November 7, 2019
Women's NCAA "At Large"
Yesterday the NCAA released the most recent, and final before the tournament field is set, Regional Rankings. These were developed by the RAC’s using match results through Sunday November 3. These rankings in the past have been powerful predictors of the final Pool C berths that will be announced next Monday. Pacific Lutheran and Washington U. have earned AQ’s which are awarded to the regular season champion in the NWC and UAA respectively. Forty-one more AQ’s will be awarded this coming Saturday and Sunday as tournament championships are settled
It makes sense that the Week 3 Regional Rankings can tell us a great deal about the final “At Large” berths. Week 3 is based on roughly 90% of the season. Through this past Sunday, D-III women’s teams had played 18 official contests on average. Those teams who have reached their conference tournament, most certainly a group full of potential “At Large” candidates, will have one to three matches added to their Regional Ranking resume.
For the women’s tournament, one “Pool B” and twenty “Pool C” or “At Large” bids will be announced at the Division III Women’s Soccer Selection Show on Monday, November 11 at 1 pm EST.
I will not dig into the NCAA’s selection process. If you need more information, Chris Shirk has written two excellent columns:
Narrowing the Field
We have analyzed eight years of Week 3 Regional Rankings from 2011 to 2018. There have been minor changes in the ranking process during that time but fundamentally the approach is the same.
From Chris’ columns we know the RAC’s use five primary criteria to make their selections along with three secondary criteria. An analysis of selections made over the last several years indicates that three factors matter most: Strength of Schedule (SoS), Win-loss percentage (WLT%) against Division III opponents and Results versus ranked Division III teams (RvR). Analysis of the Results against Ranked Opponents reveals that the key measure is the number of Wins vs Ranked (WvR) rather than overall record or winning percentage.
The first, and possibly most important finding is that with a single exception, every team that received a “Pool C” invitation appeared in the Week 3 Rankings. To be specific, there were a total 161 “Pool C” berths from 2011 through 2018; 160 of those teams were ranked in Week 3. In 2013, Augustana was not among the six Central Region teams in that year’s Week 3 ranking but later earned a Pool C invitation. Let’s explore what unique circumstances likely led to the Vikings making that improbable jump. Understanding the situation behind this exception perhaps may give some glimmer of hope to teams, who are not in line to win their conferences AQ and not in the 2019 Week 3 Regional Rankings.
An important factor is that the Central Region only ranked six teams. Three of those six, Washington U., Wheaton (Ill.) and Aurora, had very impressive “resumes” consisting of high Win% while playing strong schedules and posting superior RvR’s. And the teams in lower three rankings also carried impressive resumes. Augustana, meanwhile, had WLT%, SoS. And RVR numbers that would have earned them a Regional Ranking in any other Region that year and all years studied. Fortunately for them four of the six ahead of them in the Week 3 Rankings earned AQ’s which probably made room to get into the tournament ahead of a team in another region which was ranked there but did not have a profile as strong as Augustana’s.
Filling the Field
The earlier discussion narrows the universe of data points we need to study. From this point forward, all analysis is focused on teams that appeared in Week 3 rankings in the eight years 2011 through 2018.
From 2011 to 2018 the total field was 512 teams (8 * 64). The teams in the field earned their berth:
|
Ranked |
Not Ranked |
Total |
|
|
AQ |
182 |
163 |
345 |
|
Pool B |
2 |
4 |
6 |
|
Pool C |
160 |
1 |
161 |
|
Totals |
344 |
168 |
512 |
Over those eight years a total of 520 teams appeared in the Week 3 Regional Rankings. 182 earned AQ’s and two were Pool B leaving 336 teams which were split as follows: 160 earned Pool C bids while, 176 were not selected. With further analysis can we learn why some did and some didn’t? Well, yes; yes, we can!
Effect of Win-Loss Percentage (WLT%) and Strength of Schedule (SoS)
The graph below charts SoS on the vertical axis against WLT % on the horizontal. The green dots are teams that were awarded Pool C berths, while the red diamonds did not make the cut. As already noted, teams with AQ’s are excluded.
Click to view in full screen
There were 68 teams in our dataset with WLT% equal to or greater than 0.750 and SoS equal to or greater than 0.575 in Week 3; all except two received a Pool C bid.
Those unfortunate sides were Scranton in 2015 and Berry in 2013. Let’s dig into those two exceptions. The statistical line below is WLT% / SoS / WvR / Rank of number of teams ranked in region all from Week 3.
- Scranton 2015
- Week 3 resume: 0.765 / 0.579 / 0 / 8 of 9
- Tournament week 1-1-0, final WLT% 0.732, no additional WvR’s
- SoS of tournament opponents was 0.570 and 0.532
- AQ’s earned by teams ranked 1, 5, 6; Pool C awarded to 2, 3, 4
- To recap, both WLT% and SoS dropped below the magic cutoffs and no WvR’s so missing on all three principal factors and three “At Large” bids were awarded to teams above in their Region.
- Berry 2013
- Week 3 resume: 0.824 / 0.582 / 0 WvR / 7 of 9
- Tournament week 2-1-0, final WLT% 0.800, no additional WvR’s
- SoS of tournament opponents was 0.544, 0.520 and 0.560
- To recap, a strong WLT% but final SoS probably in the 0.560 range and no WvR’s. Christopher Newport, ranked sixth, had nearly identical WLT% and SoS but a 1-2-0 record versus ranked compared to Berry’s 0-3-0 and earned the last Pool C bid in this region.
On the other side of the equation in the “Red Quadrant” there are eight that missed both WLT% and SoS in Week 3 yet still received bids. Two had four Wins vs Ranked teams which we analyze next. One each were in 2017 and 2018 and we learn that those teams had successful conference tournament runs, one win and one draw losing on PK’s in the final, against strong teams which drove up their WLT%, SoS, and added a WvR. All of which together moved them up in the Final Regional Ranking to a point where they earned a Pool C invitation.
Effect of Wins versus Ranked (WvR)
Let’s look at this across all three dimensions which we have to do with two charts. This graph charts Wins vs Ranked on the vertical axis against WLT % along the horizontal.
Click to view in full screen
Forty-six of the forty-seven teams with four or more WvR received “At Large” berths. Demonstrating the importance of this statistic the magic WLT % cutoff appears to drop at least from my somewhat arbitrary 0.750 to at least 0.650.
The sole exception was the 2011 Carnegie Mellon side:
- Week 3 resume: 0.643 / 0.625 / 5 / 3 of 7
- Final WLT% was 0.573 while SoS likely increased to 0.630 range
- Centre 0.775 / 0.560 (Est.) / 2 / 4 of 7 moved ahead during the last week and earned the final Pool C bid in the Great Lakes Region.
This next chart has Wins vs Ranked still on the vertical axis but now plotted against SoS along the horizontal.
Click to view in full screen
For the most part this pattern is similar to the WLT% vs WvR chart except on the five win line Carnegie Mellon has moved to the right side with their traditionally high SoS. Three other outliers jump out in the lower left with extremely low SoS (for Regionally Ranked teams), under 0.500, yet still gaining Pool C bids. On the WLT% vs WvR chart these three are the green dots to the furthest right, very high WLT%, on lines 0, 1 and 2.
Two of the three, Texas-Dallas, 2011 and Texas-Tyler, 2015, were in the West Region and both years, every team above them in the Week 3 Rankings appeared in the NCAA tournament via AQ’s. It seems likely that these two were selected to fill out a pod rather than a team with a stronger resume located much farther away.
The other exception was Virginia Wesleyan in 2012. While the Marlins had the third lowest SoS to receive a Pool C bid (one of eighteen Regional Ranked teams in our eight years of data under 0.500), they also had the third highest WLT% (0.972) and a 2-0-1 record versus ranked. They were most likely one of the last, if not the last team awarded a Pool C bid that year.
Wins versus Ranked Standalone
To finish this investigation of the WvR factor, we isolate WvR alone. This graph looks only at the WvR statistic. We’ve already established that it is a dominant factor when there are four or more wins. Looking at all teams in the sample:As previously noted, forty-six of the forty-seven teams with four or more WvR received “At Large” berths.
At the other extreme, the dataset has 62 teams with no WvR’s. Fourteen of those were AQ’s leaving 48 Pool C candidates, with only eight or 17%, earning bids. Seven of those were classic “Bubble” teams. They typically exceeded either the WLT% or SoS thresholds usually by a good margin. The remaining exception was the previously discussed 2015 Texas-Tyler team.
Conclusions
So, to recap. Here’s the good news:
- Essentially it is a “must” to appear in the Week 3 Rankings. If a team is ranked and they do not win their conference’s AQ, they still have a chance at a ticket to the “Big Dance” via Pool C.
- Teams with a 0.750 WLT % combined with at least a 0.575 SoS should feel very good about their cause. IMPORTANT: These cutoff values are not used by the RAC’s; they are arbitrary values I selected while studying the data patterns.
- WvR matters, a lot; WLT% and SoS stats in the “Green Quadrant” plus three Wins vs Ranked seals the deal. Teams with that profile are almost certain to be in the tournament.
- Four or more WvR’s regardless of WLT% or SoS virtually guarantees a berth.
As for the downside:
- If your team did not appear in the Week 3 rankings it is a near certainty that they must win an AQ bid by winning their conference tournament.
- If your team is in the “Red Quadrant” (WLT % < 0.750, SoS<0.575) and have zero WvR, you better plan on winning your conference tournament if you want to be certain of playing in the NCAA tournament this year.
The “Bubble” typically contains roughly half the teams in the Week 3 Regional Rankings. They move out of the “Bubble” for better or worse, given their results during conference tournament week and the Final Rankings. “Bubble” teams are those:
- That meet only one of the WLT% (>.750), SoS (>.575) or WvR (4 or more) historical cutoffs.
- Also, teams just above or below one or more of the cutoffs in Week 3 could move to the other side of the cut line depending on their results this week.
Some Caveats
Only games through last Sunday (November 3) were considered. Teams near the WLT% or SoS cutoffs may move from one side of the cut to the other and team with 2 or 3 WvR’s could pick up more either by beating a currently ranked team or by a team they previously defeated moving into the Final Rankings and therefore picking up another WvR that way.
I have no "inside information" to incoporate into this analysis, it is simply my interpretation of the data from the last several years.
There are exceptions to every rule. Each year it seems there is a “headscratcher” or two: “Why did XYZ get excluded?” or conversely “What were ‘they” thinking?”
And, as the financial folks say, “Past performance in no guarantee of future results” but unlike the market, the committees appear to be pretty consistent and the line may actually be “When Past Performance Actually Is a Guide to Future Results”.
Wrap Up
Early Sunday evening all the AQ’s will be known, leaving the remaining candidates for Pool C invitations.
We at D3soccer.com will project the Final Week 4 WLT%, SoS, and WvR’s for the Week 3 ranked for teams who did not earn an AQ. Our estimates may differ, SoS is estimated, from the RAC’s actual final numbers, but will be directionally correct and in most cases “close enough”.
Using this analysis, we will make a projection of what we think the field will look like when announced Monday afternoon breaking the candidate teams into categories ranging from “Stone Cold Locks” to “Great Season, See You Next Time”.
While we wait, below are the projections using the analysis above against the Week 3 rankings recently released.
Highly Likely
Region
School
D3 Record
D3 Win %
D3 SoS
RvR
NEW ENGLAND
MIT
17-1-2
0.900
0.591
4-1-2
NEW ENGLAND
Middlebury
12-1-3
0.844
0.592
4-1-3
NEW ENGLAND
Tufts
12-2-2
0.813
0.624
6-2-2
NEW ENGLAND
Williams
9-4-2
0.667
0.632
5-3-2
NEW ENGLAND
Amherst
12-3-1
0.781
0.567
5-3-1
EAST
William Smith
14-1-1
0.906
0.630
5-1-1
MID-ATLANTIC
Messiah
16-1-1
0.917
0.619
6-1-1
MID-ATLANTIC
Dickinson
13-1-4
0.833
0.614
4-1-2
MID-ATLANTIC
Johns Hopkins
12-2-4
0.778
0.662
5-2-3
MID-ATLANTIC
Arcadia
14-2-3
0.816
0.603
5-1-1
MID-ATLANTIC
Stevens
13-3-1
0.794
0.619
2-3-1
MID-ATLANTIC
Gettysburg
12-3-1
0.781
0.593
3-2-0
MID-ATLANTIC
Haverford
12-3-2
0.765
0.582
2-3-1
SOUTH ATLANTIC
Emory
12-4-0
0.750
0.671
4-4-0
GREAT LAKES
Case Western
13-4-0
0.765
0.593
4-4-0
CENTRAL
Washington U.
15-1-1
0.912
0.636
8-1-0
CENTRAL
Wheaton (Ill.)
16-1-1
0.917
0.618
7-1-1
CENTRAL
Chicago
12-2-2
0.813
0.653
6-2-2
NORTH
St. Thomas
14-2-2
0.833
0.598
3-2-1
NORTH
Wartburg
12-2-3
0.794
0.592
3-1-2
NORTH
Augsburg
13-3-1
0.794
0.579
3-1-0
WEST
Pomona-Pitzer
14-1-1
0.906
0.595
5-1-1
WEST
Trinity (Texas)
12-3-0
0.8
0.595
1-2-0
On the Bubble
Region
School
D3 Record
D3 Win %
D3 SoS
RvR
NEW ENGLAND
Wesleyan
9-5-2
0.625
0.569
4-5-1
NEW ENGLAND
WPI
14-4-1
0.763
0.514
2-3-0
NEW ENGLAND
Brandeis
10-7-0
0.588
0.630
1-7-0
EAST
Geneseo State
14-2-1
0.853
0.559
3-2-0
EAST
New York University
11-6-0
0.647
0.633
2-5-0
EAST
Rochester
9-5-2
0.625
0.631
2-5-2
EAST
RIT
11-4-2
0.706
0.600
2-3-1
EAST
St. Lawrence
11-4-0
0.733
0.579
1-3-0
EAST
Vassar
9-7-1
0.559
0.595
2-4-0
MID-ATLANTIC
McDaniel
13-5-0
0.722
0.592
2-4-0
MID-ATLANTIC
Swarthmore
12-5-0
0.706
0.622
2-5-0
MID-ATLANTIC
Susquehanna
14-3-1
0.806
0.513
0-0-1
SOUTH ATLANTIC
TCNJ
14-1-1
0.906
0.574
3-1-0
SOUTH ATLANTIC
Christopher Newport
15-1-3
0.868
0.556
2-1-2
SOUTH ATLANTIC
Centre
18-0-0
1.000
0.523
3-0-0
SOUTH ATLANTIC
Randolph-Macon
18-0-0
1.000
0.527
1-0-0
SOUTH ATLANTIC
Montclair State
12-5-2
0.684
0.592
1-3-0
SOUTH ATLANTIC
Roanoke
13-5-1
0.711
0.603
0-3-0
SOUTH ATLANTIC
Salisbury
13-2-3
0.806
0.532
1-2-1
GREAT LAKES
Ohio Northern
14-2-2
0.833
0.561
3-2-1
GREAT LAKES
Carnegie Mellon
10-5-1
0.656
0.651
4-5-1
GREAT LAKES
Otterbein
14-3-1
0.806
0.550
2-2-1
GREAT LAKES
Wooster
12-2-3
0.794
0.528
1-1-2
GREAT LAKES
Denison
8-6-4
0.556
0.615
2-4-3
CENTRAL
Illinois Wesleyan
12-5-0
0.706
0.583
3-4-0
CENTRAL
Concordia (Wis.)
15-4-0
0.789
0.479
1-1-0
NORTH
Saint Benedict
10-5-2
0.647
0.596
3-3-0
NORTH
Loras
13-4-0
0.765
0.556
1-4-0
WEST
Cal Lutheran
10-2-3
0.767
0.562
2-1-3
WEST
Claremont-M-S
9-4-3
0.656
0.593
3-3-1
WEST
Hardin-Simmons
13-2-1
0.844
0.558
0-2-1
Quite Unlikely
Region
School
D3 Record
D3 Win %
D3 SoS
RvR
NEW ENGLAND
Hamilton
10-5-1
0.656
0.571
2-5-1
NEW ENGLAND
Connecticut College
10-5-1
0.656
0.574
2-5-1
NEW ENGLAND
Springfield
11-5-2
0.667
0.547
1-3-1
NEW ENGLAND
Emerson
10-6-2
0.611
0.550
0-4-0
GREAT LAKES
Capital
14-5-0
0.737
0.563
2-3-0
GREAT LAKES
DePauw
11-5-2
0.667
0.557
1-2-2
CENTRAL
Adrian
12-4-2
0.722
0.537
0-3-2
NORTH
UW-Stevens Point
11-4-2
0.706
0.562
2-3-0
NORTH
Dubuque
12-4-2
0.722
0.571
1-3-0
NORTH
UW-Eau Claire
14-5-0
0.737
0.538
3-3-0
WEST
Chapman
11-3-3
0.735
0.532
2-3-1
Good luck, good weather, good health and great soccer this week to all.
Comments or feedback for the author? Email Jim Hutchinson.
