Division I women's volleyball · an open education project
The system used to rank teams isn't a neutral scoreboard — it's a policy choice. It decides who makes the NCAA Tournament, who earns a top-16 seed and hosts, who has to fly across the country, and even which non-conference games teams are willing to schedule. Change the formula and you change the bracket. This project exists to make that visible — to show the same season scored seven different ways, side by side, so fans, media, and administrators can see the trade-offs for themselves. Two of these systems — RPI and the KPI — are metrics the NCAA committee actually uses; the others are well-known alternatives.
We don't argue for a single "best" system. Each one answers a different question — who won the most, who is actually the strongest, who earned a bid — and each helps the sport in some ways and hurts it in others. The goal is an honest, unbiased look at all the angles: what's fairest, what's most accurate, what rewards ambitious scheduling, and what's best for the sport and the people who watch it.
The NCAA currently uses the RPI to help select and seed the 64-team field. Whatever metric is used flows into four real outcomes: which at-large teams get in, who is seeded in the top 16 (and therefore hosts the first two rounds), how the rest of the field is bracketed geographically (who buses, who flies, who has to face a powerhouse early), and — indirectly but powerfully — what teams are incentivized to schedule in the non-conference. A formula that punishes losing to a weak team teaches everyone to avoid risk; one that rewards beating a strong team encourages marquee early-season matchups.
These are the tensions worth arguing about. Reasonable people weigh them differently, and different priorities point to different systems.
An example of the tension. A mid-major that goes 30-2 and wins everything on its schedule may be rewarded by a resume metric like Wins Above Bubble — it captures "this team did everything asked of it," which is a strong argument for access. But a quality metric like PAVE, which is excellent for seeding the teams that are genuinely strongest, might rate that same mid-major lower because it never had to beat an elite opponent — so a PAVE-driven field could leave a deserving mid-major out. Neither is "right." One protects opportunity; the other protects competitive accuracy. That is exactly the kind of choice this project is meant to surface, not settle.
RPI is the NCAA's official metric and is listed first as the reference point. PAVE = opponent-adjusted set-margin ridge · Massey = set-margin least squares · PABLO = Bradley-Terry · Elo = sequential win/loss · WAB = wins above bubble. 2025 rankings exclude the NCAA tournament, so they show the selection-day picture.
64-team field = 32 conference champions (auto-bids, the same in every model) + the 32 best remaining teams by the selected model. Top 16 seeds host 4-team pods; teams are placed to minimize travel (bus < 350 mi, else fly), avoiding first-round conference rematches. "vs RPI" shows who this model swaps into and out of the field the committee's RPI would pick.
Method: PAVE (the best win predictor) is fit on games played to date, then the remaining schedule is simulated times. Each simulated season is then scored by the selector you choose above — the metric a committee would use to pick and seed the field. Bars show the share of simulations in which a team seeds (top 16) or makes the field. The projection holds current strength constant, so it is conservative for rapidly improving programs.
Every rating system can be gamed. This is exactly what it takes to earn the highest possible rating in each — the single biggest lever, what to do, what to avoid, and the kind of schedule it produces if a team optimized purely for that number. Read across them and the trade-offs jump out: some systems reward the marquee non-conference matchups fans want, and some quietly punish them.
"Produces" describes the scheduling behavior each system rewards. No system is only good or only bad — comparing what each one incentivizes is the whole point.