by Matthew Webb
Senior Writer, D3hockey.com
It's hard to believe we're already onto week three of Men's Bracketology but indeed we are as the third set of men's NCAA regional rankings were released on Tuesday. It's doubly hard to believe that Selection Sunday and our final edition of Bracketology are a mere five days away. But they are which means it's time for us to take our second-to-last stab at projecting the Men's NCAA Tournament field (were the season to end today) based on the published process the NCAA National Committee should be using to select and seed the NCAA tournament.
|A borderline ranked team just two weeks ago, Utica has vaulted itself into the Pool C discussion. But will the Pioneers have enough to find their way into this week's field?
Photo: Utica Athletics
This week's regional rankings were a little less volatile than those of a week ago, but the changes that did occur are going to alter this week's analysis -- at least when it comes to Pool C.
This is thanks to the East Region, which saw some major shuffling in spots 5-10. Meanwhile, in the West, we've got the same five teams in the same order they have been in the previous two weeks.
How much different will this week's field look compared to last? Tough to tell, and there's only one way to find out so let's get on with the show and see what our penultimate analysis reveals...
But first, our standard housekeeping: our goal, and our only goal, is to illustrate the process as it is published in the NCAA Men's Division III Ice Hockey Pre-Championship Manual. To do so we will implement the process based on known regional rankings and NCAA-generated statistics for the relevant teams, and we will do so to yield the bracket that we feel maximizes fairness and equity of the field while concurrently adhering to NCAA rules. To this end, we remind ourselves of a few governing principles that we consider to be paramount:
- Process, process, process. The published process is the only thing that matters. You might not like it and we might not like parts of it, but it is what it is. To worry about things outside of the process muddies the waters and tends to lead to more confusion amongst fans than it resolves. Learn more about the process here.
- It has been numerous years since the tournament selection yielded an outcome that was not easily explained by the tenets of the process itself. While some, including us, may have differed with the committee's decisions at times, they nonetheless have fallen completely within the framework of the process.
- To cross the boundary from interpretation and application to delve into the world of speculation does a disservice to everyone. There is enough of that swirling around out there this time of year and it is something we will not engage in. Our goal is to shed light on the process, not obfuscate it for the sake of our own egos.
- The 2017 Men's Division III Ice Hockey Championship will consist of 12 teams.
- Seven conference playoff champions will receive automatic qualifying bids (Pool A) to the tournament. These conferences are the: CCC, MASCAC, MIAC, NCHA, NEHC, NESCAC & SUNYAC.
- One team from a conference that does not possess a Pool A bid will receive a Pool B bid. This will be awarded to an independent, ECAC West or WIAC team.
- Four teams that do not receive Pool A bids nor the Pool B bid will receive at-large (Pool C) bids into the tournament. Every team that did not receive a Pool A or Pool B bid is eligible for a Pool C bid.
For the purpose of Pool C selection, as well as team comparisons for regional ranking purposes and tournament seeding, the NCAA committee will rely mainly on what it refers to as its Primary Criteria, which are as follows:
- 1/3 Opponents' Opponents' Winning Percentage (OOWP)
There are also three Secondary Criterion the committees may look at if they are unable to come to a determination by using only the primary criteria, and they are as follows:
- The NCAA regional committees will release three editions of regional rankings prior to tournament selection, with the release dates being February 14, 21 and 28. A fourth will be generated on Selection Sunday (March 5) and is slated to be released to the public on March 6.
- The East Region rankings will contain ten teams, the West Region five.
- For the sake of RNK, the NCAA considers a team ranked only if it is ranked at the time of tournament selection (in the March 6 final rankings).
NCAA Regional Rankings - February 28
|EAST REGION||WEST REGION|
|1. Norwich||1. St. Norbert|
|2. Oswego State
||2. UW-Stevens Point
|4. Hobart||4. Augsburg|
|5. Hamilton||5. UW-Eau Claire|
10. Plattsburgh State
Analysis: There isn't a lot to analyze here beyond noting that Utica and Trinity are now up in the spots Williams and Babson were in last week, which is ultimately going to change the set of teams we examine for Pool C.
Beyond that, we're just going to take these as they are as recall, we care far more about what these are than why these are what they are.
