RPI is unintelligible

[MDW79]

1 Kansas
8 Kentucky
9 Butler
18 Ohio State
27 Purdue
30 Louisville
41 Xavier
89 Dayton
95 Cincinnati
147 Miami
167 Indiana


This may be a rhetorical question, but could someone please explain to me how this is even possible? I know we have not played the toughest schedule (not even close) but I don't understand how out of whack it can be this time of year.

Leaving out the obvious facts that we beat Dayton, and boast a 12-0 record as opposed to a 10-3 record. Dayton has not played a ranked team all year. Now, I'm sure some of their wins are better than ours, but it's not as if they're great. They beat a Seton Hall team, who is way down right down, an average Ole Miss team, and that's really it. Unless we're giving huge points for Western Carolina and Akron?

Now he's what really irks me, and it should be no surprise. How in the **** is Xavier's RPI 41? Seriously? Not joking, almost every year, and in every week, I ask myself that same question. Why is X's RPI what it is? Even later in the season. I know our schedule is beyond bad right now, so let's just throw us out of the equation. But can someone explain this to me? Look, does this look like the 41st best team:

Overall record 7-3

Wins: Western Michigan, IPFW, Iowa, Seton Hall, Wofford, Butler, Wake

Losses: Old Dominion, Miami, Gonzaga

Some will point to Iowa, Seton Hall, Butler, and Wake as "quality" wins and Gonzaga a "quality" loss. The records of those teams:

7-5, 6-6, 9-4, 6-6, 7-5. Old Dominion may be a decent loss, but they're 0-1 in the Colonial Conference, feature no ranked wins, and are Old Dominion for Christ's sake.

So the question is, where Xavier getting bumped so much? Those wins, and losses, I listed are against near .500 competition. X features zero ranked opponents on their schedule. The computer's doesn't play the "name game" so how are the wins worth a crap? Not too mention, and I know it's not accounted for, but half of those wins came by a very close margin. (Wofford 3OT???)

I don't mean to turn this into a bash X thread. (Feel free though) But I have no clue how that they are over 50 spots higher than us, and 40 more than Dayton. (Is X's schedule that much stronger than Dayton's, that 7-3 trumps 10-3 by 45+ spots? And is Dayton schedule that much better than ours, that it trumps a head to head loss, and 12-0 record?

And I guess both answers are "yes". Which goes to show me that the RPI is a useless system this time of year, and, maybe, all together. Like I said earlier, I'm sure I'll look at this even come March, and go "Why is X -- in the RPI?' Maybe it just my bias, but half the time, it beats the **** out of me. I consider myself at least somewhat knowledgeable on this stuff. But this, I can't make heads or tails out of this, today.

[GoBearcats31]

I've been looking at another site with similar ratings (X is 42, UC is 94) and Xavier was actually #21 before the Gonzaga loss. Iowa, SHU, Wake all are mediocre at best.

But not to talk too much about Xavier, though my next point is Xavier-related. Someone posted on Twitter a week or two ago how UC's S.O.S. was 330-something or whatever but was projected to be around 61 or 62 (one spot ahead of Xavier). This is further proof how demanding of a schedule a Big East team is forced to play the last 10 or so weeks of the year.

Everything will play out - UC will show they are good or prove that the naysayers right in that maybe they are not that good. RPI now (and Bracketology and regular AP and coaches polls) is worthless. It's still fun to look at. If none of it existed, it might make certains games less interesting, or it could inflate a team's win. Xavier fans, for instance, might act like they beat a Chris Paul- or Tim Duncan-led Wake team without realizing they are not that good in actuallity.

[Josh Rexhausen]

MDW, you helped me so I'll try to help you.

Remember RPI is only about winning percentage. Thats all. I think it's 25% your winning percentage, 50% Opponents winning percentage and 25% Opponents Opponents winning percentage. So if a 12-0 UC team played 12 teams whose records were 7-5, and all those teams played teams who were 7-5, the RPI is weighted as 75% 7-5 (in general, I don't know how they drop opponents games against the rated team or if the rated team is part of the Opp Opp record) and 25% 12-0. IT DOESN'T MATTER WHO YOU BEAT, JUST WHO YOU PLAY. A 9-1 team who beat a 9-1 team but lost to a 1-9 team gets the same RPI as if they beat the 1-9 team and lost to the 9-1 team. This is why it is only a tool that the committee, along with quality wins. Also I mentioned before it is not an indicator of anything during the season except the level of competition played so far. I like Sagarin who takes into account how much you won by and therefore has more credibility in my book.

[MDW79]

I've been looking at another site with similar ratings (X is 42, UC is 94) and Xavier was actually #21 before the Gonzaga loss. Iowa, SHU, Wake all are mediocre at best.

But not to talk too much about Xavier, though my next point is Xavier-related. Someone posted on Twitter a week or two ago how UC's S.O.S. was 330-something or whatever but was projected to be around 61 or 62 (one spot ahead of Xavier). This is further proof how demanding of a schedule a Big East team is forced to play the last 10 or so weeks of the year.

Everything will play out - UC will show they are good or prove that the naysayers right in that maybe they are not that good. RPI now (and Bracketology and regular AP and coaches polls) is worthless. It's still fun to look at. If none of it existed, it might make certains games less interesting, or it could inflate a team's win. Xavier fans, for instance, might act like they beat a Chris Paul- or Tim Duncan-led Wake team without realizing they are not that good in actuallity.


Yeah, I know, all good points. I just don't get how X is getting an inflated ranking for that schedule. Before the season that schedule may have looked better, but in actuality, it's really not all the good. It actuallykinda sucks. Frankly, I think it should be a tough sell, to say they played a strong OOC, fo try counter their conference's lack of quality opponents. (Opposite of us, what they TRY to do most years)

I don't know, sometimes it just seems off. Even leaving us out of the mix, b/c our schedule is so bad. I would think Dayton should at least be equal with X. 10-3 vs 7-3, should equal out their schedule disparities, which I still don't see much more in X's. Apparently that's not the case though.

[MDW79]

MDW, you heled me so I'll try to help you.

Remember RPI is only about winning percentage. Thats all. I think it's 25% your winning percentage, 50% Opponents winning percentage and 25% Opponents Opponents winning percentage. So if a 12-0 UC team played 12 teams whose records were 7-5, and all those teams played teams who were 7-5, the RPI is weighted as 75% 7-5 (in general, I don't know how they drop opponents games against the rated team or if the rated team is part of the Opp Opp record) and 25% 12-0. IT DOESN'T MATTER WHO YOU BEAT, JUST WHO YOU PLAY. A 9-1 team who beat a 9-1 team but lost to a 1-9 team gets the same RPI as if they beat the 1-9 team and lost to the 9-1 team. This is why it is only a tool that the committee, along with quality wins. Also I mentioned before it is not an indicator of anything during the season except the level of competition played so far. I like Sagarin who takes into account how much you won by and therefore has more credibility in my book.

Ok, cool, good stuff. I do always seem to forget some of that. So, question: Would it be safe to say Xavier is getting more of bump b/c of who their opponents played thus far? Cuz I would guess the winning % of the teams on X's schedule is not that strong. So I'm guessing, it's gotta be the teams those teams played, and the teams those teams played, etc.

Is that right, am I following correctly? (Ex a 6-6 team, that has played a really, really good schedule may be a better opponent, in terms of lifting your RPI, than a 9-2 team with and average schedule)

Thanks for breakdown. Btw, Josh, that ESPN3 had me about doing cartwheels. I couldn't believe how simple it was and how pumped I am to see the slate of games they have. Enjoy! :D

[Bearcat Cafe]

MDW, you helped me so I'll try to help you.

Remember RPI is only about winning percentage. Thats all. I think it's 25% your winning percentage, 50% Opponents winning percentage and 25% Opponents Opponents winning percentage. So if a 12-0 UC team played 12 teams whose records were 7-5, and all those teams played teams who were 7-5, the RPI is weighted as 75% 7-5 (in general, I don't know how they drop opponents games against the rated team or if the rated team is part of the Opp Opp record) and 25% 12-0. IT DOESN'T MATTER WHO YOU BEAT, JUST WHO YOU PLAY. A 9-1 team who beat a 9-1 team but lost to a 1-9 team gets the same RPI as if they beat the 1-9 team and lost to the 9-1 team. This is why it is only a tool that the committee, along with quality wins. Also I mentioned before it is not an indicator of anything during the season except the level of competition played so far. I like Sagarin who takes into account how much you won by and therefore has more credibility in my book.

Close. Its not just who you play its where you play them. You forgot to include the weighted portion. Home wins count as 0.6, true road wins count as 1.4, neutral court wins count as 1 in the equation.

[Bearcat_DF]

Ok, cool, good stuff. I do always seem to forget some of that. So, question: Would it be safe to say Xavier is getting more of bump b/c of who their opponents played thus far? Cuz I would guess the winning % of the teams on X's schedule is not that strong. So I'm guessing, it's gotta be the teams those teams played, and the teams those teams played, etc.


Actually, you only have to look at the records of our opponents vs. X's opponents. This second factor of the RPI formula accounts for 50% of the rating (more or less).

Of Xavier's opponents - only 2 currently have losing records - Wofford 5-6, Miami-OH - 5-7.

Contrast that to UC's opponents - 7 currently have losing records including these whoppers: St Mary 3-10, Savannah 1-13, Toledo 3-9, St. Francis 2-10.

These percentages significantly pull this factor down.

I actually suspect the opponents of these teams (the 3rd factor) will be better than Xavier's.

[Josh Rexhausen]

You forgot to include the weighted portion. Home wins count as 0.6, true road wins count as 1.4, neutral court wins count as 1 in the equation.

Thanks I knoew that but my wife was pressuring me into doing christmas construction for the kids. I love building new toys for the kids!!

[steveocincy]

And I see X moved up to 35 from 41 following their Titanic victory over Albany (at home) last nite. I agree that the RPI is not a good indicator of how good a team is, but it is the commonly viewed ranking used in the media and by the NCAA committee.

[MDW79]

And I see X moved up to 35 from 41 following their Titanic victory over Albany (at home) last nite. I agree that the RPI is not a good indicator of how good a team is, but it is the commonly viewed ranking used in the media and by the NCAA committee.

I saw that. What a joke.

We did move up to 78 (From 91) though.

[GoBearcats31]

I saw that. What a joke.

We did move up to 78 (From 91) though.

One spot behind 4-7 Robert Morris (who has played a decent non-conference schedule, boosted by Cleveland State having a good RPI).

X is inflated because Butler is #8 despite being 0-3 against the top 50 and losing to Evansville.

So basically, your nonconference schedule is more important than your results at this point in the season.

[bearcatn8]

8 of our 13 wins have come against teams with an RPI of 250 or worse. That is just killing us right now from an RPI standpoint.

I really believe that the scheduling this year all comes down to $$$. It costs money to schedule home games against teams with relatively good RPI's (mid majors or low-mid majors expected to do well in their conferences) because everybody wants to schedule those folks. Given that the athletic department is broke, it can only afford to bring in teams that no one else wants to schedule.

[Bearcat Cafe]

UC jumped 19 spots today by beating DePaul. DePaul jumped 12 spots by losing to UC.

[waterhead]

UC is 26 in Sagarin and X is 78. That is much more realistic to me. No sure what Ken Pom has them both.

[STKohls]

There are numerous problems with the RPI, and they are well documented:

1) as Josh pointed out, only your record and opponents records matter, not actually who you beat.

2) Margin of victory is not included. A 20 point blowout matters exactly as much as a 1 point OT squeaker. This type of digital conclusion is fine for ELO chess (where all victories are the same) but clearly limited in its application to college basketball.

3) Recent games are not weighted any more than older games. A victory in November means exactly the same as a victory in March.

The RPI was originally intended as an end-of-year snapshot that the committee could determine quickly and easily. My understanding is that it was primarily used as a kind of tie-breaker when schools had similar records or resumes. It was a way to quantify what exactly a 22-10 record meant with the specific team's schedule. Unfortunately, simple computer programs and ubiquitous sports websites have morphed the RPI into some kind of mid-season daily barometer of a teams performance, something it is clearly unqualified to be.

The RPI is just a tool, and a very crude tool at that. Using it at this point in the season will be misleading at best.

[kpzero]

UC is 26 in Sagarin and X is 78. That is much more realistic to me. No sure what Ken Pom has them both.

35 and 77(X)

[bearcatbret]

UC is 26 in Sagarin and X is 78. That is much more realistic to me. No sure what Ken Pom has them both.

Ask the great one. He is by his own admission KenPom.

[steveocincy]

Yet another example of the value of the RPI, X went from 55 to 52 based on the spank-down they took last night. RPI = Who you play means everything... how you play means very little.

[ucat4]

1 Georgetown
5 Kentucky
11 Purdue
17 Ohio State
30 Butler
41 Dayton
47 Louisville
52 Xavier (up from 55 Thur)
55 Cincinnati (up from 66 Thur)
116 Miami
125 Wright State
194 Indiana

These RPI numbers were copied from Lance's blog.

So when exactly will the RPI actually make sense and be a true indicator? Late Jan, Feb, March?? Seriously, because right now both Dayton and Xavier are ahead of us which makes absolutely positively zero sense!!

And like Steveo pointed out - Xavier's RPI actually got BETTER after a 20 point beat down??

[STKohls]

1 Georgetown
5 Kentucky
11 Purdue
17 Ohio State
30 Butler
41 Dayton
47 Louisville
52 Xavier (up from 55 Thur)
55 Cincinnati (up from 66 Thur)
116 Miami
125 Wright State
194 Indiana

These RPI numbers were copied from Lance's blog.

So when exactly will the RPI actually make sense and be a true indicator? Late Jan, Feb, March?? Seriously, because right now both Dayton and Xavier are ahead of us which makes absolutely positively zero sense!!

And like Steveo pointed out - Xavier's RPI actually got BETTER after a 20 point beat down??

When will it be a true indicator? Never.

The RPI just a blurry snapshot, one that gets a little less blurry near the end of the season. It was derived decades ago when the most powerful processing tool the committee would use was a calculator. Virtually the only thing it's useful for is if you have 2 teams with near identical records, and want to know which team is better, and although it kind of does that, it certainly doesn't do it very well.

Moreover, consider the numbers: The RPI is 25% record, 50% SOS, 25% opponents SOS. Are these ratios statistically proven or valid in any way? No. Are they just easy to enter simple fractions that were some random guy's guess as to what was important? Yes.

Perhaps we should consider the RPI like the college basketball equivalent of the slide rule. Yes, it works in very specific circumstances if you know exactly what you're doing with it. No, you really shouldn't be bothered using it, since there are newer and more accurate tools.

[red_n_black_attack]

How is it that before the Shootout Blowout, X was ranked in mid 50's and we were ranked in mid 60's, and now they are ranked #49 to our #54 ranking?

Because we are better than people thought, losing to us moved them up - WTF???

I know the RPI only takes season up to this point into consideration, and is really only useful at the end of the season. Nevertheless, that is a very confusing cause and effect sequence.

[waterhead]

RPI is simply how many wins and losses you have. How many wins and losses the teams you play have. How many wins and losses do the teams that they play have. Margin of victory means ZERO in RPI. I don't know the actual multipliers but I would guess something like 1 point, 1/2 point, and 1/4 point respectively.

Sagarin and Kenpom are better indicators right now because they take into account more than just wins and losses. There we are 21 and 20 respectively. I don't think that is too far off from where we should be at this point. A win over Nova would do wonders for us and would probably put us in the top 15. It would also help our RPI a lot since they have a bunch of wins.

[waterhead]

stkohls thanks for the breakdown on the multipliers...

