How My College Basketball Résumé Rating (BidW) Works
BidW scores what every team's results have earned, one game at a time, on a scale where 100 is the last team that usually gets an at-large bid. NCAA Basketball
What BidW measures
BidW is short for Bid Worthiness. It asks the question the selection committee asks about every team that didn't win its conference tournament: have this team's results earned it a spot anyway?
It looks at results and nothing else: who you played, where, whether you won and by how much. How talented a team looks, who's injured and what the polls think all stay out of it.
A score of 100 is the typical last team to get an at-large bid. Every 10 points is about one win, so a 90 is roughly a win short of the line and a 120 has about two to spare. The best résumés in a season clear 200.
- NET
- The NCAA's own team ranking and the one the committee sorts by. BidW uses it to rate how hard each opponent was.
- At-large bid
- A tournament spot the committee hands out to a team that didn't win its conference's automatic bid.
- The bubble
- The teams fighting over the last few at-large spots. BidW measures everyone against them.
- Team sheet
- The one-page summary of each team's season that the committee works from. All of BidW's inputs are on it.
How a score is built
Every game gets graded against what a bubble team would have done in the same spot, then the grades get added up.
The odds in step 1 come from a curve I fitted to six seasons of real games: 21,620 men's games, and a separate curve for more than 21,000 women's games. Neither curve ever saw who the committee picked, so nothing in BidW is tuned to the committee's habits.
With the field growing to 76 teams, eight more at-large teams get in, so the 100 line now sits a few spots lower on the list than it did at 68. NET comes out in early December. Until then Bracket Studio shows last season's final BidW, labeled with its year.
One résumé, up close
Every team's card in Bracket Studio lists the results that added and cost it the most.
Every team's card in Bracket Studio lists the results that added and cost it the most.
Reading a BidW rank
Here's every team from 2021 to 2026 that needed an at-large bid, grouped by where it ranked in BidW, and how often the committee took it.
BidW and my bracket model
They answer different questions, and it helps to keep them apart.
It grades results, game by game, and tells you what the résumé says. Predicting the committee is the model's job.
Half BidW, half an average of the other team-sheet ratings. The committee seeds on how good a team is as well as what it's done, so the model needs both.
Adding BidW to the men's model cut its at-large misses from 15 to 12 over 2021 to 2026.
On the Forecast page, every bracket, mine included, gets a count of the at-large teams it took that BidW would have left out. The real committee did that 17 times in 219 at-large bids from 2021 to 2026, about three a season.The real women's committee did that 15 times in 215 at-large bids from 2021 to 2026, two or three a season.
See how each bracket compares · How my college basketball model works
How it's been tested
I ran BidW against the real fields from 2021 to 2026, one season at a time. For each season the curve was fitted on the other five, so no season got to grade itself. Then I filled that season's at-large spots in BidW order and checked them against the committee's.
That matched 202 of the men's committee's 219 at-large picks.That matched 201 of the women's committee's 215 at-large picks. The test uses end-of-season NET, which includes the tournament, so it has a little hindsight built in.
What's different for the women's field
- Its own curve. The odds in step 1 come from more than 21,000 women's games, fitted separately from the men's.
- A longer margin cap. The women's test came out best with margin counted up to 15 points, so a 15-point win is a full win. The men's cap is 10.
- It's the only résumé number. The women's team sheets carry NET and nothing else, with no Wins Above Bubble, KPI or SOR, so BidW is the one résumé rating that field has.
The scale is the same for both: 100 is the typical last at-large team in.
Where the data comes from
Every game result and every opponent's NET rank comes from the public team sheets on Warren Nolan, which mirror the ones the committee uses. I collect them nightly. They go back to 2021, which is why the record starts there.
Those sheets also carry KenPom, T-Rank, KPI, SOR, BPI and the NCAA's Wins Above Bubble. BidW uses none of them. Some come with terms that rule out building on them, and a rating built out of other people's ratings wouldn't be mine. You can still see them all in Bracket Studio, and my bracket model averages them into its other half.
Why build it when WAB exists
The NCAA already prints Wins Above Bubble on the men's team sheets, and it asks a similar question. I built BidW anyway for three reasons:
- I can show every game that moved it and by how much. The published number doesn't come with that.
- It's mine, so I can set its bubble for the 76-team field and change it when the evidence says to.
- The women's team sheets carry no WAB at all, and women's BidW gives that field a résumé number it didn't have.
What it can't see
- Anything off the scoreboard. Injuries, who played in which game, how a team looks in March next to November, and whatever the committee hears in the room.
- NET's mistakes. Opponent strength comes from NET, so if NET has a team wrong, every résumé that played it inherits the error.
- How good a team is. BidW measures what a team did. The committee also weighs how strong a team is, and that's why BidW's order and the real seed list part ways, and why my model blends BidW with the predictive ratings.
Every team in Bracket Studio carries its BidW and rank, and its card shows the games that moved it most. Sort the pool by BidW and build your own field from there.
Open Bracket Studio