Wins Above Bubble
What is it, essentially?
Wins Above Bubble, or WAB. The goal of WAB is t measure how much better your team is compared to a typical "bubble" team. If your WAB is higher, you've done better than that typical bubble team.
A few more details, please...
It subtracts:
- Your record, how you performed against your schedule (your actual W-L), and
- The expected record of a typical bubble team if they played your schedule (what their hypothetical W-L would be).
So a positive WAB value means you have done better than what a bubble team would have done against your schedule. A negative value means that they would have done better than you did. So ultimately, it helps tell us whether a team's record is more legitimate, or that the record is inflated by a weak schedule.
What can it *not* do?
WAB is one of the "results-based" metrics -- it is not meant to predict scores. There are other methods for that (Sagarin, KenPom, Massey, and dozens more). So that puts it in the category of systems like the RPI; Colley Matrix; Wilson; or Rothman (and dozens more).
How do you calculate it?
Hold on tight. To calculate WAB yourself -- for any league or season -- you need a few things.
- You obviously need a set of games, with outcomes. So you will have everyone's W-L-T record.
- And, you need a ranking system that will allow you to calculate win probabilities for games. Do I have a 95% chance to beat this really weak team? Do I have a 30% chance to pull this upset on the road? A method to calculate these is crucial.
- You need to choose a "typical bubble team" strength to compare against. For me, in college basketball, I take that to mean the teams that are ranked between 40th and 50th. So I might use the average ranking of those teams as my typical bubble team strength.
- Then, for each opponent on a team's schedule, ask: how would the bubble team have performed instead? What would their probability of winning this game have been?
Here's an example set of calculations for Indiana. These are made-up numbers, so please relax if you think I've offended your team.
| IU's opponent was | IU's actual result | Probability of a "typical bubble team" beating my opponent |
Difference |
| Purdue (really good) |
IU wins, so 1.00 | 44%, or 0.44 | 1.00 - 0.44 = +0.56 |
| Oregon (really good) |
IU loses, so 0.00 | 17%, or 0.17 | 0 - 0.17 = -0.17 |
| Penn State (really bad) |
IU wins, so 1.00 | 76%, or 0.76 | 1.00 - 0.76 = +0.24 |
Now, total up those differences for every game on a team's schedule. "I actually had this set of wins and losses playing these opponents. And a typical bubble team would have had this other expected total number of wins if it played my schedule. The difference is my team's WAB." For the three games above, Indiana has a WAB of 0.56 - 0.17 + 0.24 = 0.63. They have done slightly better than what a typical bubble team would have done against these three made-up opponents. That's good -- it's a resume booster.
I mean, getting the actual outcomes will be easy. But where do these probabilities come from?
Yeah, this is the tough part.
Quick Links
• Want to create your own ranking?
• Want to browse other ranking systems?
• ...including this author's?
• More to add here...