, UW Whitewater, Wartburg all done.
Region 10--Willamette, Mary Hardin Baylor, either St. Thomas or Trinity have auto bids.  Pacific Lutheran and St. Thomas Trinity loser in the conversation.  Whitman and Southwestern done.

Pool C Candidates:
Region 1 (7) Middlebury, Amherst if they lose, Bowdoin, Tufts, Hamilton, Williams, maybe Conn if they do not win
Region 2 (1) Endicott
Region 3 (3) Cortland St, St. Lawrence/Skidmore loser, Vassar
Region 4 (2)  Montclair St., NYU
Region 5 (2) Franklin and Marshall, Catholic
Region 6 (2)--CNU/Mary Washington loser, Lynchburg
Region 7 (3)--Kenyon, Ohio Northern, Case Western
Region 8 (1)--North Park
Region 9 (1)--Gustavus Adolphus
Region 10 (2)--Pacific Lutheran and Trinity/St. Thomas loser

That is a total of 24 teams for 20 spots.

[paclassic89]

The conference central/NCAA tourney page incorrectly has JHU as the champion already.  They still need to beat Muhlenberg tomorrow

[College Soccer Observer]

I stand corrected on Hopkins.  If they lose, 25 teams for 20 spots.

[PaulNewman]

Good work by CSO...

Stay tuned for a SC productions release later today/evening...

My possible divergencies from CSO's work above.

I think the only way NESCAC gets 6 is if Conn wins today..and could only get five even in that scenario.

Given where they were this week in the RR I agree with the sentiment that W CT is out...but I will not be absolutely shocked to hear their name called.  Their resume stacks up with some other candidates but there's no obvious or even less obvious way to see how they jump the teams ahead of them in Reg 1.  All that said, compare resume with Pac Lutheran.

I think Endicott is out with the caveat that this means not a single at large will come from Reg 2.  Too bad W CT isn't in Reg 2.

I'll be shocked if Vassar and Skidmore both get in (if Skid fails today), and I won't be shocked if neither do.

Catholic has a good shot and would guess on right side of bubble.  Also don't think Drew is totally out of the question when looking at the resume blindly against some others.

Gut tells me Lynchburg gets in but very much on the bubble and the Hornets are sweating.

I don't think CMU is out.  Record and SoS good.  Only one ranked win but that's true of the majority of bubble teams.  May come down to Reg 7 not going beyond ONU and CWRU.

Imo UW-Platteville and Wartburg are still very much in play (compare Wartburg to like a Williams or Tufts...very similar except for somewhat lower SoS).

My gut tells me yes on Pac Lutheran but if St Thomas falters against Trinity today (which I probably expect), then the cmte will have to decide between the two or obviously take both. 

And I am gonna add Brockport to the mix...very competitive resume with some others and two ranked wins are impressive ranked wins.

[PaulNewman]

Disclaimer....I estimated updates to SoS for teams that played in past week...

Locks to Highly Likely (alphabetical)....Bowdoin, Case Western, Cortland St, F&M, Gustavus Adolphus, Kenyon, Middlebury, Montclair St, North Park, NYU, Ohio Northern, Christopher Newport/Mary Washington loser

**** Amherst, Hopkins, SLU take bids if lose today, and St Thomas (TX) joins bubble

EIGHT SPOTS LEFT (and maybe less than eight)

Brockport St – 9-3-7 // .565 // RvR 2-2-1 (ranked wins over Oneonta and Cortland, ranked draw Oneonta)
Carnegie Mellon – 11-4-1 // .585 // RvR 1-3-1
Catholic – 11-3-4 // .580 // RvR 1-1-2
Conn Coll – 7-5-5 // .615 // RvR *2-3-5
Drew – 12-2-4 // .565 // RvR 1-1-1
Endicott – 12-5-4 // .550 // RvR 1-1-1
Hamilton – 8-4-4 // .593 // RvR 2-4-1 (3-4-1 if Conn ranked week 4)
Lynchburg – 11-2-5 // .559 // RvR 1-0-3
Pac Lutheran – 17-3 // .535 // RvR 1-3
Skidmore -- *9-3-6 // .570 // 1-2-5
Trinity (TX) -- *9-4-3 // .560 // RvR 1-2-1
Tufts – 7-2-7 // .587 // RvR 2-1-5
St Thomas (TX) -- *13-1-3 // .520 // RvR 3-1-1
UW-Platteville – 12-4-3 // .565 // RvR 2-4-1
Vassar – 9-3-5 // .600 // RvR 1-3-4
Wartburg – 7-2-9 // .553 // RvR 2-1-4
Western CT – 20-0-2 // .515// RvR 1-0-1
Williams – 6-1-10 // .590 // 2-0-6

* for teams playing today

If you think you're sure who is getting in and who isn't try looking at the resumes without the names attached. 

[4samuy]

Pn

Your thoughts on pac Lutheran resume and western conn can be very much debate IMO.  The advantage pac will have is IMO is they should be up to the plate quicker and longer than western conn. I'm just not sure western conn ever gets to the plate.

[College Soccer Observer]

St. Lawrence up early on Skidmore thanks to 7th minute goal by Sibanda.  Amherst and Conn tied at 1-1 25 min left in 2nd half.

[College Soccer Observer]

Conn and Amherst heading to OT tied at 1.  Saint Lawrence up 2-0 at half on Skidmore.