"My rule was I wouldn't recruit a kid if he had grass in front of his house.
That's not my world. My world was a cracked sidewalk." —Al McGuire

Monday, March 08, 2021

Championship Week Runs & the Bubble: The Data

Data Explanation

Green/Red Highlights: These are teams that moved in (green) or out (red) of the field in the period between the last Bracket Matrix results before Championship Week and Selection Sunday.

Last Safe: These are the first four at-large teams about the traditional 16 considered to be on the bubble. They are included because those teams are investigated as possible bubble teams that played their way off the bubble with strong Championship Week runs.

Last Byes: The last four at-large teams in the field that don't have to play a play-in game.

Last Four: These are the last four at-large teams that do need to play a play-in game.

NIT 1-Seeds: The top NIT teams, presumably the first four teams out of the field.

NIT 2-Seeds: The next NIT tier, presumably the next four teams out of the field.

CT Record: This was the team's record in conference tournament games for the given year.

R1/R2/R3: These are the results the teams had based on the above record. W or L in the column indicates a win or loss. Q1, Q2, Q3, or Q4 indicates that said result was in that respective quadrant.

Pre-ES/Post ES: Because every field is different in terms of bid thieves and the placement of the automatic bids, we are using "ES" which is shorthand for Effective Seed. This is to create equity among the 5 different seasons, and effective seed is based solely on how far the team is from the at-large cutline. For the purposes of this exercise, we are assigning five primary values for the 5 analyzed categories. A "9" would refer to teams in the Last Safe category, a "10" would refer to teams in the Last Byes category, an "11" would refer to teams in the Last Four category (as 11-seeds are the typical location for at-large play-in games), a "12" would refer to teams in the NIT 1-Seed category, as 12-seeds are frequently automatic bids and thus would be the first teams pushing at-large candidates into the NIT, and "13" would refer to teams in the NIT 2-Seed category, as they are the next batch of teams. In the Pre-ES category, there are teams assigned "7," "8," and "14." "7" and "8" indicate these teams were placed higher than the Last Safe category by the Bracket Matrix aggregate before conference tournament play. "14" indicates a team that was not on the Bracket Matrix aggregate, and thus presumed to be a team that bracketologists saw as a NIT 3-Seed. There are also teams in the Pre-ES category delineated with "11-A." This indicates the team was previously an 11-seed as an automatic qualifier.

ES Change: This is the degree of change the team had between the last Bracket Matrix report to start Championship Week and the actual Selection Committee placement on Selection Sunday. This is measured strictly by the Effective Seed, so it is batching teams by category, not S-Curve position. Teams listed with a "0" had no change. Teams listed with a positive colored number did experience a change. Green numbers indicate the team's seeding improved and red numbers indicate the team's seeding worsened.


To evaluate whether or not these games matter in the eyes of the Selection Committee beyond the auto bids, I evaluated 20 teams from each of the past 5 NCAA Tournaments, so 100 teams in total. The 20 teams included the last 12 at-large teams to make the field and the first 8 teams out on the basis of the Selection Committee's official seed lists and the NIT 1 and 2 seeds. I broke those teams up into batches of four, with the added term of "Last Safe" to describe teams that were ahead of the first four byes but plausibly could have been on the bubble to start the week. The other terms I used, Last Byes, Last Four In, and the NIT 1 & 2 Seeds for the First Four and Next Four Out should be straight forward.

I then looked at two primary factors. The first was Conference Tournament record and the respective results. All 100 teams went either 0-1, 1-1, or 2-1 in their Conference Tournaments, so it was a pretty easy breakdown. The results included the wins and Quadrants the wins were in, since winning a Quadrant 1 game in the Big East Tournament should be more valuable than winning a Quadrant 4 game in the WCC. While the Quadrant system wasn't in place for all of these years, the neutral site Quadrant breakdown was the same on the team sheet both before and after the renovation of the Quadrant system, so those results have been applied equally across all 5 years.

Effective Seed describes a team's placement before and after Championship Week. Effective Seed does NOT mirror the actual seeds given on either Bracket Matrix nor the actual NCAA Tournament field. Because every tournament will be different with bid thieves and the placement of automatic bids, Effective Seed assigns a seed based strictly on how close to the at-large cutline a team is. I started by assigning the Last Four In a score of 11 because every at-large play-in game the past 5 Tournaments has been contested on the 11 line. As the 12-Seeds are typically the best auto-bids, I assigned 12 to the NIT 1-Seeds, which are the first and teams out of the field if not for the auto-bids needing to be placed. With that in mind, the effective seeds came naturally, with 9 assigned to the Last Safe teams, 10 to the Last Four Byes, 13 to the NIT 2-Seeds, and 14 to any teams that would've been NIT 3-Seeds or worse (not included in Bracket Matrix).

Effective Seed Change shows how many Effective Seed lines a team had changed from the start of Championship Week to Selection Sunday. A 0 means there was no change, green numbers mean a change for the better, and red numbers mean a change for the worse.

No comments: