College Basketball Rankings

[Sea Ray]

Can someone explain to me how the University of Tennessee can have an RPI of #7 in the country, tops in the SEC, and they're no where to be found in the ESPN/USA Today top 25? They barely snuck into the top 25 in the AP poll. Last year the SEC got a lot of schools into the tournament (7?) and had two in the final 4. An SEC team also ended up winning both the NCAA and the NIT. What gives?

I'm not saying they deserve to be in the top ten but how 'bout at least the top 20 or 25? Depending on how they do in the SEC tournament, they could be in line for a #2 seed and not even be ranked in one of the national polls.

[Redsland]

Well, I'm not defending the poll, which probably should have Tennessee ranked somewhere, but in answer to your question, I'd say the Vols fell off the map when they were losing six of eight games in January, particularly to teams like Auburn and Mississippi. And besides earning splits against Florida and UK, the Vols haven't beaten a top team since the Memphis game, back in December.

If it's any consolation, bracketologists have them as a four seed at the moment.

[max venable]

It's the nine losses. Fair or unfair, the nine losses makes it tough for voters to rank anyone very high. The only other nine-loss team ranked ahead of Tennessee is Duke and...well...Duke is Duke.

[dabvu2498]

There are serious problems with the RPI that have really shown their faces here towards the end of this year.

Nice column from the Louisville paper on this subject:

Eric Crawford
RPI looks like a system stuck in the past

Imagine this scenario: The NCAA Tournament selection committee sits in its conference room, gathering data from its Commodore and Tandy computers.

Maybe the committee is thinking about mandating that the short shorts come back for this tournament or playing it without the three-point shot or shot clock.

Are those Members Only jackets they're wearing? Right. No way the committee would go retro. It's ludicrous.

So answer this: Why will the committee use a computer ranking that was devised in 1981, with only minor updates since?

This is the time of year when the letters RPI get more airtime than IRS. The Ratings Percentage Index will be the statistical flavor of the week.



It's a strange place, RPI Land. In this numerical nirvana, the University of Kentucky (20-9) ranks higher (10th) than defending national champion Florida (25-5 and No. 11).

Here, Nevada Las Vegas is ranked higher than Kansas and Brigham Young is better than Georgetown. Here, Vanderbilt can beat UK twice and yet still rank 27 spots below the Wildcats.

Can we agree, whatever the "R" in RPI stands for, it isn't reality?

The RPI was designed to measure strength of schedule and performance against that schedule. The problem is, this time of year the numbers are repeated so much that they seem a part of a team's DNA.

We in the media like numbers. They're easy. They're quick. They can illustrate a great deal in a short time.

Unfortunately, they also can paint a picture that isn't quite accurate.

Scrap the RPI?
That's why I agreed when I heard ESPN's Jay Bilas say last week that it's time for the RPI to R.I.P.

"They should get rid of it," Bilas said on ESPN's "College GameNight." "The RPI has taken on a life of its own because it affects perception. … We can do better than the RPI."

Bilas called for a "precision instrument" that would better measure the quality of wins when they happened. He argued, for instance, that UK's win over the University of Louisville looks good now because the Cardinals have climbed up the charts as the season has gone on. But at the time, U of L was struggling.

I wouldn't go that far. Neither would two big-time hoops number-crunchers: Jerry Palm of CollegeRPI.com and Ken Pomeroy of kenpom.com.

"I wouldn't make that a part of it," Pomeroy said yesterday. "But I don't understand why they haven't gone to a better system. I mean, you don't see people sitting around playing games on Atari systems anymore. Back in 1981, you didn't have all the TV coverage we have now, you didn't have the Internet.

The RPI worked great then. But now, we have so much data to figure out what the important factors are historically, last 10 games, and all the other things. Why not build a system that measures that?"

Forget the numbers
One legitimate fear is that the better the ranking, the more people will defer to it. The last thing basketball needs is a system like football, with the ranking the final word.

"This isn't the BCS, where the numbers are supposed to do the thinking, and furthermore, the committee doesn't want a set of numbers to do the work," Palm said. "The RPI is fine, given how it's used."

The RPI works for the committee because the committee knows what it is for and how to use it. That's not always the case in the media and public.

Because it is a number endorsed by the NCAA, the RPI is thrown around in a round-the-clock news, talk-radio and Internet cycle leading up to the tournament. That's the perception Bilas talked about.

And as long as all of us amateur bracketologists are going to play with it, we might as well have a ranking based in reality.

Reach Eric Crawford at (502) 582-4372 or ecrawford@courier-journal.com. Comment on this column, and read his blog and previous columns, at www.courier-journal.com/crawford.

[Redsland]

Duke may be Duke, but Duke sure doesn't look like Duke this year.

[Danny Serafini]

The RPI went screwy a couple of years ago when they changed the formula. Weighting it to give more credit to road wins is a nice concept, but they overdid it and sent the numbers wacky. It's not as reliable a tool as it used to be, and even then it wasn't the be all and end all of team evaluation.

[Sea Ray]

Originally Posted by Redsland

Well, I'm not defending the poll, which probably should have Tennessee ranked somewhere, but in answer to your question, I'd say the Vols fell off the map when they were losing six of eight games in January, particularly to teams like Auburn and Mississippi. And besides earning splits against Florida and UK, the Vols haven't beaten a top team since the Memphis game, back in December.

If it's any consolation, bracketologists have them as a four seed at the moment.

No question they fell off the map in late January. That was due to the loss of the SEC's top scorer, Lofton. Even with those losses, they have the 2nd best record in the SEC.

I agree the only nice wins they've had since December is FL and KY, but what other big teams have they played since Dec? One. OSU. And they lost on a last second shot in Columbus.

I agree with the bracketologists. I think they should be ranked nationally 15-20. The RPI is not w/o its flaws. I think it should weight road wins much more than home wins.

Traditionally UT has a hard time come mid March 'cause the school is more wrapped up in Spring football than basketball, IMO. They haven't even won an SEC tournament game in a few years. A big challenge for Bruce Pearl is to change that and keep the "basketball energy" well into March.

[Sea Ray]

Originally Posted by max venable

It's the nine losses. Fair or unfair, the nine losses makes it tough for voters to rank anyone very high. The only other nine-loss team ranked ahead of Tennessee is Duke and...well...Duke is Duke.

Virginia is also ahead of UT, but I have no problem with those teams' rankings. The ESPN poll has Vanderbilt, BYU, UNLV and 8 loss Texas ahead of the Vols.

[Chip R]

Originally Posted by Danny Serafini

It's not as reliable a tool as it used to be, and even then it wasn't the be all and end all of team evaluation.

That's the point. It's not the be all and end all. It's one tool that is used along with others.

[Sea Ray]

Originally Posted by Danny Serafini

The RPI went screwy a couple of years ago when they changed the formula. Weighting it to give more credit to road wins is a nice concept, but they overdid it and sent the numbers wacky. It's not as reliable a tool as it used to be, and even then it wasn't the be all and end all of team evaluation.

I guess you're right. They do weight road wins. This surprises me 'cause Tennessee has had few big road wins as their 6-9 record attests.

[dabvu2498]

Originally Posted by Sea Ray

No question they fell off the map in late January. That was due to the loss of the SEC's top scorer, Lofton. Even with those losses, they have the 2nd best record in the SEC.

But the Vols are the #3 seed from the SEC East. HAHA!

[Sea Ray]

Originally Posted by dabvu2498

But the Vols are the #3 seed from the SEC East. HAHA!

Let Vandy have a day in the sun every now and then. The Vols won the SEC regular season last year and this is likely as high as Vandy will ever get.

[dabvu2498]

Originally Posted by Sea Ray

Let Vandy have a day in the sun every now and then. The Vols won the SEC regular season last year and this is likely as high as Vandy will ever get.

How quickly we forget that Vanderbilt has been in the Sweet 16 more recently than the Vawls.

[Highlifeman21]

The RPI is a whole heck of a lot better than the BCS

[15fan]

Originally Posted by Redsland

Duke may be Duke, but Duke sure doesn't look like Duke this year.

now that conference regular season games are over, let's take a look at how things shook out in the acc.

north carolina (25-6 / 11-5) and virginia (20-9 / 11-5) tied for first.

maryland (24-7 / 10-6), virginia tech (20-10 / 10-6) and boston college (19-10 / 10-6) finished in a tie for 3rd.

georgia tech (20-10 / 8-8) and duke (22-9 / 8-8) finished in a tie for 6th.

now let's look at the current top 25 ranking for each of those 7 schools:

unc - #8 coaches / # 8 ap
virginia - #24 coaches / NR ap
maryland - #20 coaches / #17 ap
virginia tech - NR coaches / NR ap
boston college - NR coaches / NR ap
duke - #21 coaches / #21 ap
georgia tech - NR coaches / NR ap

so duke was at best the 6th best team in the acc this year according to conference records. yet according to both the coaches and the writers, they are the 3rd best team in the conference.

let's look at how the other major conferences shook out and the relative rankings of the schools.

big ten - ohio state won the regular season title (27-3 / 15-1) and is ranked #1 in both polls. wisconsin finished 2nd (27-4 / 13-3) and is ranked #3 in the ap poll and #4 in the coaches poll. no other conference teams are ranked.

big 12 - kansas won the regular season title (27-4 / 14-2) and is ranked #2 in the coaches poll and #2 in the ap poll. texas a&m finished 2nd (25-5 / 13-3) and is #7 in the coaches poll and #7 in the ap poll. texas finished 3rd (22-8 / 12-4) and is #14 in the coaches poll and #15 in the ap poll. no other conference teams are ranked.

pac 10 - ucla won the regular season (26-4 / 15-3) and is # 3 in the coaches poll and #4 in the ap poll. washington state finished 2nd (24-6 / 13-5) and is #12 in the coaches poll and #11 in the ap. oregon finished 3rd (23-7 / 11-7) and is #18 in the coaches poll and #16 in the ap.

are you noticing a pattern? national rankings in each poll generally follow the logic that the team with the best conference record has the highest national ranking in the conference. rankings then follow in relative order. except in the acc. for the sake of rounding out the argument, let's look at the rest of the bcs conferences:

in the big east, georgetown won the regular season (23-6 / 13-3) and is ranked #9 in the coaches poll and #9 in the ap. pittsburgh (25-6 / 12-4) and louisville (22-8 / 12-4) finished tied for second. pittsburgh is #11 in the coaches poll and #13 in the ap. louisville (with the slightly worse overall record) is ranked #15 in the coaches poll and #12 in the ap. notre dame finished 4th (23-6 / 11-5) and is #16 in the coaches poll and #20 in the ap. marquette is the final big east team to be ranked. they finished 5th (23-8 / 10-6) and are #19 in the coaches poll and #18 in the ap.

pittsburgh and louisville are in different relative order in the two polls. but they are both ranked behind georgetown (who had a better conference record) and ahead of notre dame & marquette who each had worse conference records. marquette and notre dame are flipped in the two polls, but the big east #4 & #5 teams are ranked behind the #1-3 teams.

in the sec, florida won the sec east (26-5 / 13-3) and is ranked #7 in the nation. tennessee finished 2nd in the sec east (22-9 / 10-6). they are #22 in the ap poll, and unranked in the coaches poll. no one in the sec west is ranked.

so there's certainly a logic with regards to conference standings and relative national rankings in the big 10, big 12, sec, big east, and pac 10. that's 15 of the 25 schools in the coaches poll.

only one other conference has 2 teams ranked. in the moutain west, regular season champ byu (23-7 / 13-3) is #23 in the coaches poll and the ap poll. 2nd place unlv (25-6 / 12-4) is #25 in both polls.

all of the other teams that are ranked in the top 25 are the lone representative from their respective conference:

memphis (conference usa) - #5 ap / #5 coaches
nevada (wac) - #10 ap / #10 coaches
southern illinois (mvc) - #14 ap / #13 coaches
butler (horizon) - #19 ap / #17 coaches
winthrop (big south) - #22 in the ap / #24 coaches.

so aside from some very minor quirks in the big east, there is a reliable pattern among non-acc teams that national rankings coincide with relative conference standings. if team x finishes higher than team y in the conference, team x should have a higher ranking in the national polls than team y. if you go with that line of reasoning to predict national rankings, you'll be correct most of the time.

except when it comes to duke and the acc.

the only rational reason that i can think of that would explain this is discrepancy exists is that duke must have played a more difficult non-conference schedule. coach k always gets more than his share of high school all-americans, so he shouldn't be afraid to play anyone anywhere, right?

here's a url that lists all of coach k's high school all-americans through 2005:

http://www.dukeupdate.com/Records/mc...allamerica.htm

right now, coach k has 6 mcdonald's all-americans - greg paulus, josh mcroberts, and demarcus nelson, as well as 2006 participants lance thomas, gerald henderson and jon scheyer.

so back to the duke non-conference schedule this year. coach k and his team of 6 high school all-americans have wins at home against...columbia, georgia southern, unc greensboro, davidson, indiana, georgetown, holy cross, george mason, kent state, san jose state and temple. they have a road win against st. johns in nyc. they have neutral court wins against gonzaga (in nyc), and air force (in kansas city). they have a neutral court loss against marquette (in kansas city).

that's one home win over a top 25 team (georgetown) and one neutral court loss (marquette) against a top 25 team. in fairness, gonzaga did just win their conference last night. none of the other teams, though are top 25. indiana and air force are right around 35 if the top 25 poll was expanded out to include the "others receiving votes" category. that's hardly the kind of non-conference schedule that would explain such a radical disconnect between finishing tied for 6th in the conference but having the conference's 3rd highest national ranking in 2 different polls.

so the only other conclusion i can logically draw is that the voters in the coaches poll and the ap poll behave in a rational manner when voting for 24 of the top 25 teams.

they behave irrationally, though, when it's time to vote for duke.