WrestleStat Rankings - Final Regular Season - 2018-2019

[andegre]

The final regular season rankings have been generated.

Wrestler Rankings: https://www.wrestlestat.com/rankings/starters

Dual Rankings: https://www.wrestlestat.com/rankings/dual

Tournament Rankings: https://www.wrestlestat.com/rankings/tournament

Redshirt Rankings: https://www.wrestlestat.com/rankings/redshirts

Also, I made some adjustments to the RPI rankings to follow closer what the actual formula uses. It is as close to the real thing as we can get, without the NCAA telling us what the real formula is...

RPI Rankings: https://www.wrestlestat.com/rankings/rpi

 

[Old Number Nine]

Whatever:
125: Pic (28-0) should be #1 over Lee (16-2) who he beat.
133: Fix (27-1) should be #2 over Suriano (20-3) and Desanto (16-2) both who he beat
174/184: Where's Jacobe Smith
197: Where's Weigel
285: White should be #2

Dual Rankings:
OSU (15-0) should be at #2 over Iowa (14-1) who we beat and Ohio State (12-2) because of record

 

[osudog]

Dumbass ranking system

 

[ionel]

Dual Rankings:
OSU (15-0) should be at #2 over Iowa (14-1) who we beat and Ohio State (12-2) because of record

Actually dual rankings should be OSU #1, they beat the Big 10 champions. Its not our fault that the Big 10 won't wrestle everyone in the own conference, plus Cael seems to be the one who has messed up the national duals, doesn't wrestle the top teams, etc, not interested in participating. If the Big 10 declared Iowa champions then they must be their conference best team or equivalent to any other co-champ, if such exists. Given OSU spanked Iowa but squeaked by Mizzou, I'd put Missouri at #2 and then Iowa #3, just leave PSU off the dual list with not even a participation award.

But seriously regarding ranking systems. Wrestlestat is a bit like Flo, folks have suggested for a couple+ years now simple things (mathematical adjustments) they could do make the rankings meaningful. The like Flo don't listen. Its a great source of info, easy to find dual and wrestler record info easily, but yeah the rankings mean zip.

 

[andegre]

Wrestlestat is a bit like Flo, folks have suggested for a couple+ years now simple things (mathematical adjustments) they could do make the rankings meaningful. The like Flo don't listen. Its a great source of info, easy to find dual and wrestler record info easily, but yeah the rankings mean zip.

There have been massive changes and improvements to the rating algorithm this last offseason, as well as two additional fixes/enhancements during this season (I believe weeks 12 and 14). We are continually running simulations to improve the outcome of the ratings.

The version that was used last season scored dead-last in the RTR (Rank the Rankers) scoring matrix that a poster on TheMat/TheOpenMat forums did, and by a WIDE margin. Thousands of simulations went on last offseason and we got our score to go from dead-last (and well behind the 2nd to last place ranking system), all the way to 1st.

I'm VERY anxious to see what the results look like this season. We may not end up in 1st, but I believe we will be way up there when it's done.

 

[ionel]

There have been massive changes and improvements to the rating algorithm this last offseason, as well as two additional fixes/enhancements during this season (I believe weeks 12 and 14). We are continually running simulations to improve the outcome of the ratings.

The version that was used last season scored dead-last in the RTR (Rank the Rankers) scoring matrix that a poster on TheMat/TheOpenMat forums did, and by a WIDE margin. Thousands of simulations went on last offseason and we got our score to go from dead-last (and well behind the 2nd to last place ranking system), all the way to 1st.

I'm VERY anxious to see what the results look like this season. We may not end up in 1st, but I believe we will be way up there when it's done.

look at the dual team rankings

 

[Wrassler17]

There have been massive changes and improvements to the rating algorithm this last offseason, as well as two additional fixes/enhancements during this season (I believe weeks 12 and 14). We are continually running simulations to improve the outcome of the ratings.

The version that was used last season scored dead-last in the RTR (Rank the Rankers) scoring matrix that a poster on TheMat/TheOpenMat forums did, and by a WIDE margin. Thousands of simulations went on last offseason and we got our score to go from dead-last (and well behind the 2nd to last place ranking system), all the way to 1st.

I'm VERY anxious to see what the results look like this season. We may not end up in 1st, but I believe we will be way up there when it's done.

andegre, just curious, did you test how accurate your rankings were for the Big10 tourney last year (or maybe try this year)? It might be interesting to see how close your rankings/predictions are to the seeds and final results. If you have the time of course.

 

[andegre]

look at the dual team rankings

Dual rankings are not influenced by actual results (I know most will disagree with that).

They are based 100% on the starters ratings, and then running each dual through the prediction engine to determine a winner.

So, for the dual rankings, all 76 teams will get run through the prediction engine against the remaining 75 teams. That's how it generates the ranking. If there's a tie, like OSU is tied for 4th with Michigan with 72 anticipated wins, the tie-breaker is determined by the sum of the starters elo rating.

 

[andegre]

andegre, just curious, did you test how accurate your rankings were for the Big10 tourney last year (or maybe try this year)? It might be interesting to see how close your rankings/predictions are to the seeds and final results. If you have the time of course.

That's basically what the RTR scoring matrix is that the guy came up with.

Take the rankings for the wrestlers prior to the NCAA tournament. Then compare their actual finish. Points awarded for the result.

So, in that sense, yes, that's what all of the simulations used and we're now able to get a really high score (1st with last years data).

But to answer your question specifically, no, the B1G tournament isn't used (NCAA's is).

Edit: here's a link to the thread where everything is explained in the RTR process. I believe it's this same thread that shows how poorly WrestleStat did last year.

http://board.themat.com/index.php?/topic/17133-rank-the-rankers/

Edit: that first thread didn't have the results, here's that thread:

http://board.themat.com/index.php?/topic/17451-rank-the-rankers/

The modified version of WrestleStat's algorithm scored a 571 after we made all of the changes this last offseason.

 

[Wrassler17]

That's basically what the RTR scoring matrix is that the guy came up with.

Take the rankings for the wrestlers prior to the NCAA tournament. Then compare their actual finish. Points awarded for the result.

So, in that sense, yes, that's what all of the simulations used and we're now able to get a really high score (1st with last years data).

But to answer your question specifically, no, the B1G tournament isn't used (NCAA's is).

Edit: here's a link to the thread where everything is explained in the RTR process. I believe it's this same thread that shows how poorly WrestleStat did last year.

http://board.themat.com/index.php?/topic/17133-rank-the-rankers/

Edit: that first thread didn't have the results, here's that thread:

http://board.themat.com/index.php?/topic/17451-rank-the-rankers/

The modified version of WrestleStat's algorithm scored a 571 after we made all of the changes this last offseason.

Gotcha. I was asking if you did the RTR (or plan to) for Big10s in addition to nationals. And it wasn't to make more work for you...

I was thinking Big10s presumably involves more HTH rematches with guys being in the same conference. So it might tell you something about the H2H weighting in the overall ELO rating. As an example, you have Carr ranked below Moran & McKee even though he beat them both this year. If he beats one (or both) again at Big10s, maybe that's useful info for the H2H weighting? Just a thought.

 

[andegre]

Gotcha. I was asking if you did the RTR (or plan to) for Big10s in addition to nationals. And it wasn't to make more work for you...

I was thinking Big10s presumably involves more HTH rematches with guys being in the same conference. So it might tell you something about the H2H weighting in the overall ELO rating. As an example, you have Carr ranked below Moran & McKee even though he beat them both this year. If he beats one (or both) again at Big10s, maybe that's useful info for the H2H weighting? Just a thought.

Oh, I'm NOT planning on doing anything for B1G's; but I like your idea of analyzing those results for future enhancements.

Once March 1st rolls around, I'll be pretty busy monitoring/maintaining all of the Tourney Pool stuff...

 

[Old Number Nine]

Your web site has a lot of nice features but it can't be relied on because your rankings are not good.

I've forgotten the match, but earlier in the season I was very busy and didn't have time to preview the match very much and I only looked at your head-to-head review. I took from it that we would have no trouble in any match but to my surprise, several of the matches were quite close and we lost a couple as more accurate rankings proved to be much closer.

I believe you should keep all the functionality and rely on other rankings or just use you own head to set rankings. Maybe you could do something like Rotten Tomatoes and write and algorithm that combines all the existing rankings.