[CincyEngGrad02]

kenpom is the answer. I look at it daily.

[CaptainProbasco]

Current RPI top 10:

1) Georgetown
2) BYU
3) Kansas
4) Syracuse
5) Kentucky
6) Duke
7) UConn
8) St. John's
9) Pitt
10) SD State


That's right kids, BYU is currently 2nd in the RPI. And St John's is currently 8th. No further proof is necessary to prove how awful the RPI is as a computer system.

[shadyo]

kenpom is the answer. I look at it daily.

kenpom has UC #20 and Villanova #16. Sagarin has UC #21 and Villanova #20. I saw their game against Temple and I if we defend the perimeter as effecively as we defended the post last night, this game will be in reach.

[bearcatgary]

Good point. This season has enlightened me, also, to insanity of the RPI.

The RPI is:
a) 25% your own record
b) 50% your opponents record
c) 25% your opponents' opponents record

So for X:
a) went down (0.667 winning % to 0.615)
b) went up because UC's record is 15-0
c) probably went down because UC's SoS is low.

Apparently b) caused the overall RPI to go up more than a) and c) caused it to go down.

[steveocincy]

It's not that I don't know the basics of the calculation or what it means. The frustration is the use of the RPI by media, including analysts, that it is the equivalent of a computerized rating, such as Kenpom or Sagrin. The RPI really in no way indicates a computerized ranking. I wish the NCAA would drop the use of it.

[Josh Rexhausen]

Caps lock on:

REMEMBER THE RPI IS FOR DETERMINING WHICH OF TWO TEAMS, WITH SIMILAR CREDENTIALS WHO DID NOT PLAY EACH OTHER, IS BETTER BASED ON SCHEDULE. IF TWO TEAMS PLAYED EACH OTHER, THEY WOULD NOT USE THE RPI. IF THE SEASON ENDED TODAY AND THE LAST SPOT COMES DOWN TO x AND UC, UC WOULD GET IN OVER x.

Caps lock off.

Strange, even in caps lock I can't get a capital x.:D

[bearcatbret]

Remember that X had the same record at this time last year and still made the NCAA because of the powerhouse conference that they play in. Maybe both teams can make the tourney and be in the same bracket. It would be a great season to beat X twice in the same year.

[Josh Rexhausen]

Delete me.

[Lobot]

Remember that X had the same record at this time last year and still made the NCAA because of the powerhouse conference that they play in. Maybe both teams can make the tourney and be in the same bracket. It would be a great season to beat X twice in the same year.

I think I remember correctly that the committee sets the bracket so that no rematches happen before the S16

[red_n_black_attack]

I think I remember correctly that the committee sets the bracket so that no rematches happen before the S16

I thought that was only for conference games, can't meet before Sweet 16 if from the same conference. There are too many OOC games to have to also incorporate a system to avoid all rematches.

[bearcatbret]

The best run in my lifetime was when UC beat Memphis State four times in one season to make the final 4. One thread was to list the most hated, I would have to say coach Finch. He would throw his jacket and scream at the refs all game. We used to throw tube socks at him from the student section.

[Panamacat]

It's not that I don't know the basics of the calculation or what it means. The frustration is the use of the RPI by media, including analysts, that it is the equivalent of a computerized rating, such as Kenpom or Sagrin. The RPI really in no way indicates a computerized ranking. I wish the NCAA would drop the use of it.

I don't know that too many news organizations are using the RPI to rank UC at all. Rather I have seen the RPI stated to demonstrate how UC's "bad" scheduling might influence the way people (NCAA/voters) look back on this season. The RIP is only used by the NCAA at the conclusion of the season, and that is the only time at which it becomes a halfway decent reflection of the relative quality of teams. It is not a predictor. There are other sites for that......though people seem to like to dis those as well.

With regards to being behind Dayton and Xavier....after UC plays Villanova, and certainly by the end of next week we will be ahead of both teams. We have risen 51 places in the past twenty two days as our quality of competition has increased, and as some of those so-called weak teams have improved their results (note that Dayton is now 13-3 with a good SOS).