Simulator Results – Breaking Down the Conferences
Last week, I shared the results of the Bracketology simulator for the 2020 season. That post covered the ACC, Big East, and SoCon in some detail. Today, I’ll do the same for the other conferences.
As a reminder, this data is based on running thousands of simulations of the 2020 season based on each team’s schedule and seeing how things shake out at the end. Because the committee hewed so closely to RPI last year in making selections, we can project how each team will fare on Selection Sunday.
Without further ado, here is the Big Ten breakdown…
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Penn State
- 12.1 – 2.7
- 8
- 5
- 17
- 88.8%
- 62.8%
- Maryland
- 9.9 – 5.5
- 13
- 14
- 4
- 64.8%
- 19.5%
- Ohio State
- 7.0 – 6.9
- 20
- 24
- 3
- 36.4%
- 7.0%
- Johns Hopkins
- 6.8 – 7.1
- 23
- 25
- 3
- 28.6%
- 6.7%
- Rutgers
- 7.1 – 7.6
- 31
- 31
- 15
- 10.8%
- 3.7%
- Michigan
- 3.8 – 10.4
- 48
- 52
- 11
- 0.4%
- 0.1%
I was a bit surprised when I saw the table for the Big Ten. Penn State is head and shoulders above the rest of the conference. Shockingly, the Big Ten only had two teams (PSU and MD) end the 2019 season in the top 20 of the Lax-ELO ratings. The Terrapins look strong, and with a very challenging schedule will have a chance to prove they can challenge the Nittany Lions.
But yeah, there is a bit of a muddle after that with Johns Hopkins, Ohio State and Rutgers all bunched up around the .500 mark. The Bucks and Blue Jays both boast difficult schedules, which gives them a punchers chance at an at-large berth should they not bring home the Big Ten title. If either gets in with a close to .500 mark, especially if it comes at the expense of a team like Richmond, expect the RPI debate to flare up again come Selection Monday.
Overall, it’s interesting to see that even with a Big Ten schedule, in the simulations, Penn State’s eventual SOS ranking comes in only at #17 nationally. The Lions scheduled some cupcakes this year, and you could easily imagine that coming back to hurt them if they are in the mix for a top-8 seed or, (fans close your ears) on the bubble.
An RPI rating that is lower than a team’s SOR indicates that RPI is expected to look favorably on that team’s schedule. This is true for 4 of the Big Ten teams. Penn State is the only one where their schedule actually would cause RPI to look less favorably on them than their resume would deserve.
Could the Big Ten end up as a 2-bid league? As it stands, the league is projected to earn roughly 2.3 bids. But for several of those teams in the central muddle, the RPI boost they get from a top-10 schedule could be the difference between getting in and being left out. That is a precarious situation.
Imagine a world where just one of those three teams has a melt-down this year. Suddenly, the Big Ten looks very average, and that RPI-boost may not be coming for the other B1G teams in contention. Of course, the opposite hypothetical could also occur. If Penn State drops a few conference games to the JHU/OSU/RU troika, it’s easy to imagine this ending up as a 4-bid league.
Point is, like a gerrymandered state legislative map, the benefit of a strong conference slate can be very protective…until it’s not. And that’s when things start to fall apart.
Ivy League
Let’s shift our focus over to the Ivies, where Yale looks to continue what has been a pretty impressive run over the past few seasons.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Yale
- 13.5 – 2.3
- 5
- 4
- 22
- 94.9%
- 62.9%
- Penn
- 8.2 – 5.2
- 11
- 13
- 3
- 67.7%
- 15.8%
- Cornell
- 9.6 – 5.7
- 13
- 14
- 5
- 60.3%
- 14.4%
- Princeton
- 8.4 – 5.5
- 25
- 21
- 21
- 21.5%
- 5.1%
- Brown
- 8.4 – 8.2
- 30
- 34
- 21
- 10.1%
- 1.7%
- Harvard
- 3.7 – 8.4
- 51
- 44
- 19
- 0.2%
- 0.1%
- Dartmouth
- 2.1 – 11.9
- 65
- 66
- 29
- 0.0%
- 0.0%
Deja-vu? This table feels very similar to the Big Ten table. A top team staring down a less-than-scary schedule. A second tier of teams that could challenge for supremacy if things break right. A rump at the bottom (although I’m sure the new coaching staff at Harvard has other thoughts).
If nothing else, the IL projections show just how important a strong schedule is for bubble-positioning. There are 4 IL teams that project with between 8.2 and 9.6 wins. They all have less than a 16% chance to win the conference’s auto-bid. So on the surface, you are looking at 4 fairly similar situations, and mostly, you are right.
But in that all-important Selection Committee meeting room, things couldn’t look more different. Penn and Cornell both project with a top-5 schedule. As a result, they currently enjoy just about even odds at an at-large berth should they not win the IL crown. Princeton and Brown on the other hand, are sitting there with less than stellar chances. In terms of overall resume strength (per SOR), Princeton is 7 slots behind Cornell at #21. But per RPI, the gap is 12 spots.
Such is the difference between a top-5 schedule and a schedule that projects as the 21st most difficult.
Nonetheless, the Ivies should feel pretty good collectively. As we embark on 2020, they are projected to have 2.5 bids to the NCAA tournament. Not bad.
One thing to keep an eye on though. The comparison with the Big Ten table highlights a potentially interesting situation. College lacrosse is not like college football, where schedules are set decades in advance (only a slight exaggeration). Teams are always tweaking schedules and opponents and there is an interesting underlying drama around schedule creation. Up and coming teams angling for games against big-time opponents. Historical powers arranging inter-sectional games or maintaining long-time geographic rivalries.
But the point I wanted to make about the Big Ten and Ivy League is that both have a team that had a great season and shows up the next year with an underwhelming schedule. At least as compared to the other teams in the conference. To the extent that every coach has a primary goal of getting into the NCAA tournament, is this a manifestation of some sort of over-confidence on the part of dominant teams?
Is the Penn State/Yale staff effectively saying: “we are good enough to win enough games to be shoe-ins, so what benefit is there in scheduling a gauntlet for ourselves?” They haven’t been burned recently by being a bubble team with a strong resume and a weak schedule.
Contrast that with teams like Maryland, Ohio State, Hopkins and Cornell who spent at least part of last spring in the bubble conversation. Every one of those teams projects with a top-5 schedule. Of course, the downside of a tough schedule is very likely a less-than-immaculate win-loss record, which is still a huge component of the RPI calculation (25% to be exact).
It’s certainly possible that all of these teams are over-reacting. Strong teams getting complacent and assuming they won’t need a top schedule to get in. Bubble teams assuming the opposite at the expense of having a better record. I’ll be interested to see how these schedule decisions work out for all of these teams.
Patriot League
And now to the Pat-Spencer-less Patriot League.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Loyola
- 12.2 – 3.7
- 6
- 8
- 14
- 89.8%
- 54.6%
- Army
- 11.8 – 5.1
- 20
- 18
- 36
- 38.6%
- 22.2%
- Boston U
- 11.9 – 4.8
- 25
- 21
- 44
- 24.2%
- 14.8%
- Bucknell
- 9.3 – 6.9
- 34
- 34
- 40
- 7.4%
- 3.3%
- Lehigh
- 8.6 – 6.7
- 39
- 31
- 49
- 3.9%
- 3.1%
- Holy Cross
- 7.4 – 8.4
- 43
- 45
- 37
- 1.7%
- 0.9%
- Navy
- 5.6 – 8.1
- 44
- 42
- 28
- 1.5%
- 0.8%
- Colgate
- 4.2 – 10.3
- 49
- 53
- 18
- 0.4%
- 0.3%
- Lafayette
- 1.1 – 12.9
- 64
- 67
- 15
- 0.0%
- 0.0%
It would be fair to dismiss this table because Lax-ELO doesn’t care that Spencer is no longer a Greyhound. The system is (purposely) ignorant about injuries, graduations, coaching changes. And I’m fine with that. If you read my calibration post, you’ll know that Lax-ELO has been very effective in projecting outcomes over the past seasons.
Still, I’d forgive you if you were to look at Loyola‘s projections and mentally downgrade them a bit. Not saying you should because a dumb model with no subjective biases is better than cherry-picking teams to mentally move up and down. But I can’t control your thoughts.
Regardless, Charley Toomey’s bunch leads the pack in the projections. That said, their chances of bringing home the auto-bid are worse than both Yale in the Ivy and Penn State in the Big Ten. And that is largely thanks to Army. BU clocks in 3rd and after that, no team has greater than a 4% chance at bringing home the title.
