|
| -Overview |
The PICture 'RPI' Engine was developed at the request of a basketball fan who wants to publish RPI (Ratings Percentage Index) data for sports teams on the Internet. The default PICture 'RPI' Engine is my independent implementation of the 25-50-25 standard RPI algorithm model. These percentages are adjustable with optional arguments to the program. If you feel that the NCAA's original 20-40-40 RPI algorithm model was better, then you can calculate with that model by supplying the appropriate arguments when the program is started. While a spreadsheet and/or database has more than enough capacity to handle the management of the data for teams and their respective scorings, the RPI calculation and its associate the SOS (Strength of Schedule) calculation, while straight forward on their own, become fairly complex in their relationships one to the other. What with opponents' and opponents' opponents' and all. For comprehensive descriptions of what the RPI and the SOS are, and their calculation formulas, visit, CollegeRPI.com - College Basketball Ratings Percentage Index (RPI) at http://www.collegerpi.com/rpifaq.html and, Notes on My RPI Calculations at http://members.tripod.com/veneziano/rpi/notes.html These two sources seem to calculate very nearly the same results. One of those sources hint at where their two methods could differ in averaging results and the ramifications of each. Their results should converge as more games are played. Again. you have the choice of averaging methods with optional arguments to the program.
"We have met the opponents' opponents and he is us!"*
In order to test my perception of the formulas involved, and the procedures to attempt to minimize data input errors, I developed this C++, Win32, Console application. I call it the PICture 'Rpi' Engine. This application is fairly easy to learn and use in conjunction with your spreadsheet and/or database. While I am still experimenting between the two calculation methods mentioned above, the method I have adopted seems to be outlined in the Google Group's thread... The Official RPI Document. While I have no idea whether or not this is indeed an "official document", it does seem to indicate one method of averaging that could be used. This method is consistent with the method that we would use in the surveying industry for averaging sets of data. It should yields results consistent with one or both of the methods outlined in the above paragraph.
A word or two about myself. I am not a statistician. I am a surveyor who 'backed' into computing and programming as a logical extension to a surveyor's toolset. Most of my experience is chronicled on my homepage. I soon learned the value of accurate datasets. Garbage in, garbage out, is still true. While researching the various RPI ratings sites, I determined that there was no readily available tool that the average person could use /her own ratings of his/her favorite league, high schools, whatever. To this end, I decided to develop a freeware/shareware RPI /or an online Java2 applet/application to do the calculations. The PICture 'RPI' Engine will become the heart of any new program. It may not happen this century though.
*"We have met the enemy and he is us.", Pogo by Walt Kelly
|
| - |
|
times.
