Must be getting close to match time....edges are starting to show.
You are right. It is not the end all....that’s why it is one of three measures used when determining how many allocations per weight per conference.
Coaches Rankings II and RPI I have been released
- Thread starter ccdiver
- Start date
You are right. It is not the end all....that’s why it is one of three measures used when determining how many allocations per weight per conference.
Did not intend to be rude or snarky; however, I would be curious to know what RPI formula is used rather than merely be informed that the "standard" formula is not used. So perhaps I was a bit frustrated in that regard. Again, no snarkiness intended.
Snarkiness. Lol. Funny word.
Your win % x opponents win % x opponents opponents win %. That’s it.
In response to your other post, yes, I'm familiar with the RPI metric. I was out-and-about, and on my phone...hard to post a longer response given what I was in the middle of...and was going to respond later. It's now later, and I see Cowbell responded.
For wrestling, it's simple math, though getting all the way to "opponents-opponents win %" is cumbersome and not worth the effort to me...just wait for the NCAA to publish the list, though I wish they would release the calculated RPI.
There is no home vs away factors, and yes, it is all about strength of schedule. In general, tougher conferences have an advantage, because guys in those conferences wrestle more highly-ranked guys, more frequently.
Lastly, yes, of course it's a tool, and no one said it was "be all-end all". Win % and Coaches Rankings have shortcomings too...though I prefer to look at the positives of all 3. They measure different aspects of success, and need to be considered together to understand a wrestler's body of work for the year.
RPI, Zain as an example (his win % x opponents win % x opponents opponents win %)
Zain has a win % of 100%
His opponent's win % is 57%
I don't know his opponents-opponents win %
So his RPI, today, is 100%x57%x"his opponents-opponents win %", or .57 x his opponent's-opponents win %, which probably calculates to a final number in the .30 to .34 range.
At an average of 20 bouts per wrestler say, and Zain having 21 bouts, I'd have to find the composite wins and losses of 420 bouts (20x21), to finish the math. Now imagine doing that for 760 wrestlers. The database used (NCAA's ISRF on the NWCA OPC system) keeps track all season, and probably is retrieved by the push of a button. Too cumbersome for anyone to do manually, as I said.
Last edited:
Thanks. That formula appears very flawed in that it's not weighted. So any single low component can really drive down the RPI value. Hard to believe, really.
Not sure I follow. The opponents win% and the opponent's-opponents win % are made up of the win percentage based on something north of 340 bouts, and 5000 bouts respectively. From a statistician's perspective, those are large sample sizes, and have a high level of confidence.
Keep in mind that the winning percentages of the opponents (and their opponents) are figured in, not the straight wins and losses. In this regard, the difference between Brandon Sorensen (just to use an example) wrestling Zain Retherford and a .500 wrestler is the same as the difference between him wrestling that .500 wrestler and Britt Malinsky, despite the fact that we all know that he has a .001% chance of beating Retherford, a 99.999% chance of beating that .500 wrestler, and a 100% chance of beating Mr. Malinsky. That's what he means when he says the RPI is flawed. The WrestlingByPirate Dual Impact Index, for what it's worth, suffers from a similar flaw (albeit to a lesser extent, due to the inclusion of bonus points) in that beating a REALLY bad wrestler, even by fall, will bring down your rating in the abacus.
Mine or Bruce's?You're words make it sound way more complicated than it really is...though I understand.
My comment was based mostly on construction of the formula, not sample size. I've never seen RPI expressed as purely multiplication of the three winning percentages. In this form RPI is always limited by the lowest of the three factors, and probably tends to converge in a relatively small range. Typically RPI is additive of three weighted factors. Unclear why college wrestling decided to differ in it's use of RPI from every other sport that I'm aware of. The method used by wrestling would appear to accentuate the built-in flaws of RPI (as noted by SHP), though I'm not certain about this. As noted before, RPI is not justified by any statistical basis.
once again, the Pirate is underestimating his abilities. Do you ever brag about anything?
once again, the Pirate is underestimating his abilities. Do you ever brag about anything?
I'd rather let my work do the talking.
Not sure I understand the post-conferences comment. He'll be the 3rd seed at Big Ten's, pretty sure.
