WrestleStat Rankings - Week 14

[andegre]

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

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

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

Statistical Rankings: https://www.wrestlestat.com/rankings/statistical

Hodge Watch: https://www.wrestlestat.com/rankings/hodge

(Create your own rankings) Tournament Projection Tool: https://www.wrestlestat.com/tourney/projection

 

[walterg]

I believe PSU won the dual on 2/3/18

 

[andegre]

I believe PSU won the dual on 2/3/18

I'm assuming you're referring to the 75-1 dual rankings record. That's what's generated when running Penn State vs all 76 other D1 teams through the prediction engine (dual comparison). That's how the dual rankings are generated.

For next season, I plan on swapping out ACTUAL dual results if they actually occurred during the season...

 

[KCLion]

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

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

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

Statistical Rankings: https://www.wrestlestat.com/rankings/statistical

Hodge Watch: https://www.wrestlestat.com/rankings/hodge

(Create your own rankings) Tournament Projection Tool: https://www.wrestlestat.com/tourney/projection

So PSU wins and OSU jumps ahead in both duals and tournament rankings. Hmmm, either you have BCS judges making selections or a tweak in your algorithm might be appropriate.

 

[pawrestlersintn]

I believe PSU won the dual on 2/3/18

Doesn't matter. The dual is figured into the results, but so is our backup 157. Remember, all of Wrestlestat is "on paper." On paper, without Jason, OSU is the better team. Good thing we wrestle to prove the paper wrong.

 

[jack66]

For next season, I plan on swapping out ACTUAL dual results if they actually occurred during the season...

What a unique idea.

 

[andegre]

Doesn't matter. The dual is figured into the results, but so is our backup 157. Remember, all of Wrestlestat is "on paper." On paper, without Jason, OSU is the better team. Good thing we wrestle to prove the paper wrong.

See reply above...

 

[pawrestlersintn]

I'm assuming you're referring to the 75-1 dual rankings record. That's what's generated when running Penn State vs all 76 other D1 teams through the prediction engine (dual comparison). That's how the dual rankings are generated.

For next season, I plan on swapping out ACTUAL dual results if they actually occurred during the season...

If you do that, then doesn't it become a reactive tool, rather than a predictive tool? Let's face it, very few people, were giving PSU a significant chance Saturday night. And, if we were to wrestle that dual again this Saturday, very few people would predict PSU to win again.

I don't know if you follow our KYPSW game at all, but of the 50 people that put predictions in for this match, 18 picked PSU, 15 picked OSU and 17 picked a tie. And, that's all blue-glasses-wearing PSU people picking.

Also, ELO is all about prediction, right? A guy with a higher ELO score is predicted to beat a guy with a lower ELO score. The dual results are a secondary prediction to the ELO rankings, so if you start putting actual dual results in place of predicted, I think it makes it look even stranger than it does now, with PSU behind OSU. ELO says X>Y, but Y beat X in one dual, so Y>X, even though your prediction right now would be for OSU to beat PSU. Taking my example to the absolute extreme, would you pick the 1980 U.S. men's hockey team over the Russians, ever?

 

[dwkeller]

The results make sense. On a neutral mat we are very close. Keep in mind this is a computer generated result. When on the mat, grit, determination and will are all in play.

Go PSU!

 

[Lyons212]

If you do that, then doesn't it become a reactive tool, rather than a predictive tool? Let's face it, very few people, were giving PSU a significant chance Saturday night. And, if we were to wrestle that dual again this Saturday, very few people would predict PSU to win again.

I don't know if you follow our KYPSW game at all, but of the 50 people that put predictions in for this match, 18 picked PSU, 15 picked OSU and 17 picked a tie. And, that's all blue-glasses-wearing PSU people picking.

Also, ELO is all about prediction, right? A guy with a higher ELO score is predicted to beat a guy with a lower ELO score. The dual results are a secondary prediction to the ELO rankings, so if you start putting actual dual results in place of predicted, I think it makes it look even stranger than it does now, with PSU behind OSU. ELO says X>Y, but Y beat X in one dual, so Y>X, even though your prediction right now would be for OSU to beat PSU. Taking my example to the absolute extreme, would you pick the 1980 U.S. men's hockey team over the Russians, ever?

Based upon what I saw Saturday night, I would pick Cassar over Moore if they wrestle again. To me that was not a shock, Moore has had some close matches this year.

Wrestlestat is a tool that is fun too use.

 

[andegre]

If you do that, then doesn't it become a reactive tool, rather than a predictive tool? Let's face it, very few people, were giving PSU a significant chance Saturday night. And, if we were to wrestle that dual again this Saturday, very few people would predict PSU to win again.

I don't know if you follow our KYPSW game at all, but of the 50 people that put predictions in for this match, 18 picked PSU, 15 picked OSU and 17 picked a tie. And, that's all blue-glasses-wearing PSU people picking.

Also, ELO is all about prediction, right? A guy with a higher ELO score is predicted to beat a guy with a lower ELO score. The dual results are a secondary prediction to the ELO rankings, so if you start putting actual dual results in place of predicted, I think it makes it look even stranger than it does now, with PSU behind OSU. ELO says X>Y, but Y beat X in one dual, so Y>X, even though your prediction right now would be for OSU to beat PSU. Taking my example to the absolute extreme, would you pick the 1980 U.S. men's hockey team over the Russians, ever?

Good points, now you're making me re-think my decision

 

[RoarLions1]

I'm assuming you're referring to the 75-1 dual rankings record. That's what's generated when running Penn State vs all 76 other D1 teams through the prediction engine (dual comparison). That's how the dual rankings are generated.

For next season, I plan on swapping out ACTUAL dual results if they actually occurred during the season...

If you do that, then doesn't it become a reactive tool, rather than a predictive tool? Let's face it, very few people, were giving PSU a significant chance Saturday night. And, if we were to wrestle that dual again this Saturday, very few people would predict PSU to win again.

I don't know if you follow our KYPSW game at all, but of the 50 people that put predictions in for this match, 18 picked PSU, 15 picked OSU and 17 picked a tie. And, that's all blue-glasses-wearing PSU people picking.

Also, ELO is all about prediction, right? A guy with a higher ELO score is predicted to beat a guy with a lower ELO score. The dual results are a secondary prediction to the ELO rankings, so if you start putting actual dual results in place of predicted, I think it makes it look even stranger than it does now, with PSU behind OSU. ELO says X>Y, but Y beat X in one dual, so Y>X, even though your prediction right now would be for OSU to beat PSU. Taking my example to the absolute extreme, would you pick the 1980 U.S. men's hockey team over the Russians, ever?

Good points, now you're making me re-think my decision

Question andegre...

The Compare Dual Lineup engine uses the "starters" to determine the dual winner, currently showing Bo Pipher as the 157 starter. When I change the starter to Jason Nolf, the match score flips to a PSU win, when adding the points at each weight class, yet the Wrestlestat Prediction is unchanged.

My question, and your thoughts...what if the prediction engine delivered a result based on changes to the starters (in this case, replace Bo with Jason)? Fans can then play with different line-ups to see how, in theory, they change the team results.

 

[andegre]

Question andegre...

The Compare Dual Lineup engine uses the "starters" to determine the dual winner, currently showing Bo Pipher as the 157 starter. When I change the starter to Jason Nolf, the match score flips to a PSU win, when adding the points at each weight class, yet the Wrestlestat Prediction is unchanged.

My question, and your thoughts...what if the prediction engine delivered a result based on changes to the starters (in this case, replace Bo with Jason)? Fans can then play with different line-ups to see how, in theory, they change the team results.

The concept is good, but the processing time required doesn't work.

The rankings process is VERY fast, EXCEPT for the dual rankings. (Actually, rankings for future seasons are processed on a daily basis, EXCLUDING the dual rankings. Future seasons dual rankings are going by the old-school: add up EloRank of the 10 starters, rank teams based on those values). Re-generating the rankings for all wrestlers actually only takes about 1 or 2 minutes. It's the dual rankings portion that takes about 20 minutes. So if you figure that's for all 77 teams, (where it remembers comparisons that have already been run), it would be a little more than 4 minutes.

Here's what I mean by "where it remembers comparisons that have already been run"...

It processes all teams, against all other teams, so:
1 vs 2
1 vs 3
1 vs 4
....
2 vs 1 (this already calculated, retrieve those results)
2 vs 3
2 vs 4
2 vs 5
.....
3 vs 1 (this already calculated, retrieve those results)
3 vs 2 (this already calculated, retrieve those results)
3 vs 4
3 vs 5
....

So, that would require a 4 minute process to execute every time a starter is changed. No user would wait that long, so I'd need to offload that to a background process. I'm currently in the process of building that "background processing" for all of my batch jobs right now, so that piece isn't done yet.

But, the downside to this approach, other than waiting for 4/5 minutes to see updated dual rankings, that causes more "resources" to be used on my cloud hosting platform which costs me more money. With the amount of times starters are changed on a daily basis on the site, that could get quite expensive...

 

[PSUcup1]

Based upon what I saw Saturday night, I would pick Cassar over Moore if they wrestle again. To me that was not a shock, Moore has had some close matches this year.

Wrestlestat is a tool that is fun too use.

I would as well. I picked Rasheed to beat Moore prior to the meet. He just doesn’t impress me all that much (relatively speaking for a #1 ranked kid of course)

 

[RoarLions1]

Thanks for the quick response!!

 

[pish69]

If you do that, then doesn't it become a reactive tool, rather than a predictive tool? Let's face it, very few people, were giving PSU a significant chance Saturday night. And, if we were to wrestle that dual again this Saturday, very few people would predict PSU to win again.

I don't know if you follow our KYPSW game at all, but of the 50 people that put predictions in for this match, 18 picked PSU, 15 picked OSU and 17 picked a tie. And, that's all blue-glasses-wearing PSU people picking.

Also, ELO is all about prediction, right? A guy with a higher ELO score is predicted to beat a guy with a lower ELO score. The dual results are a secondary prediction to the ELO rankings, so if you start putting actual dual results in place of predicted, I think it makes it look even stranger than it does now, with PSU behind OSU. ELO says X>Y, but Y beat X in one dual, so Y>X, even though your prediction right now would be for OSU to beat PSU. Taking my example to the absolute extreme, would you pick the 1980 U.S. men's hockey team over the Russians, ever?

We've had our disagreements over Elo in the past, but I think I agree 100% with you. The beauty of Wrestlesats is that there is no human element to it. Once you start messing with that, I think it losses it's original luster.

All that being said..I still don't understand some things (Ashnault #1 all year without wrestling) and I don't think I ever will but I'm fine with it. I love Wrestlestats and am grateful (Cael wearing off on me) for all Andegre and Obrats do to keep this site alive