Stick with me for these next few paragraphs, because while they’re technical, they show how we can
game the system – and how the system currently games us.
If I were to make computer rankings to seed the NCAA Tournament, I’d base them upon how you fared
against your schedule compared to how the average NCAA Tournament team would be expected to fare
against your schedule.
These rankings would be based on win probability and not opponent record. What does that mean? Say
Team A beats RPI No. 3 Michigan State (28-8) and RPI No. 344 Binghamton (2-29). In my ranking they’d
get credit of say, 85 percent for beating Michigan State – I estimate the average NCAA Tournament
team would lose to Michigan State 85 percent of the time – and 0.3 percent for beating Binghamton,
as a tournament team would only lose 3 in 1,000 times to that squad. Combine a team’s performance
in each game into a season-long rating and tada, we have a rational way to select and seed the
tournament. Indeed, most credible computer rankings (read: not the RPI) rank teams based upon similar
logic.
Instead, we have the RPI, which is one-half your opponent’s win percentage, one-quarter your
opponents’ opponents’ win percentage and one-quarter your win percentage. (Ignore your own record and now you have the formula the NCAA uses to calculate strength of schedule.) On the surface this sounds reasonable, the more you win and the more your opponents win, the better you’ll be. However,
in the above example, MSU and Binghamton are a combined 30-37, a losing record.
So you would actually lose ground on your opponents’ win percentage and, with Binghamton in
such a crappy conference, your opponents’ opponents’ win percentage.( Sure, your record would go up
if you won both games, but that’s only 25 percent of the formula.)
Bottom line: Your RPI would likely go down for beating MSU and Binghamton. You would probably be
ranked higher if you hadn’t played those two teams at all.
Instead, you would have been better off playing two middling teams who happened to have good
records, say No. 117 Norfolk State (25-9) and No. 136 Georgia State (22-12). Those teams were a
combined 47-21, so you would get a lot of credit in the RPI and SOS rankings for wins there (and your
ranking might actually improve with losses, remembering that your own record is just 25 percent of the
formula). A fairer set of rankings, however, would recognize that an average NCAA team would only
lose to those teams, say, 20 percent of the time, so you’d rightfully get more credit for beating MSU and
Binghamton.
Life, however, isn’t fair, and neither is the RPI. It’s high time we got onboard.
How can you game this system? Simple: play crappy teams with good records. How do you get
absolutely screwed by this system? Simple: play crappy teams with crappy records.
Okay, the non-math majors can tune back in: it’s time for some actual numbers. With its NIT
championship, Stanford finished No. 69 in the RPI and No. 101 in the associated strength of schedule
(SOS) ranking, the two numbers the committee most heavily leans on when selecting and seeding the
field. That RPI number is really unfair to us; we finished No. 32 to KenPom, No. 51 to Warren Nolan, and
No. 29 and No. 41 in Sagarin’s two calculations. The difference is the disproportionate weight the RPI
puts on our awful out-of-conference schedule.
This season, Stanford played seven teams from subpar conferences with sub-.500 records: Central
Arkansas (6-20), Fresno State (11-20), UC Davis (4-26), Pacific (9-19), Seattle (9-15), San Diego (11-18),
and Bethune-Cookman (16-17). All told that’s a 66-135 mark.
Had we been able to replace those seven teams with teams nearly as overmatched but with an
opposite 135-66 mark, we would have finished 21st in the RPI and 20th in strength of schedule!
That’s a 48-team jump in the RPI and a 81-team jump in strength of schedule, even though our schedule
wouldn’t have had to get that harder. We could have substituted with Norfolk State and Georgia State
from above, two games we should win. We could add in-state Long Beach State (24-8) and San Diego State (24-8) – two teams that should absolutely become staples on our schedule moving forward. Heck,
we could have added any of the 25-30 teams with RPIs between 50 and 150 and at least 20 wins, all of
whom an NCAA contender should handle: Denver (22-9), UT-Arlington (23-9), Mercer (26-11), Loyola
Maryland (24-9) and on and on. (Note: I gave us one extra loss and one fewer win in reflection of the
slightly tougher schedule and was generally conservative in my estimates, so the real results could have
been better yet.)
Alternatively, had we simply not played UC-Davis and Central Arkansas, we would have finished No.
55 in the RPI and No. 65 in SOS, 14- and 36-team jumps respectively. Put another way, playing and
beating UC-Davis and Central Arkansas dropped us 14 slots in the RPI and 36 slots in SOS.
A wonderful case study is Colorado State, who despite a worse postseason performance (one and done
in the tourney), a worse conference, and a significantly worse record (19-12 vs. 26-11), finished the year
with an RPI of No. 29 and a SOS of No. 13, light years ahead of us. We beat those guys by 12, yet we
missed the Tourney by the mile, while they made the Big Dance, almost exclusively because of their RPI
and SOS.
They’re a 19-11 team from a so-so conference who never received a top-25 vote the entire season, and
whose best wins were against Colorado and San Diego State. By the head-to-head result and the eyeball
test of schedules, I think we were a better team than they were. (Here is Colorado State’s schedule. Here is ours. Judge for yourselves.) So how
did the Rams do it?
Simple: their out-of-conference opponents racked up great schedules. Seven of their 13 non-Mountain
West opponents would finish with at least 21 wins, while only one team, Northern Colorado (8-18),
finished with fewer than 10 wins. (They did play a Division II team in Nebraska Omaha, but by the
strange NCAA logic, Nebraska Omaha could have been 0-30, because Division II games don’t count in
the RPI.) We went 10-2 out of conference, while Colorado State went 10-4, had but we and the Rams
switched out-of-conference results, they definitely would not have been in the tournament, while we
very well might have been dancing on the big stage.
Anyhow, if I (and many others) can point out the importance of getting out-of-conference teams with
awful records off the schedule, there is absolutely no excuse for the athletic department and basketball
front office. Luckily, our scheduling patterns can change overnight.
I know people will point to the fact that we’re going to play tough out-of-conference teams next year.
That’s not the point at all. We played tough out-of-conference teams this year (Syracuse and Butler, for
starters), but we finished 48 slots lower in the RPI than we could have had we simply replaced the bad
teams on our schedule with nearly as bad teams who happened to have better records. Again, it’s
not about playing star teams in or out of conference; it’s about avoiding the god-awful ones.
People will also point to the fact that we can’t know in advance how many games each team will win.
That’s true, but we can use common sense and make pretty educated guesses. I bet we win more games
than Utah and USC next year. I bet Duke wins more than Boston College (9-22). I’ll also bet some team
which has won 20-plus and made the NCAA Tournament five of the last seven years from a league
you’ve never heard of is going to do better than whoever finished seventh in their league.
It’s crazy that beating Binghamton and Michigan State would lower your ranking, that not playing UC-
Davis and Central Arkansas would raise your ranking 14 slots, or that replacing bad out-of-conference
teams with slightly better but still bad out-of-conference teams would catapult your RPI from No. 101 to
no worse than No. 20. It’s also crazy that because of these gaping holes in the RPI logic Colorado State
made the NCAA Tournament, while Stanford, a better team with better results, was lucky to make the
NIT.
But this is the world we’d live in, and next year, I’d rather be a Colorado State than a Stanford, and even
if we exceed expectations and the Tourney is assured, I’d rather be a five-seed than a 10-seed. All the
Athletic Department has to do is make some strategic decisions and some phone calls.
Bob Bowlsby and the basketball front office should consider all the blood, sweat and tears the players
and coaches expend each season to try to scrape out one extra win, and contrast that with how much
impact, easily the equivalent of five wins, they could achieve with some critical thinking.
P.S.: If you’ve got friends in high places, please pass this along.