- Pool A
As Bracketology assumes this would be the tournament field were the season to end today, we will use the teams leading the seven Pool A eligible conferences as our automatic qualifiers. And it's the same seven teams we saw a week ago. The seven Pool A bids go to:
- Pool B:
As only ECAC West, WIAC, and independent teams are eligible here, this should be easy enough. No independent teams are anywhere in the mix we'll compare the highest-ranked ECAC West team, Hobart, with the highest-ranked WIAC team, UW-Stevens Point:
|RNK|| .5000 (2-2-2)
|| .6426 (4-2-1)
Analysis: For the third week in a row we are looking at a very similar comparison between the same two teams, and for the third week in a row our conclusion comes rapidly: Stevens Point gets the bid. At this point it hardly matters, however, as both of these teams are again in the field one way or another. Fun fact: A Hobart win paired with a Stevens Point loss this weekend hands this comparison to the Statesmen in a sweep.
Now, we got that settled so fast we're going to use this comparison to illustrate something that has been gnawing at us, and as it's a concept that could factor into this bid or Pool C on Selection Sunday we'd like to address it. Apologies to Hobart and Stevens Point as this doesn't directly relate to either, but it's a convenient example to use.
For Hobart to win this comparison as-is we'd have to give it the nod entirely on the strength of a very narrow edge in SOS. Is it possible the committee could do so? Of course, and it wouldn't be that outrageous considering how much we've seen SOS get valued in the past. Which brings us to a belief that has brewing within us for a long time:
The committee severely overvalues SOS.
And here's why.
It's one thing to just look at numbers and say x is greater than y thank you very much have a nice day team A gets the bid, but the entire exercise is meaningless, if not downright flawed, if we don't understand the actual relationships those numbers represent.
In the Hobart-Stevens Point comparison, the SOS difference amounts to .005, which is five thousandths of a point, or one half of one percent. That's pretty small. But how small is it really, and what does it tell us?
To illustrate, let's wander over into the Division I standings and take a look at Providence (20-9-5) and Notre Dame (19-10-5). Those records amount to win percentages of .662 and .632, respectively. The difference in their winning percentages is six times the difference in SOS between Hobart and Stevens Point. Or think of it this way, if one team played Providence 25 times on the year and another played Notre Dame 25 times, at this point in the year the difference in SOS between the two would be .030.
Translating into Pool selection terms, that's the same as the difference between teams with SOS values of .530 and .500 -- a gap which has historically been treated as a very significant margin, if not a downright dealmaker or heartbreaker.
So, does a SOS gap of .030 between teams actually tell us anything useful about the two teams in question? We contend it doesn't tell us much, unless we're prepared to accept that 20-9-5 is vastly superior to 19-10-5, which is preposterous.
But that's exactly what's being done when a gap of .030 is treated with such great significance, which rings a little odd to us. And remember, the SOS difference between Hobart and Stevens Point is one-sixth of that. So, one-sixth of the difference between 20-9-5 and 19-10-5.
It doesn't tell us a thing.
Now, despite this we're still going to have to take SOS seriously as it's one of the criteria and we follow the process, but we felt this was a good spot to illustrate the concept that without understanding what the numbers are actually telling us, the numbers are useless. This concept will surface again within the next six days.
The Pool B bid is awarded to: UW-Stevens Point
- Pool C:
We now must consider which teams will get at-large bids to the tournament. That requires looking at the highest-ranked teams in the regional rankings that are not projected to win Pool A, and those are:
East: Hobart, Utica, Trinity
West: St. Norbert, Augsburg, UW-Eau Claire
Let's begin by constructing a table that shows each of these six teams' relative winning percentage (WIN), strength-of-schedule (SOS), and record against ranked teams (RNK). We'll eye things up and see if we can come to any obvious conclusions.
Analysis: We're going to get the easy part out of the way in a hurry here and take St. Norbert and Hobart immediately. They are the highest-ranked teams remaining in their respective regions and fare quite well by the numbers. Simply put, we find it very, very difficult to think both would not grab a bid here, so we're going with it.
Now things get a bit challenging as we have four teams left but only two tournament bids to dole out. This week we're going to shift gears and move to the expanded table as for the first time this year we have some meaningful COP and H2H data to look at:
|Utica||.6920||.5200||.5000 (2-2-0)|| 1-1 v. Eau Claire
4-1-0 v. Trinity
|Trinity||.6740||.5230||.5830 (2-1-3)|| 2-0-2 v. Utica
|Augsburg||.7200||.4920||.6667 (2-1-0)||11-2-5 v. Eau Claire||2-0 v. Eau Claire|
|UW-Eau Claire||.7220||.5280||.0000 (0-5-0)|| 1-1 v. Utica
11-1-5 v. Augsburg
|0-2 v. Augsburg|
Well, ain't this just grand? This is tight. In past weeks we've sort of cheated this process by using regional rank to see who is up for comparison first, which is fine when things are pretty straightforward but not fine in an instance like this. Why? Because that's not really how teams are compared. Rather, each committee member ranks the teams, the votes are tallied and the top team gets the bid. Then repeat until all spots are filled (Does anyone really think St. Norbert and Hobart wouldn't be out of the way early here?).
