The writing was probably on the wall Sunday night when the selection committee announced the 16 host sites for regionals. Those host sites very closely followed RPI: 14 of the 16 teams selected were in the top 16 and the two that weren’t ranked 17 (Oklahoma State) and 19 (Auburn).
When the full NCAA Tournament field was announced Monday at noon, again RPI appeared to be the leading factor. The selection committee picks 34 at-large teams and all but two of those teams ranked in the top 41. The two that weren’t ranked 45 (Arizona) and 47 (Louisiana-Lafayette). Left out were Nos. 42 (Georgia) and 43 (Louisville), which had 20 conference losses apiece, and No. 46 (No. 44 Xavier won the Big East Tournament and, therefore, an automatic bid).
There’s grumbling every year that the committee relies too heavily on RPI. What is the point of a selection committee, after all, if things like conference record or the regional advisory committees or how a team finishes the season—all data points the committee is allowed to use to build the field that aren’t captured in the overarching RPI metric—are seemingly not going to be considered? And it’s hard to see how exactly those other data points were considered this year.
“I don’t think RPI is the best straight-line indicator,” one mid-major coach said. “There’s value to it for sure, but it’s dangerous to base everything off RPI. But that seems to be what they’re doing.”
Most glaringly, Arizona (RPI 45) got in the tournament after finishing eighth in the Pac-12 standings, five games behind Southern California (RPI 53) and 4.5 games behind Arizona State (RPI 52). Yes, the Wildcats reached the Pac-12 Tournament championship game and, yes, the conference record the committee looks at includes not just regular season conference games, but also any game a team plays against a conference foe. That includes conference tournament games, and any games conference foes play outside the conference setting, like Arizona’s midweek game against ASU. That meant the committee saw Arizona as a 16-19 Pac-12 team instead of a 12-18 one, but it’s still the first Pac-10/12 team with as many as 19 conference losses to get an at-large bid in the super regional era.
While Arizona’s inclusion was unusual, the deference to RPI wasn’t. What was truly different on Monday, however, was committee chairman John Cohen appearing on the selection show on ESPN2 and openly calling for RPI reform.
“There needs to be some statistical experts to come in and help us recreate this thing,” Cohen said. “I think basketball has done a great job of changing their RPI to the NET and moving it in a different direction.
“The thing about baseball is it’s a geographical sport so there’s advantages to warmer weather, we know that in the South, Southeast and out West. I would love to see it changed.”
Basketball ripped up its own RPI formula in 2018 and introduced the NET (NCAA Evaluation Tool). It includes a component much like RPI that adjusts winning percentage to give more weight to road and neutral site wins than to home wins and it, also like RPI, is influenced not only by a team’s adjusted winning percentage, but that of its opponents and its opponents’ opponents. What’s different is that the NET includes an efficiency metric (an advanced metric meant to be predictive that’s based on Bill James’ Pythagorean expectation) and margin of victory (up to 10 points).
While I’m not a college basketball expert, I think it’s fair to say that NET has generally mixed reviews. It seems that those in the sport like it more than RPI but don’t view it as a perfect metric. And that’s generally how the basketball selection committee treats it: as a tool, not as the tool.
Getting baseball to a similar place would make a lot of people in the sport happy. The trouble is getting baseball to that place.
College basketball has many privately run metrics meant to rank teams. Most of them are predictive metrics—which RPI is not—but there was a robust space for the NCAA to draw from when creating NET. It brought some of those people into the fold to help it create its own metric. College baseball does not have such a robust space. Boyd Nation, who runs BoydsWorld.com, has long produced his own Iterative Strength Rankings. Kenneth Massey, whose football ratings were a part of the old Bowl Championship System, has ratings for baseball. But college baseball is generally not awash with clear alternatives to RPI.
Such a metric could be created, but as Cohen said later on a conference call with reporters, it won’t be simple.
“It’s easy to have an idea, it’s really difficult to make it come to fruition,” he said. “We need some great mathematical minds to come together.”
Cohen won more than 600 games in 16 seasons as a head coach at Mississippi State, Kentucky and Northwestern State before becoming an athletic director at Mississippi State and now Auburn. He knows college baseball as well as any administrator in the country.
So, what does he think belongs in a revamped RPI?
Cohen said he wants to see a system that disincentives canceling games for RPI purposes and one that gives more credit for teams that travel long distances to play road games. The cancellation of games in an attempt to preserve RPI has long been a quiet part of the end of the college baseball season but has come into sharper focus in the last couple years. Some coaches have proposed dropping the worst few RPI games from a team’s metrics, which hockey does. That would, in theory, make coaches less likely to cancel games simply because they don’t want to damage their RPI.
If baseball were able to adopt rules that disincentive dropping bad RPI games, it would solve a PR problem but probably wouldn’t do much for actual RPI reform since it would affect every team in the country. SEC teams would get the same benefit as a team like Minnesota, which has few midweek options and just has to schedule what it can. Cohen’s second goal would be more likely to effect actual change.
The question would be how to reward teams who travel long distances. Presently, a road win is worth 1.3 wins, a neutral site win is 1.0 and a home win is 0.7. Before basketball changed to NET, its RPI credited teams with just 0.6 for a home win and 1.4 for a road win. Baseball, when it last changed its RPI formula a decade ago, declined to go that aggressive, noting that baseball is an outside sport, and its schedule is built around series and not single games.
The easiest thing would be to get more aggressive in rewarding road wins. Tweaking the formula to where basketball was would change the value of a home three-game sweep from 2.1 wins to 1.8. Even if that formula change only applied to non-conference games, perhaps it would encourage more teams to go on the road early in the season or reward those teams that have to for weather-related reasons.
One prominent coach from a conference that’s outside the Power Five said he thought Cohen’s ideas made for a fine theory, but more research is needed.
“It feels like there should be a deeper dive into the math,” he said. “I would say this is nonscientific, but I’d look at a penalty of playing a certain percentage of non-conference games at home. If 95% of your non-conference games are at home, there should be a penalty for that.”
Doing anything to RPI won’t be easy because changing anything in college baseball never is. But perhaps after this Selection Monday, when the SEC got eight hosts and many coaches around the country were left scratching their heads about the relative standings of several teams, there’s enough impetus to make it happen. Cohen has one more committee meeting—in August—and he said he’s confident the committee will “fully investigate” RPI reform.
Will it happen? As with so many things related to baseball and the NCAA, here’s hoping, but maybe don’t hold your breath.