Talk:Rating percentage index

Strength of Schedule
The SOS calculation -- at least for basketball -- is 100% your opponent's record. There is no 2/3 and 1/3. That's a common misconception that was confirmed to me by NCAA Selection Committee admins during an NCAA Mock Selection invitation in Indianapolis a few years ago. I calculate the SOS 100% on opponent's record and it is congruent with the NCAA's official numbers. I can only speak to mens basketball. As of this writing, the SOS was carried over from the old RPI and has not changed, even though MBB now uses the NET team rankings. Here is a current RPI/SOS ranking with accurate calculations for SOS based on my notes here: http://udpride.com/images/rpi.htm I can re-write the Wiki page, but would rather someone else take the initiative. --70.60.14.146 (talk) 22:12, 2 March 2020 (UTC)


 * @ 85.18.30.105 (talk) 09:25, 12 January 2024 (UTC)

realtimerpi
I have removed realtimerpi from the external links section *repeatedly* because it is unnecessary with the official NCAA RPI above it.

The RPI calculation is so simple (I can almost do it in my head) that there are many sites on the Internet that have the RPI available in realtime: this in itself is non-notable. You may know of some media outlets that quote their data, but I have seen articles that quote the RPI of each of kenpom.com, warrennolan.com, boydsworld.com, collegerpi.com, espn.com, rpiratings.com and teamrankings.com. As such, being quoted by the media is also non-notable.

You may have an affection for this site (or a commision from it), but if wikipedia needs a link to realtimerpi, then it should have links to all the rest of the sites that provide this commodity service, and that would be over the top for a 2 1/2 paragraph article.

Thanks, 208.127.59.165 07:16, 6 March 2007 (UTC)
 * I disagree with your removal. The NCAA's RPI numbers are only updated weekly, and it may be useful to our readers to have a source which updates daily (or more frequently) such as RealTimeRPI or Ken Pomeroy's site. Either of these is frequently cited in mainstream media. I personally favor Pomeroy's site as it has fewer ads, but I think it is not out of place to have one of those links. (ESkog)(Talk) 14:43, 6 March 2007 (UTC)
 * OK, then we will go with Pomeroy then. I am agnostic as to the site but it should be free (which rules out Jerry Palm's CollegeRPI) and cited in mainstream media. I would prefer if the site was a mainstream auto-updated site but ESPN and the like charge for their real time versions. By the way it helps your credibility greatly if you register for a user name. Thanks, Calwatch 02:15, 7 March 2007 (UTC)

Wow, RPI can also be used for ranking any leaders, can it? hoh hoh.....
look forward to seeing the applicaion in other fields —The preceding unsigned comment was added by 72.52.66.10 (talk) 06:18, 18 March 2007 (UTC).

But, but is it a validated method?....I mean if it is validated by sets of training and validation data, or by another relevant method?

If the article can not provide any scientific references on the method validation, please indicate it somewhere in the article. Otherwise, it is misleading the public. I have seen an article abstract at

http://www.bepress.com/jqas/vol2/iss3/3/

However, I'm not sure what the ranking results are using RPI and OLRE. If the ranking results are consistent with each other by these two methods, then RPI is a validated method. Any people who is able to access to the journal article, please add some relevant info to the front article

Question
Are the standings prior or after the game? -- Howard  the   Duck  17:27, 3 July 2008 (UTC)

Deletion due to copyright
This article is the copyvio according to Wayback. The other article dates back to '04. Listing for speedy. --MWOAP (talk) 21:10, 13 December 2009 (UTC)
 * Wayback only shows the main home page of the site - which was last archived at the end of 2004 - but it is not the current home page, so the site has changed pages in the last 6 or so months. Also if one plays around enough, you can find the php definition files at http://www.btnsports.com/cbb/, which show that rpicalc.php date is 9th March 2009, which is after the data was added to the Wikipedia page. One could therefore conclude that the web site has used data from Wikipedia. In fact all reachable directories show all file dates post Feb 2009 - almost looks like a relaunch of the site. I have therefore removed the speedy tag.  Ron h jones (Talk) 21:52, 13 December 2009 (UTC)


 * I am sorry, this is about the third or fourth time I have marked something like this incorrectly. I got a hold of the Wayback tool, is there anything else I should know. Please let me know. --MWOAP (talk) 01:46, 14 December 2009 (UTC)


 * Easy done. Been there myself (the only way to learn is sometimes the hard way!). I now always look in the page history to see how long the text has been in place here, if it's many months, then I would have thought that the CorenSearchBot would have found it first, so I look much deeper, if it was created yesterday - then it probably is a CopyVio.  Ron h jones (Talk) 01:53, 14 December 2009 (UTC)


