RPI SOS Update

[Chili]

From realtimerpi.com

Men's Basketball - Marquette Golden Eagles (2006-2007)

RPI Rank: 24    SOS Rank: 19     Record: 17-4 (5-2)
Conference: Big East

Not bad at all.

[mu_hilltopper]

I'm surprised by that a little, since the BE's SOS is still #7.

1     Atlantic Coast     0.5914    0.5665    2    12
2     Pacific-10     0.5907    0.5692    1    10
3     Southeastern     0.5901    0.5663    3    12
4     Big Ten     0.5774    0.5592    4    11
5     Missouri Valley     0.5759    0.5587    5    10
6     Big 12     0.5678    0.5445    7    12
7     Big East     0.5674    0.5523    6

[77ncaachamps]

But MU's played some tough off-season opponents.

[tower912]

Thank goodness we didn't paly UWM this year.  Would have dragged that sucker right down.

[Wade for President]

Sure is nice to see Texas Tech get 2 wins over Top 10 teams in a week!

Love the clips of Knight trying to encourage students not to storm the floor....act like you've been there.  Kinda hard to stop a fired up bunch of kids when you've knocked off your second Top 10 team in a week.  Let the kids have fun.

[Marquette84]

I'm surprised by that a little, since the BE's SOS is still #7.

Then maybe you should learn how individual team RPIs as well as the conference SOS are calculated.

[mu03eng]

^^^Well thats a shot across the bow^^^


I think what people never took into account was that yes we were playing no name buy schools in the non-conference season.....but they were good teams at "their level".  Those teams go back and are successful in their conferences.  Remember with the adjustment to the RPI mid-majors earn extra points for road losses that majors don't.  As a result teams like UMBC play us, get an rpi boost even though they lost, go to their conferences and beat teams....it rises our SOS somewhat.

[mu_hilltopper]

I'm surprised by that a little, since the BE's SOS is still #7.

Then maybe you should learn how individual team RPIs as well as the conference SOS are calculated.

I admit I did misspeak and say conference SOS when the table was actually conference RPI.

Is it wrong to think that the lower the conference RPI, the weaker the conference, and thus, the harder it is to have a high SOS within that conference?  Please explain, instead of only alluding to how stupid I am.  Thanks for that, by the way.

[Marquette84]

Ken Pom had a good explanation of the "Conference RPI" a couple of days ago.

An individual team's RPI is calculated based on three elements
1.  Winning Percentage (25%)
2.  Average Winning Percentage of all opponents (50%)
3.  The average of the opponents' oppnents winning percentage (25%)

The SOS is calculated in a similar manner, comprised of #2 (66%) and #3 (33%) above.

First Note:  the typcial reporing of RPI is rank, not the RPI itself.  MU doesn't have an RPI lf 24--its RPI is .6188.  Of course, .6188 the 24th highest RPI, 

Second Note:  The "Conference RPI" is simply the average of all conference members' RPIs.  From Pitt's .6609 to Rutgers .4897.  Add up 16 numbers, divide by 16. 

Third note:  The "Conference RPI" rank simply orders the conference average.

Now, why is it not surprising that our SOS has improved, even though the Conference RPI has not?  Simple:  The Conferce RPI includes USF, and we haven't played them yet. 

Fourth Note:  Conference RPI is not a smooth distribution.  The Big East is just about as close to #3 as it is to #8.  Its closer to being #1 than it is to being #9.  In other words, while a precise rank puts the Big East 7th, its by a very small margin compared to the conferences ranked 9th and below. 

Its like in baseball--sometimes the 2nd place team is  1/2 game behind, sometimes it's 21 1/2 games behind.  The Big East is more like the former.

One more thing--You have to know which element of an opponents RPI is driving them lower.  Lets say a team has a low RPI because they have a poor SOS but a good w/l record.  UConn is a goo example to use. When we play UConn, the weighting of the relevant RPI elements is reversed. 
--For UConn, 25% is for a good record against lousy opponents, 50% is based on those opponents average w/l record.   
--For MU, 50% is based on UConn's good record, 25% is based on UConn's opponents.

DePaul is the opposit type of team.  Worse Record--better SOS.  That puts them ahead of UConn in the RPI rankings, but a worse opponent for us.

And for the purists, yes, I'm glossing over the impact of the opponents' opponents average w/l record.  I've found that it's rarely that significant in calculating RPIs.

[spiral97]

Ah.. now THAT was SOOOO much more useful than insulting topper - thanks for posting it!

Good points about the distribution not being smooth (or linear)... hmm.. I smell some graphs coming on here... 

[mu_hilltopper]

After reading that, I'm afraid I still don't understand why it's wrong to believe, the lower your conference RPI rank, the more surprising it is you have a high SOS.  (As I said above, a little surprising.)

Obviously, exceptions occur, but in general?   Isn't it relatively easier for team X in a high RPI conf to have a high SOS, while it's relatively harder for team Y in a low RPI conf to have a high SOS, since the family of teams they play are weaker and therefore can't boost their SOS relative to everyone above you?   Certainly, each team has its own data, but can't any generalizations be made?

And agreed, from the chart, those 7 confs are a mere +-.03 points from eachother.   While that indicates things could change with a smallish number of wins in the right places, it doesn't change the rankings or counts of teams in conferences with better numbers, at this exact point in time, on a macro level.   They are close, no doubt, but those ahead, are on average, ahead, right now.

[WashDCWarrior]

Think of it this way.  If you take USF and Rutgers out of the conference, the BE would probably have the highest RPI of any conference.  Marquette hasn't yet played either, so our SOS is very high.  It is likely to drop some following those games.

[mu_hilltopper]

Certainly agreed.  If you look at the details, all things can be explained.  I don't believe that invalidates (to the point of derision) the "little surprise" I had when finding the BE RPI hadn't improved a rank or two.

[Marquette84]

Certainly agreed.  If you look at the details, all things can be explained.  I don't believe that invalidates (to the point of derision) the "little surprise" I had when finding the BE RPI hadn't improved a rank or two.

Sorry for the "derision," which I assume was mine--   

I'll offer the same defense as many others--as one of those who bleeds blue and gold, I misinterpreted the comment as an unwarranted attack on MU.   

After re-reading, I realize that it wasn't, and i apologize for my "over the top" reply.
 

[rocky_warrior]

Bringing this back to the topic, Thursday was a big RPI night for Marquette without even playing a game.  8 previous opponents played, 6 won.  As a result, Marquette's #21 in RPI, with an SOS of #15.  wow.