In terms of NCAA berths, it’s not a pretty picture for the Patriot League offices. Only two teams in the league project to have a top-15 schedule. Loyola and (checks his notes and SOS calculations) Lafayette. As a result, no team beside Loyola projects to have a better than top-20 RPI. If the Greyhounds win the tournament, it’s very likely a one-bid league.
To have a realistic shot at an at-large, Army would have to probably be a top-7 or -8 team in the country to overcome what is expected to be an RPI hit from their schedule, which projects as the 36th toughest in the country.
America East
If you are like me, you are still getting used to not seeing Albany as the top-dog in the AEC.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Vermont
- 9.7 – 3.9
- 27
- 23
- 62
- 48.1%
- 44.3%
- Albany
- 7.4 – 7.1
- 30
- 33
- 26
- 28.0%
- 24.0%
- UMBC
- 7.7 – 5.7
- 33
- 31
- 48
- 25.1%
- 22.2%
- Stony Brook
- 6.9 – 8.0
- 50
- 49
- 58
- 7.2%
- 7.0%
- UMass-Lowell
- 6.0 – 10.4
- 57
- 61
- 60
- 1.9%
- 1.9%
- Binghamton
- 2.9 – 11.2
- 62
- 65
- 40
- 0.3%
- 0.3%
- Hartford
- 5.0 – 10.2
- 65
- 64
- 71
- 0.3%
- 0.3%
But that is absolutely the state of play as the America East conference kicks off this year. The Catamounts have a lofty 44% chance to win the conference, and the Dane Train is only a few points ahead of third-place UMBC.
Vermont looks like one of those cases where a very poor schedule (62nd most difficult) could be somewhat incongruous with the overall resume. If you project as the 23rd most impressive team in 2020, but your schedule is that weak, you would probably benefit from a few tougher games to really test that strength. Vermont isn’t going to do themselves any favors if they end up with an 12 win season but a bottom-15 schedule. In that case, at least from an at-large perspective, you’d probably rather be at 11 wins but a top-15 schedule.
Certainly, it’s not that simple, but for mid-tier teams with designs on an at-large-worthy resume, there seems to be room to improve schedule design. (Again, the committee can render this moot by looking at resume strength instead of RPI=schedule strength.)
Colonial
The Colonial appears to be a two-horse race, but there is a whole season ahead of us.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Towson
- 8.6 – 6.0
- 14
- 19
- 9
- 64.0%
- 36.6%
- UMass
- 10.3 – 5.2
- 19
- 20
- 27
- 50.0%
- 34.8%
- Delaware
- 9.8 – 5.6
- 29
- 27
- 43
- 24.2%
- 18.9%
- Drexel
- 7.4 – 7.5
- 37
- 40
- 32
- 9.2%
- 6.8%
- Hofstra
- 7.4 – 8.2
- 49
- 47
- 49
- 3.0%
- 2.9%
- Fairfield
- 3.0 – 14.1
- 61
- 70
- 29
- 0.0%
- 0.0%
Lax-ELO pegs Towson and UMass as the top dogs in the Colonial this year. Towson’s schedule is strong enough to keep them in the bubble conversation even with a loss in the conference tournament; they are probably the only team that can say that. Still, given that the CAA has two teams that project in the top-20, they are projected with ~1.5 tournament teams. Certainly not destined to be a one-bid league by any means.
That said, UMass could easily end up as a team with a record that looks like it should be part of the bubble conversation yet be an after-thought on Selection Sunday. There are just too many teams with too strong of a schedule, especially in the mid-tier (i.e. Richmond).
But I don’t want to be too harsh on the Minutemen. In roughly 15% of the simulations, they didn’t win the CAA, but still got a bid. The key to their at-large chances is probably the Yale game. The Gorillas earned a slot in the NCAA tournament in 60% of the simulations where they beat Yale. If they lose to the Elis, they are in just 48% of the time. It’s not the largest gap for them, but it’s their best chance to create a tournament-worthy resume.