 * Thanks A lot!--MWOAP (talk) 01:57, 14 December 2009 (UTC)
 * I've tagged it as a backwardscopyvio. In addition to the above, there is compelling evidence in the history. Some of the language in the other page is included from the very beginning: March 2004. A year later, we have additional text added by an IP that is clearly used by the external source and later developments of our article. Almost two years later, we seem some heavy copy-editing that also made it into the external site. These are strong indicators of reverse infringement. Though written for administrators, Copyright problems/Advice for admins has some good advice for checking for backwards copyvio. Well, I think it's good advice, but I may have a COI.;)) --Moonriddengirl (talk) 14:21, 15 December 2009 (UTC)

This shouldn't be deleted, and that's not why I tagged it. This topic is clearly notable and shouldn't be speedied. It should be overhauled with new and original language that cites reliable sources. Unless I am just misunderstanding what you all are talking about ... Daniel J Simanek (talk) 21:09, 14 December 2009 (UTC)
 * A contributor saw your copypaste tag and tagged the article for deletion for copyright concerns. Copyrighted material must be deleted, in accordance with policy. However, investigation shows that the material originated here first. Reliable sources are always good. :) --Moonriddengirl (talk) 14:22, 15 December 2009 (UTC)

Adjustments
I beleive that while the initial formula is straight forward and you seem to have the details of that base formula. .25(WP) + .5(OWP) + .25(OOWP) 1.4's and .6's yada yada.

However, I'm fairly sure that there is als an adjustment when the NCAA uses theirs (bonus points if you will as well as penalties) for teams that play different percentages of Non-conf. games against teams ranked in the top half of the rpi. I'm working on finding a source for this, again. I used to know the specifics, but was trying to find them again this year. Again I'm fairly sure that the NCAA uses these fairly quantifiable adjustments, but I'll have to find the source again to verify.

Will watch your entry to see who finds those details first. —Preceding unsigned comment added by 76.210.61.248 (talk) 22:38, 16 December 2010 (UTC)

Reading through the short example that describes the WP, I think it states that a home loss counts as 1.4 Games, an away loss as .6 games if a team loses to Syracuse at home, beats them away, and then loses to Cincinnati away, their record would be 1-2. Considering the weighted aspect of the WP, their winning percentage is 1.4 / (1.4 + 1.4 + 0.6) If that is a correct reading, then the extended example shows Minnesota: (0 + 0 + 0) / (0.6 + 1.4 + 0.6) which would mean 1 home loss and 2 away losses however the table shows they lost at home twice and away once ... — Preceding unsigned comment added by Baldrix (talk • contribs) 04:33, 12 May 2015 (UTC)

Example Correction
(correcting my correction...OOWP is not agnostic of games against original team...so changed text below and removed example)

I have some issues with the example...some already pointed out. But my main beef is the calculation of the OWP should be "agnostic" of the Team for which the RPI is being calculated. So to list only four OWP numbers is incorrect...Our sample has four teams...so you would have an OWP for each team "with respect to" each of the other teams...so 4x3 = 12 different OWP calculations are needed to figure the RPI of all four teams. In the same way you would calculate 12 OOWP values using the 12 OWP values calculated.

I am wondering if the WP calculations are correct for the 4 teams listed in the Extended Example. The current calculation for UConn shows: (0.6 + 0.6 + 1.4 + 0) / (0.6 + 0.6 + 1.4 + 1.4) = 0.6500. I believe it should be UConn: (0.6 + 0.6 + 1.4 + 0) / (0.6 + 0.6 + 1.4 + 0.6) = 0.8125 if the denominator for Away losses is 0.6 instead of 1.4. If all games were weighted 1, the WP = 3/4 = 0.75. It would seem intuitive that if only loss is an Away game weighted at 0.6, then the WP would be higher than 0.75 not less. Maybe another way to summarize the 4 possibilities are 1) Home win: numerator = denominator = 0.6 2) Home loss: num = 0, den = 1.4 3) Away win: num = den = 1.4 4) Away loss: num = 0, den = 0.6 Nateman88 (talk) 19:39, 9 March 2016 (UTC)


 * You are correct regarding the WP calculations. I edited the page.  If I understand you correctly, I do not agree with your "beef" regarding OWP.  You are correct that there would be 12 different OWP calculations if each of the four teams play each of the other three teams.  However, in this example Kansas does not play Duke.  Therefore, only 10 different "individual OWP" calculations are required for this example.  If you look at the OWP example, you will see that all 10 "individual OWP" calculations are used.  Remember, the "total OWP" for a team is simply the average of its opponent's "indiviual OWPs" from all its games.  For example, Kansas has an "individual OWP" of 1.0 with respect to UConn as an opponent and an OWP of 0.5 with respect to Minnesota.  UConn has an OWP of 1.0 with respect to Kansas and an OWP of 0.6667 w.r.t. both Duke and Minnesota.  These OWP's naturally change as the season progresses.  Also, if a team plays another team two or more times, then that opponent's OWP is included in the average two or more times, respectively. JWKNYC (talk) 09:45, 6 May 2016 (UTC)