Projecting The Field of 64 With Analytics

Using analytics is far more complicated when trying to build predictive models for players. But for schools, it’s far less so. Albeit harder this year than most due to two conferences only playing themselves, and another basically cancelling the season; there are still historically-based and far less volatile metrics (like ELO) available to rank teams.

There are 31 conferences, but the Ivy league won’t have any representation. So there are 30 spots up for automatic bids, and 34 spots up for at-larges.

To make things easy, we’re picking the team with the highest ELO to win their conference. Obviously, this won’t happen, so a handful of at-larges will get stolen by the obvious choices. But the 30 conference winners are: Arkansas, TCU, Notre Dame, Arizona, East Carolina, Louisiana Tech, Indiana, Georgia Southern, UC Santa Barbara, Gonzaga, Dallas Baptist, Mercer, UNLV, Northeastern, Liberty, Connecticut, Southeastern Louisiana, Morehead State, South Carolina Upstate, Oral Roberts, VCU, Grand Canyon, Stony Brook, Central Michigan, Army, Wright State, Fairfield, Bryant, Florida A&M, Jackson State.

Now onto the at-large teams, the methodology is quite simple, a combination of adjusted runs scored/allowed (accounts for strength of schedule), ELO Delta, ELO ranking, opponent record, Quadrant one through four ELO records, and lastly, but most importantly, the teams actual winning percentage adjusted for strength of schedule. Here’s how those metrics work and why I’m using them:

Adjusted runs scored/allowed (differential/margin): Takes runs scored/allowed and adjusts for strength of schedule.

Quadrant One: Record against teams with records inside the top 50 in ELO.

Adjusted win percentage: Similar to adjusted run differential, takes win percentage and accounts for strength of schedule to put more context into the most important statistic.

ELO: a method for calculating the relative skill levels of players in zero-sum games such as chess. It is named after its creator Arpad Elo, a Hungarian-American physics professor. The Elo system was originally invented as an improved chess-rating system over the previously used Harkness System, now it’s being used in association football, American football, basketball, and Major League Baseball. The reason ELO is better then RPI is because it has better judgement on schools only playing in-conference as it’s a touch more historical with less volatility.

ELO Delta: Amount of points received based on ELO calculation.

Disclaimer: the teams below aren’t ranked from top to bottom, it’s just the best 34 at-large cases according to the numbers I’m valuing. So just because team A is mentioned first, they aren’t necessarily better than team D. 

Texas: They’re 36-11, have the 5th highest adjust run differential, the 4th highest ELO. They’ve yet to lose a game to an opponent with an ELO below 150, and they’re 11-7 against teams from 1 to 50.

Mississippi State: Are 2nd in ELO, 8th in adjusted run differential, they’ve lost to a team with an ELO below 150, but have the 3rd highest SOS and are 21-9 against teams with an ELO below 100.

Vanderbilt: They have the 9th highest winning percentage, are 7th in adjusted run differential and are one of just three schools (Mississippi State/Arkansas) with a top 16 SOS and ELO.

Ole Miss: Have the 19th highest winning percentage with the 9th hardest SOS while having the 6th most runs adjusted runs per game. They have the 6th highest ELO and are one of just 8 schools to have zero losses against schools with ELO’s below 100.

Tennessee: They have the 3rd most wins in the country while having the 15th highest SOS. They’re also 15-7 against top 50 ELO teams and are 19th in adjusted run differential.

Texas Tech: Also haven’t lost a single game against the bottom two quadrants while having just one loss against teams with ELO’s below 50. They’re just 10-10 against top 50 ELO teams but rank 13th in adjusted run differential, and 18th in win percentage against the 22nd highest SOS.

Oregon: Surprise team. They’re 20th in adjusted run differential with a superb 14-5 record against top 50 ELO teams. They’re just 13-6 against teams outside of the top 50, but rank 24th in ELO and have the 21st highest SOS.

Old Dominion: Have the 32nd highest SOS, rank 28th in ELO, are 8-6 against top 50 teams with a  11-7 record against the top 100.  They’re the only mid major with an adjusted run differential higher then 1.8 as well as at least one win against a top 50 team. A 1.98 adjusted  run differential per game has them 7th in the nation, and ahead of Vanderbilt, Mississippi State, and Arizona.

Florida: Rank 5th in ELO, have won just over 70 percent of their games. Somewhat low quadrant one record of 9-11, but 15-12 when you expand it to the top 100. Rank 52nd in adjusted run differential with the 42nd highest SOS.

Baylor:  18th in ELO with an SOS ranked 65th. But they have the 25th highest adjusted run differential with a 4-6 record against the top 50. They’re just 7-5 against the teams ranked 51st-100th, but have the 33rd highest win percentage.

Charlotte: Lack an Old Dominion like record against top 50 teams, but they’re 10-8 against the top 100, rank 31st in win percentage, have 31 wins against the 12th hardest schedule while having the 22nd ranked offense.

UCLA: 14th in ELO. Have the 21st ranked offense with a .641 win percentage against the 38th highest SOS. They’ve won 8 of 15 games against the top 50 with zero losses against the bottom ~150.