Is there anything we can immediately conclude here to try to help us whittle away at things? We think so, and it's that we'd rank Augsburg over UW-Eau Claire every time. The Auggies do have a pretty big SOS deficit and WIN is a push, but they win RNK handily thanks to beating none other than Eau Claire. Twice. That's enough for us in this instance as those games need to matter, and it seems our thinking is in line with the West committee as they have ranked Augsburg ahead of Eau Claire for three weeks and counting.
So that means we're down to three for the next bid. Augsburg has the edge in WIN and RNK but is facing large SOS deficits to both -- at least in terms of how it is historically interpreted.
Which, of course, does not coincide with how we think it should be interpreted which is why we hijacked the Pool B discussion to explain why we believe small differences in SOS get highly overvalued, but this puts us at a crossroads. Do we interpret it in a way that is in our minds is correct, or do we try to mimic the process as it has bee implemented for years and years? Our hearts and senses of self-pride say the former, but our heads say the latter. The .492 SOS and resultant deficits with Utica and Trinity are, sadly, enough to take the Auggies down here as we don't think its small edges in WIN and RNK are enough to overcome a SOS deficit that been historically viewed as large.
Which means, albeit in rather resigned fashion, that we're done here. The Pool C picture is bound to change by Sunday and we say good as this would be a painful comparison to deal with when it matters.
The Pool C bids are awarded to: St. Norbert, Hobart, Utica, Trinity
Setting the Field
Thus, our full tournament field is:
|Pool B:||UW-Stevens Point
|Pool C:||St. Norbert|
Seeding the Field
Now the field must be seeded by region. Using this week's regional rankings (and a bit of guesswork on a tight comparison between Plymouth State and Nichols that could certainly be wrong), it would look something like this:
1W St. Norbert
Setting the Bracket
To quote ourselves from last week: "We again have ourselves an 8-4 split, which presents the increasingly familiar question that is probably becoming second nature to many by now: Do we go with two West Region quarterfinals, two West Region first round games, or send Adrian east and roll with one quarterfinal in the West?
We continue to think this is a pretty balanced field numerically which means we are again going to choose the middle road here as we find it the most equitable thing to do. That means one West Region quarterfinal and Adrian-Hobart-Oswego together in a quarterfinal. In this instance Adrian, by virtue of its higher regional seed, would host Hobart in a first round game with the winner traveling to Oswego State for a quarterfinal.
From there, we simply drop the remaining East teams in by seeds and we're done."
Our bracket is:
4W St. Thomas @ 2W UW-Stevens Point
|Click to view full-size.|
4E Hobart @ 3W Adrian
8E Plymouth State @ 5E Hamilton
7E Trinity @ 6E Utica
4W St. Thomas/2W UW-Stevens Point @ 1W St. Norbert
8E Plymouth State/5E Hamilton @ 3E Endicott
4E Hobart/3W Adrian @ 2E Oswego State
7E Trinity/6E Utica @ 1E Norwich
As we have, in our opinion, maximized bracket integrity while adhering to travel limitations, the only question left is which teams should be lined up to face off in the semifinals? You may recall that last week we lined up the Oswego quarterfinal with the St. Norbert quarterfinal so as to give the highest-seeded team, Norwich, the lowest-seeded potential semifinal.
We didn't like it at the time as it created a very unbalanced bracket in terms of regional seeds but went with it anyway so as to present the option. This week, we're switching it back in the name of respecting regional seeds, which is an emphasis of the NCAA. St. Norbert, which we surmise is the No. 2 overall seed, is already lined up to play 2W in the quarterfinals which means it's not far-fetched to think the Green Knights could draw the lower-seeded quarterfinal in the semifinals. It's happened before.
What we're now left with is regional seeds of 1,2,3,4,5,8 on one side of the bracket and 1,2,3,4,6,7 on the other. That's as good as it gets.
Potential Frozen Four pairings are:
St. Thomas/UW-Stevens Point/St. Norbert v. Plymouth State/Hamilton/Endicott
Hobart/Adrian/Oswego State v. Trinity/Utica/Norwich
Questions or comments you'd like further explanation on or you simply think we're nuts? No problem, we'd love to hear from you. Feel free to jump in with your own questions or ideas in the Bracketology Discussion over on the D3sports forums, comment below, or you can always yell at us on twitter @d3hky.