Well, I don’t think they need to be included for a variety of reasons. You’re pretty much supposed to beat a Division 2, Division 3, NAIA school if you’re a Division I team.
There’s PlENTY of RPI tweaks that not only would I like to see, but I think are NECESSARY to undo the horrific effects of what we have.
The RPI formula was invented in 1981, when the powerful teams of college basketball were spread out. You had 22 really good programs in 20 conferences/independents:
Big 8: Kansas
Big East: Syracuse
Big Sky: Boise St
Big Ten: Michigan State
Eastern 8: West Virginia
East Coast: St. Joseph’s
Independent: South Carolina, Dayton
Pac 10: UCLA
PCAA: Fresno State
Sun Belt: VCU
WAC: Utah, BYU
Well now, those same 22 teams are in NINE conferences: ACC, B12, BE, B10, SEC, A10, MWC, P10, WCC.
The RPI was a good tool back in 1981 because everyone played over 50% of their schedules against non-conference teams and about 40% of their schedules in 20 relatively evenly matched conferences.
And NOW everyone’s playing 18 conference games, 62% of their teams against conference opponents. The difference between “Wichita State’s MVC schedule vs Xavier’s MWCC schedule” in 1981 wasn’t that much. But the difference NOW between WSU’s MVC slate and Xavier’s BE slate is enormous.
Because everyone is playing 18 conference games, and EVERY CONFERENCE goes .500 against itself, the VAST MAJORITY of each team’s SOS doesn’t come from “who you schedule OOC.” It comes from “How the bottom team in your league does OOC.”
Those things need to be addressed.
Yet another thing that should be addressed is this: Every NCAA at-large candidate is expected to go about 8-1 against teams outside the Top 100 of the RPI. But the RPI counts the difference in record between every single one of those teams. If you beat a 7-22 team, you just played a .2414 team. But if you play a 12-18 team, it’s .4000. Who cares? You beat ‘em.
One of the secrets to game the RPI is what the Pac-12 did this year. They all scheduled the conference favorites of bad leagues. The teams that will finish 17-14 and be 175 in the RPI instead of the 6-24 teams that will be 300 in the RPI. And that’s why their RPIs are all ridiculously high this season; yet almost ALL their marquee wins are against each other. USC, UCLA both lost at home to Monmouth (Notre Dame and Georgetown lost to them too). The Pac-12 is not that strong. Monmouth’s RPI continues to fade.
And finally, when you have these power conference teams playing a bunch of nobodies OOC, racking up ridiculous OOC win percentages, then playing each other… these 6 power conference teams have about 42 of the top 50 of the RPI.. which is 6-8 teams each. Which means they’re playing 10 Top 50 games against each other. And the NCAA Selection Committee COUNTS Top 50 wins. As if Colorado’s 3-7 is better than Monmouth’s 2-2.
A fourth thing that’s jacked about the RPI is they made a switch to make road wins worth more than home wins. Which is “fine” I guess, but it didn’t encourage teams to go face people on the road OOC. It basically encourage all the big boys to play Neutral site games against each other. There HAS to be a better way. There IS a better way.
With the statistical revolution in baseball, where we have WAR telling you how much better than a replacement player everyone is… why can’t we develop a formula that:
#1 - Puts the value of win percentage and SOS in the correct ratios. I’d say 2-1 Win Pct to SOS, and eliminate Opp SOS. This eliminates the Conference effect and sets a 1-350 list of teams.
#2 - Sets a FLOOR on the SOS for bad teams you beat (say, .400). You beat a team under .400, you get .400 on your SOS. You lose to them? Count it fully. Now “bad losses” are built in.
#3 - Takes the average win percentages of teams in the Top 68 of our new RPI against each category:
Home vs 1-40, Road vs 1-40, Neutral vs 1-40 (NCAA locks!)
Home vs 41-80, Road vs 41-80, Neutral vs 41-80 (NCAA bubble!)
Home vs 81-120, Road vs 81-120, Neutral vs 81-120 (NIT/Postseason teams)
Home vs 121-200, Road vs 121-200, Neutral vs 121-200 (Respectable teams)
Home vs 200-350, Road vs 200-350, Neutral vs 200-350 (inferior teams)
Then breaks down each teams’ win percentage against each category, and assesses a value for “expected win percentage vs each group, assuming a balanced schedule.” and assigns everyone their expected wins against a 250-game schedule (10 home/10 road/5 neutral) vs each category. And it’s decimal place wins.
So, let’s say Kansas is 3-0 on the road against 81-120, and the average of Top 68 is .806. We wouldn’t make Kansas 10-0. We’d say they’d go between 1.00 and .806 in their remaining 7 games. You build in how much better than expected they should go based on what they’ve done so far.
Then we add it all up into one neutralized, expected win percentage. We could then multiple that win percentage by the number of games into the season we are.
On January 1st, a 12-0 team might be: 11.788
On February 1st, a 17-3 team might be: 16.677
On March 1st, a 22-6 team might be 20.011
On Selection Sunday, a 24-8 team might be 23.415.
1897-1898 | 1900-06 | 1926-27 | 1929-30 | 1939 | 1942