MAAC
There is something pure about a one-bid league. Schedule strength doesn’t matter. Conference standings/seedings matter. The tournament decides the NCAA participant. Without having to worry about RPI and all that, things become simple. There is something nice in that simplicity.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Quinnipiac
- 10.1 – 6.4
- 32
- 36
- 61
- 35.1%
- 33.2%
- Canisius
- 9.0 – 6.4
- 31
- 38
- 59
- 30.7%
- 27.6%
- Detroit
- 8.3 – 6.9
- 36
- 41
- 56
- 19.4%
- 17.7%
- Marist
- 7.7 – 7.2
- 42
- 44
- 62
- 13.9%
- 13.1%
- Monmouth
- 7.5 – 7.2
- 51
- 48
- 73
- 6.1%
- 6.1%
- Siena
- 5.6 – 8.6
- 59
- 57
- 71
- 1.6%
- 1.6%
- St. Bonaventure
- 3.6 – 10.6
- 61
- 64
- 55
- 0.4%
- 0.4%
- Manhattan
- 5.2 – 8.9
- 67
- 62
- 74
- 0.2%
- 0.2%
Don’t get me wrong, if the MAAC ended up more than one-bid, it wouldn’t be a huge shock. This early in the season, anything can happen. And in fact, the probabilities suggest that they should end up with 1.07 tournament bids. That would take a huge leap from someone in this league who then ended up upset in the conference tournament. It’s not impossible, but I wouldn’t bet on it.
Northeast
This deserves a Lax-ELO digression.
- Team
- Avg Record
- RPI
- SOR
- SOS
- NCAA%
- Conf%
- Robert Morris
- 11.3 – 4.1
- 22
- 22
- 63
- 45.4%
- 35.8%
- Hobart
- 9.0 – 5.3
- 26
- 27
- 46
- 30.8%
- 22.4%
- Sacred Heart
- 8.9 – 5.3
- 29
- 31
- 54
- 26.2%
- 19.6%
- Merrimack
- 9.8 – 6.0
- 43
- 35
- 70
- 8.7%
- 8.5%
- LIU
- 7.7 – 5.9
- 40
- 36
- 67
- 7.2%
- 5.8%
- Saint Joseph’s
- 6.8 – 7.7
- 38
- 41
- 38
- 6.8%
- 4.4%
- Mount St Marys
- 6.7 – 7.6
- 42
- 41
- 45
- 4.0%
- 3.1%
- Bryant
- 5.7 – 9.4
- 56
- 57
- 63
- 0.5%
- 0.4%
- Wagner
- 2.6 – 11.4
- 69
- 69
- 69
- 0.0%
- 0.0%
Lax-ELO ratings are the life-blood of our predictions and projections. ELO ratings were originally developed for chess, and they are a time-honored method for ranking teams or individual competitors. The crux of the system is that in each game, a number of ELO points are transferred from the loser to the winner. You can also use ELO ratings to convert two teams’ ratings into a probability for each team. You can read more about how I have implemented ELO ratings as Lax-ELO here.
The reason that I bring this up is because LIU and Merrimack, being new teams to D1, present an ELO challenge: what should their starting rating be?
Given the track record of teams joining D1, you might expect that a new team’s Lax-ELO rating would be below average (1500 is the average rating). On the other hand, these teams were good enough in the lower divisions that maybe they will be closer to average in D1. After all, Utah had some successes in their first year.
So we have decided to go with…1500. Average.
And frankly, the reason we set new teams to “average” is because we don’t have a better plan. I mean, where would you start them? It has to be objective for every new team. What is the right number? My philosophy is that when you are making a subjective decision, you are going to introduce some distortions. The more complexity to the reasoning, the more distortions and the greater the risk of cherry-picking. What is the least complex answer to the question of a new team’s ELO rating? 1500. Average. Easy.
It’ll take a few seasons for a new team’s ELO rating to stabilize, but that is going to be the case regardless of what you start a new team at.
As for the rest of the league, Hobart seems to be getting their fair share of pre-season attention, but Robert Morris is actually the team that Lax-ELO likes best in the NEC. RoMo has the best chance to win the conference, at 35.8%, with Hobart and Sacred Heart roughly tied in 2nd.
Interestingly, this is a case where SOS doesn’t have a significant impact on the chances of an at-large berth. In fact, Robert Morris, with the worst schedule of the top-3, actually has the best chance at an at-large berth should they not win the NEC tournament. I suppose that goes to show that for a schedule to be a net positive come Selection Sunday, it has to be objectively strong and not just relatively strong. Clearly, despite having a schedule that looks about 20 spots more challenging than Robert Morris, Hobart isn’t getting an at-large bump.
It may be worth some more research to see how schedules of varying strengths impact a team’s RPI relative to their SOR. I would bet that the difference between a top-5 schedule and a top-15 schedule is much greater than the difference between a top-15 schedule and a top-25 schedule.