Small Margins

With the NCAA Tournament halfway through, I thought it was time to look at the small margins that made up the field of 64. All year, we talked about how the NCAA Power Index (NPI) rewards wins and punishes losses. It is no longer about winning against the elite or playing a tough schedule, it’s about winning. We know that Centre was our last at-large team and that this flipped once Smith beat MIT in their conference tournament. (The very last match of our regular season.) Babson was the unfortunately recipient of the flip and in the end these two teams were separated by just 0.0485 points. As I’ve said all season, NPI is about small margins and losses cause margins to expand greatly.

What I wanted to do here is look at how many teams were one loss away from making the tournament. I don’t have to do the math to know that if Babson had not lost to Clark that they would have been in the tournament. Same goes for second team out, Calvin, and their loss to Edgewood. Southwestern was the third team out at 58.8275. That’s almost 0.8 points behind Centre. If you take away their loss to HSU, their NPI would jump to 59.60. If you factor in the HSU win then their NPI goes to 59.81. Remember, I’m not factoring in the ripples this change in the time-space continuum would cause. Regardless, that NPI moves them ahead of William Smith and would have secured an at-large bid. In other words, HSU was Southwestern’s small margin in 2024.

Eastern was the next team out and they finished at 58.8095. Their worst loss was to Haverford. Flip that match and Eastern’s NPI moves to 59.6429. That puts them ahead of UW-Stevens Point and secures an at-large bid.

Moving down the list, we next come to Swarthmore. Their worst loss was also to Haverford. Well, I guess we see who the naughty team of 2024 was! Santa is not going to be happy. Flipping that match and Swarthmore goes to 59.7345 and congratulations to Haverford for knocking two teams out of the tournament!

Next up is Gustavus Adolphus who finished the year 58.5424. Their worst loss was to Aurora. Flipping that result moves GAC to 59.3700 and well behind Centre for an at-large spot. I guess they should have lost to Haverford? Let’s check one more to make sure we’re done here.

Coe College finished at 58.4405 and their worst loss was to Wartburg (twice). Flipping one of those losses jumps Coe to 59.3227 and still out of the at-large picture.

That means we had 5 teams needing bids that were one loss away from an at-large and I’m not flipping their hardest opponent but their easiest (or, maybe better to say, their lowest NPI opponent). This is the match that NPI says they should have won. In most cases, this is a match that the team had no clue that this was their Waterloo. I’m not doing the math, but a couple of other teams would probably have had an at-large by flipping a better opponent. Just one match is the difference for a number of teams outside looking in.

Back to Southwestern. This is obviously a team I follow very closely so I made sure to track them as they marched towards their conference tournament. They went into their next to last weekend playing TLU away and ETBU at home. I knew they couldn’t afford to lose to both teams but with the way all three teams were playing at the time, that was exactly the result that was most likely to occur (and did occur). I wondered if the team shouldn’t just catch the flu and cancel the ETBU match. Again, not factoring in the ripples, but Southwestern would have ended the season at 59.1722 not playing that match. An improvement, but it wouldn’t even change their placement. We know flipping either result would have put them in the tournament because both teams finished with higher NPIs than HSU. Another fascinating aspect with regards to my team was how the SCAC Tournament bracket was constructed. By virtue of a tiebreaker with TLU, Southwestern was the 2-seed out of the Gold Division while TLU was the 3-seed. This meant that Southwestern would face the easier road of McMurry and then Colorado College (probably) before making the finals against Trinity (most likely). The problem was that Southwestern needed the harder road for an at-large bid. They needed Austin College and then Trinity before going to the finals. (Yeah, don’t remind me that they weren’t beating Trinity in the semis because they lost to them in the finals.) The point is that they needed the extra NPI points those two teams would have provided assuming they could have gotten the wins. (St Thomas’ upset of Colorado College in the first round also reduced the number of points Southwestern earned but it didn’t matter in the end.)

Enough of Southwestern, but I did want to look at scheduling so let’s turn our attention to Calvin. I’m using Calvin because they are more likely to play a schedule involving teams outside their immediate area as opposed to Babson.

I’m not sure the 2024 version of Calvin deserved an at-large bid but there were certainly a team on the bubble. Their best win, and only win over an at-large team, was Carnegie Mellon. Their next best win was Whittier. A fine team but one that, like Calvin, did not make the at-large field. They did have some impressive losses. Hope took them down 3-times. They also lost to Marietta, Millikin, Chris Newport and Chicago. All of those teams made the tournament. In the end, those matches helped give Calvin a 0.608 Strength of Schedule (SOS). Their record, that SOS and a couple of ranked wins might have been enough in our old selection system to get Calvin back in the dance. With the new system, their NPI was 59.27. When they played a team with an NPI at 57.7 or higher they went 2-9. It appears that Calvin might have been overrated by NPI but it’s just a formula and the real problem was Calvin’s schedule. Because of their losses (against too many good to elite teams) they had to count 15 wins. (The more wins you count, the lower the NPIs you encounter.) Another problem with their schedule is that their conference, outside of Hope, isn’t that helpful in the NPI game. What Calvin needed to do, with the ability of hindsight, was trade one or two of those SOS making matches with a safer team like Carroll, Denison or Aurora. (Teams they played and swept and who gave them match NPIs over 60.) This is the paradox for teams scheduling in 2025. How many great to elite teams do we risk and how many “safe” NPI teams can we schedule?

In years previous, I would do “Case Studies” that would look at the selection criteria and see if the committee had made a mistake. With NPI, I believe the entire formula is a mistake but there is no arguing with a number. The number is never wrong. Here, however, we can see how one different outcome would have impacted NPI and the at-large teams. Again, small margins.