San Diego: Rank 32nd in adjusted run differential with a .771 win percentage, but the 222nd ranked SOS. They’ve yet to face a team in the top 50, and are just 2-2 against the top 100 but are 15-2 against the bottom ~150. They’ve had eight losses, three by just one run, and two by just two runs. They’re 9-6 on the road, which ranks 48th.

Louisville: Only 24-14, but are 11-9 against the top 50 with the 16th highest ELO. Non-conference schedule is just 215th in the country, but they have series wins over Virginia, Florida State, Wake Forest, Boston College, and NC State. They’ve beaten Vanderbilt in the mid-week and have a winning record against all four ELO quadrants.

Miami (Florida): They’ve only faced two teams outside of the top 100, and won both. They’re 19th in ELO with a .615 win percentage, 12-9 record against the top 50 and a top 30 schedule.

North Carolina State: Only 22-14 with a bad record against the top 50 (3-12) but they’re 12th in ELO with a 15-13 record against the top 100 and just one loss to a team below the top 100. They’re 65th in adjusted run differential and have the 39th hardest schedule.

LSU: Rank 13th in ELO. 8-14 against top 50, 18-3 against teams outside of it with the 2nd highest SOS and a .614 win percentage. Also 52nd in adjusted run differential.

Tulane: 39th in ELO. Win percentage of .632 ranks 60th against the 80th SOS. They’re just 1-5 against top 50 but 5-3 against 50-100 teams, 9-5 against 101-150, and 9-1 against the rest. 

Stanford: They have the 20th best record with a 7-3 record against the top 50. They have three losses against bottom 150 teams but have a 12-5 record against the top 100. But their ELO currently ranks 41st.

Southern Miss: Rank 30th in SOS with a .674 win percentage and 4 wins against the top 50. Their adjusted run differential ranks 49th and they’re tied with Oregon State for the 20th most wins in the country.

Pittsburgh: Have one of the best quadrant one records in the country at 12-6. Their ELO ranks 29th and they’ve played a team outside of the top 50 in just 13 of the 31 games, they’re 8-5 in those games. They posted a 1.1 run differential per game against the 20th hardest SOS.

Georgia: Series win over Vanderbilt. They rank 33rd in ELO with a 10-10 record against top 50. They have a top 40 schedule, and have won 63 percent of said games.

Oregon State: Very similar to Southern Miss, just with a power five schedule. They rank 36th in win percentage. They’re 35th in ELO with the 48th hardest SOS.

South Carolina: 11 wins against top 50 teams. 17th in ELO. 7th in SOS. 44th in adjusted run differential.

UC Irvine: Has the 37th highest adjusted run differential while winning 65 percent of their games. They only have the 107th ranked schedule but are 4-6 against the top 50 and 12-10 against the top 100 with an ELO ranking in the top 45.

Georgia Tech: Good enough record against the best teams. Only 21-18 with 6.6 runs for to 6.5 against. This is against the 29th hardest schedule. They get in ever so slightly.

Oklahoma State: 20th in ELO. Have the 41st highest adjusted run differential. 6-9 against quadrant run with .637 win percentage.

Alabama: 21st in ELO. Similar profile as Oklahoma State. 

Virginia Tech: Barely got in. Have dropped in ELO and are now 53rd. They get in barely thanks to a top 50 RPI, ELO, SOS and decent 10-11 record against the top 50. They also have a decent scoring margin and have won the majority of their games pretty handily.

Florida State: 27th in ELO, 9-8 against quadrant one, 9-5 against the next, and a top 50 SOS.

Virginia: Hace faced the 8th hardest schedule, hence the record of 22-21. They’ve managed to have a scoring margin of +0.7 against said schedule. The 11-17 top 50 record isn’t great, but given the 6-0 record against teams outside the top 100, I think they’ve just had some bad luck. They are 8-5 in one-run games, which is odd, so they could be in for some regression. They need to basically win 5 of 6 against Wake Forest/Boston College and do some damage in the ACC tournament to make themselves a comfortable at-large.

Kentucky: They have 15 losses, but 7 are by two runs or less. They’re 53rd in SOS, 44th in RPI and 40th in ELO. They have a quadrant 4 loss, but it was a two-run loss to Missouri. The argument against them is a 4-6 road record, and losing record against the top 150 teams. But they’re 26-15 with a good schedule, so they get in.

South Alabama: The final mid major team making it in. They’re on the cusp in the ELO ranking, at 51st, but they’re one of the few teams not chosen to have a win percentage (against top 50 ELO teams) at or better than .500, something only 13 percent of the country can say they’ve done. They’re just 25-15, but have an SOS ranked 81st. The case against them is an 11-8, 8-2 record in the bottom two quadrants. Considering the other choices available have far worse records and quadrant one records, they get the node.

Duke: They are the final team to get in, one of the teams in consideration were Campbell, whom Duke beat by 7. They’re just 19-18, but 38th in ELO and the 24th highest SOS. They have a run differential of 0.0 (5.8 for/against) which isn’t ideal, but they’ve beaten 7 teams in the top 50.

Just Missed (in no order): Clemson, Campbell, Kansas State, Auburn, Arizona State, Michigan, Long Beach State, Iowa.

Leave a Reply

Powered by

Up ↑

%d bloggers like this: