All things Rankings, NCAA RPI and Coaches Rankings

[HFO]

Well...things are looking up. Wrestlestat data converges to NCAA RPI final list of 33.
We are at 7 now and Logan could be knocking on the door at 165.

Lehigh Wrestling Rankings
Jan 30th, 2024
Wrestlestat

125 #4 Luke Stanich FR
133 #4 Ryan Crookham FR
141 #20 Connor McGonagle JR
#22 Malyke Hines JR
#26 Carter Bailey SO
149 #21 Kelvin Griffin FR
157 #29 Max Brignola JR
165 #49 Jake Logan JR
174 #63 Connor Herceg JR
184 #86 Jack Wilt SO
197 #3 Michael Beard JR
285 #10 Nathan Taylor SO

[Richb-3]

It might be a stretch but I suspect If Seymour gets in wins in his expected 2 matches, (Army+Princeton) maybe, maybe He could reach the top 30. Although a couple of open wins on Bucknell + AzSU could improve that possibility. Are there any more opens after Edinboro?

He is at 14 bouts that count. (obviously this only matters if something untoward happens to Stanich.

[JAD77]

Wrestlestat’s algorithm can be often be eerily accurate, though - unless I’m missing something, it is not used in calculating a wrestler’s RPI. Am I incorrect, or are you saying their algorithm is similar to that which is used for calculating RPI?
Thanks.

[jdalu75]

WrestleStat has a link to their attempt at duplicating RPI calculations. I always overlook it, probably because it's labeled "RPI" on their home page in an attempt to confuse me.

They also show the actual RPI calculation, which the NCAA does not:

Official RPI [and RPI Rank] is based on a minimum 15 Division I match results at the desired weight class
Formula: (Win% * 0.25) + (OppWin% * 0.50) + (OppOppWin% * 0.25)

Thus the actual RPI is a number less than 1.0.

The kind of thing I always wonder about -- the opponents do not need 15 bouts to be counted in the calculation. Does an opponent with 30 bouts count the same as an opponent with 1? And does the wrestler's own winning percentage count in the OppOppWin%, or does the NCAA exclude those?

[martinsilvestri]

The math is over my head but 141 is a little odd. Why would McG be rated so highly at 141? If anything, Bailey should be rated higher after beating McG and then Kai Miller at the F&M open.

[Oracle]

Because Wrestlestat is a piece of elephant dung.

[Spladle1989]

I find wrestlestat to be somewhat useful in reference for the guys outside the top 25......rankings inside the top 25 are not great

[HFO]

Wrestlestat’s algorithm can be often be eerily accurate, though - unless I’m missing something, it is not used in calculating a wrestler’s RPI. Am I incorrect, or are you saying their algorithm is similar to that which is used for calculating RPI?
Thanks.

Not rocket science. If you go year by year, but time final RPI is out, it is highly correlated to total individuals and to rankings between RPI and Wrestlestat. Check for yourself and run the data. How NCAA does RPI is their own process. How Wrestlestat does their ARMA model is their own process. Given that by final RPI you are looking at a pool of about 40 potential wrestlers per weightclass, the R-squared explanatory of what NCAA RPI or Wrestlestat are expected.

Wrestlestat gets knocked around, but it is about spot on by year end when enough data samples are in.

[HFO]

Because Wrestlestat is a piece of elephant dung.

I'll show you that it is not after NCAA. The explanatory statistics of All Americans in final wrestlestat rankings will be VERY HIGH. As is their correlation of wrestlestat rankings to NCAA RPI for who gets into the tournament.

All models are wrong but some are useful. George Box

[JAD77]

Wrestlestat gets knocked around, but it is about spot on by year end when enough data samples are in.

Yeah - that's the key thing - enough data. The more data Wrestlestat has, the more accurate it tends to be - as I said, sometimes almost eerily so.

The "elephant dung" comes in when there isn't a lot of data to work with.

Regardless of what one thinks about their rankings, one thing is for certain - they are the only "well known" wrestling rankings that are in no way, shape or form subjective. All the others have some sort of bias - unintentional or otherwise, built in.

Also, I love playing with their dual team matchups.