RPI SOS

[wapiti]

I was looking at the RPI rankings and noticed that Incarnate Word is the number one rated Strength of Schedule.
They have 3 non-Div 1 opponents they have already played. How are they ranked as having the toughest schedule?
Their schedule does not pass the eyeball test as being a difficult schedule.

How is the SOS calculated? or is the RPI messed up? or am I missing something?

[clenz]

RPI and SOS is worthless until Christmas, at the extreme very earliest.

It takes nothing about future opponents into account, and only looks at D1 opponents.

Here's a must read on the topic

http://kenpom.com/blog/index.php/weblog/entry/rpi_help

The RPI is calculated by adding three parts.
Part I (25% of the formula): Team winning percentage. For the 2005 season, the NCAA added a bonus/penalty system, where each home win or road loss get multiplied by 0.6 in the winning percentage calculation. A home loss or road win is multiplied by 1.4. Neutral games count as 1.0. More on the effect of these changes can be found here (http://kenpom.com/blog/index.php/weblog/an_rpi_with_road_rage/).
Part II (50%): Average opponents’ winning percentage. To calculate this, you must calculate each opponent’s winning percentage individually and average those figures. This is NOT calculated from the opponents’ combined record. Games involving the team for whom we are calculating the RPI are ignored.
Part III (25%): Average opponents’ opponents’ winning percentage: Basically taking all of the opponents’ Part II values and averaging them.
Only games against other teams playing a mostly D1 schedule count when computing the RPI.
Here’s a simple example of how it’s calculated:
Team A has played two games and beat Team B and Team C.
Team B has a 1-1 record. Team B beat a team with a 2-1 record and lost to 2-0 Team A.
Team C has a 1-2 record. Team C beat a team with a 0-2 record, lost to a team with a 2-0 record, and lost to Team A.
So now let’s see how to compute Team A’s RPI.
At 2-0, their winning pct. is 1.000.
Their average opponents’ winning percentage (OWP):
Team B is 1-0 (1.000) without the loss to Team A. Team C is 1-1 (.500) without the loss to Team A. So the OWP is (1.000 + .500)/2 = .750.
Their average opponents’ opponents’ winning percentage (OOWP):
To do this, we need Team B’s and Team C’s OWP. When removing Team B’s results, Team B’s opponents are 2-0 and 1-0 for an average of 1.000. Team C’s opponents are 0-1, 1-0, and 1-0 for an average of .667. The average of these two numbers is .833.
So Team A’s RPI is 1/4 X (1) + 1/2 X (.75) + 1/4 X (.833) = .8333
Here’s how other columns in the RPI grid are computed…
SOS (Strength of Schedule): This is the last two components of the RPI formula:
(2/3) X Opponents Winning Pct. + (1/3) X Opponents Opponents Winning Pct.
Non Conference RPI: This is computed applying the basic RPI formula only to a team’s non-conference games. A common misconception is that a team’s non conference RPI will remain the same after they have played their last non-conference opponent. This is not true because a team’s non-conference opponents’ records and their opponents’ records change as they play their conference games. This can have a substantial effect on the NCRPI. For example, in 2003 Alabama’s NCRPI had a ranking of #23 when they entered conference play. By the end of the regular season it had risen to #3.
L10 (Last 10 Games): A team’s record in their last 10 games against Division I opponents.
RD/NT (Road/Neutral Record): A team’s record away from its home court.
How important is the RPI?
From the NCAA web site:

The Rating Percentage Index (RPI) was created in 1981 to provide supplemental data for the Division I Men’s Basketball Committee in its evaluation of teams for at-large selection and seeding of the championship bracket.
The RPI is intended to be used as one of many resources used by the committee in the selection, seeding and bracketing process. It never should be considered anything but an additional evaluation tool. No computer program that is based on pure numbers can take into account subjective concepts, e.g., how well a team is playing down the stretch, what the loss or return of a top player means to a team, or how emotional a specific conference game may be.
Several independent elements are combined to produce the RPI. These elements are a part of the statistical information that can be utilized by each member in an objective manner.
Each committee member independently evaluates a vast pool of information available during the process to develop individual preferences. It is these subjective opinions—developed after many hours of personal observations, discussion with coaches, directors of athletics and commissioners, and review and comparison of objective data—that dictate how each individual ultimately will vote on all issues related to the selection, seeding and bracketing process.
While the various elements of the RPI are important in the evaluation process, the tournament bracket each year is based on the subjectivity of each individual committee member to select the best at-large teams available and to create a nationally balanced championship.
To read more about the details of the tournament selection process go to this page (http://www.ncaasports.com/basketball/mens/story/6985142). Prepared to be bored though.
Why are the RPI ratings different on other pages?
I don’t know and I don’t care. For most of the 2005 season, every site (including this one) that posted the RPI differed significantly from what the NCAA was using. That saga is documented on this page (http://kenpom.com/blog/index.php/C20/). As of August 2005, ESPN continues to carry bogus RPI data for last season. I am confident what I post is correct, subject to game data errors which are rare but do occur. Now that home/road wins matter, small errors can occur if the NCAA considers certain games neutral that I do not.
How does the RPI factor in margin of victory?
It doesn’t.
How does the RPI handle home site advantage?
As of the 2005 season, the RPI gives more weight to road wins and home losses (see above description of Part I). Prior to 2005, site of game was not a consideration in the formula.
Any more complaints about the RPI?
I’ve written a little on the RPI in an effort to defend it, here (http://64.62.152.83/blog/index.php/weblog/defending_the_rpi_part_1_of_3/), here (http://64.62.152.83/blog/index.php/weblog/defending_the_rpi_part_2_of_3/), and here (http://64.62.152.83/blog/index.php/weblog/defending_the_rpi_part_3_of_3/). It doesn’t mean I think it’s perfect, it’s far from it, but it’s better than most folks think. Especially considering the selection committee is there to clean up any messes, although they do fail occasionally.