WrestleStat - Final PreSeason Rankings updated

[andegre]

The preseason rankings have been updated on WrestleStat to prepare for the start of the Fantasy Pick'Em contest tomorrow.

These should contain all transfers, weight changes, redshirts, etc.

In the past few days I've also made numerous improvements to the speed of each of these pages. If you notice anything taking longer than you think it should, don't hesitate to contact me.

Here are the rankings:

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

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

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

 

[sunshinedynomite]

I have a feeling If

They finish with a 4,1,1,1,2,1,1

We would be ok with that.

 

[NoVa Lion]

I have a feeling If

They finish with a 4,1,1,1,2,1,1

We would be ok with that.

Nope, have to have 8 scorers.

 

[jack66]

I have a feeling If

They finish with a 4,1,1,1,2,1,1

We would be ok with that.

But don't we need 8 scoring wrestlers?

 

[Chickenman Testa]

I have a feeling If

They finish with a 4,1,1,1,2,1,1

We would be ok with that.

FIUR Cael!!!!!

 

[Used2BEerie]

Nope, have to have 8 scorers.

But don't we need 8 scoring wrestlers?

Anybody else getting a sudden sense of Deja Vu?

 

[jrd23psu]

I can't take your rankings seriously when returning champs at a weight aren't even ranked #1. 141 and 174 are ridiculous.

 

[RoarLions1]

I can't take your rankings seriously when returning champs at a weight aren't even ranked #1. 141 and 174 are ridiculous.

Best out there for what it is...an algorithm that, as much as possible, removes the human component. After reading about ELO, and seeing (from a distance) how andegre keeps improving the model, I think it's awesome...again, for what it is.

 

[jrd23psu]

Best out there for what it is...an algorithm that, as much as possible, removes the human component. After reading about ELO, and seeing (from a distance) how andegre keeps improving the model, I think it's awesome...again, for what it is.

An algorithm is never going to tell the whole story, though. The two weights I mentioned are a great example. Does anything think Yanni and Zahid aren't the #1 guys at their weight? When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use.

Sure, it's a neat exercise, but it's not an accurate portrayal of reality. There are definitely problems with all human polls and problems with hybrid polls as well. However, I think that 141 and 174 are obvious weights for who is #1 and even #2 at 174.

 

[CholleyVandine]

An algorithm is never going to tell the whole story, though. The two weights I mentioned are a great example. Does anything think Yanni and Zahid aren't the #1 guys at their weight? When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use.

Sure, it's a neat exercise, but it's not an accurate portrayal of reality. There are definitely problems with all human polls and problems with hybrid polls as well. However, I think that 141 and 174 are obvious weights for who is #1 and even #2 at 174.

What does "reality" mean in this context? The WrestleStat rankings are not an accurate portrayal of conventional wisdom. That's not the same thing as reality.

 

[pawrestlersintn]

This will only affect predictions, right? And, not rankings? You know I love to antagonize the people that are getting all bent out of shape and don't get that the rankings are just math. Er, did I say antagonize? I meant educate.

I can't take your rankings seriously when returning champs at a weight aren't even ranked #1. 141 and 174 are ridiculous.

An algorithm is never going to tell the whole story, though. The two weights I mentioned are a great example. Does anything think Yanni and Zahid aren't the #1 guys at their weight? When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use.

Sure, it's a neat exercise, but it's not an accurate portrayal of reality. There are definitely problems with all human polls and problems with hybrid polls as well. However, I think that 141 and 174 are obvious weights for who is #1 and even #2 at 174.

Am I prescient, or what?

Exactly what practical use are you using ANY rankings for? Betting? Hoping? Predicting? And, that's not a rhetorical question. I'd really like to know.

Why do you want yet another ranking service to place all of the wrestlers within one or two spots of each other? I mean, we already have Flo, Intermat, WIN, The Open Mat, AWN, TrackWrestling, Wrestling by Pirate, all which will closely agree with each other, but you want one more to confirm that Yanni is the man to beat at 141?

Wrestlestat is a career achievement ranking, not a who's on top this week ranking. It is for entertainment purposes only, just like any other ranking.

 

[tikk10]

An algorithm is never going to tell the whole story, though. The two weights I mentioned are a great example. Does anything think Yanni and Zahid aren't the #1 guys at their weight? When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use.

Sure, it's a neat exercise, but it's not an accurate portrayal of reality. There are definitely problems with all human polls and problems with hybrid polls as well. However, I think that 141 and 174 are obvious weights for who is #1 and even #2 at 174.

It doesn't pretend to tell "the whole story." Every ranking need not adhere to that (probably unreachable) standard b/c different rankings tell different stories. The idea that there need be one story else don't bother is your presumption. This same argument comes up every year and I don't get it. It also comes up in other sports; the mere mention of analytics in hockey conjures up laments of how analytics can't measure "grit," and so on. And the correct response is "Yeah, so what? It's measuring this other thing that people also find interesting and useful."

 

[HOCKEYGOD11]

How long until 25 and 33 see a PSU wrestler ranked? Shouldn't be to long I'm guessing.. This line up.. IF all stay healthy and goes as we expect, should be the most balanced from top to bottom.. first time in a long time that there is no glaring hole at any one weight... Is it time to start the season yet?!?!?!

 

[aalion]

there is also a difference between statistical accomplishment and prediction. If you mix the two they can become self fulfilling, which is wrong. Easy to see in football polls - guys will rank Wisconsin high and say "they have an easy path to an undefeated season because BIG West is soft". This is a prediction mixed with a ranking. A ranking should be based on accomplishment, at least over time. If Suriano and Spencer Lee are both 18-0, and Suriano beat every other wrestler in the Top 20 except himself and Lee, while Lee wrestled noone ranked in the Top 50, in my opinion a poll should have Suriano ranked higher because he accomplished more. Even if most still think Spencer would win head-to-head. You can argue both ways to do it, but if you don't separate ranking from prediction, the process becomes self-fulfilling. "I had Wisconsin ranked #1 because of their soft schedule, and because they are undefeated I am keeping them there". What you really meant was "I predicted Wisconsin would go undefeated". That thinking rewards soft scheduling, not accomplishment.

I believe this is a statistical ranking, gets refined over time. Not necessarily a prediction - more a measure of accomplishment. I think.

 

[NoVa Lion]

An algorithm is never going to tell the whole story, though. The two weights I mentioned are a great example. Does anything think Yanni and Zahid aren't the #1 guys at their weight? When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use.

Sure, it's a neat exercise, but it's not an accurate portrayal of reality. There are definitely problems with all human polls and problems with hybrid polls as well. However, I think that 141 and 174 are obvious weights for who is #1 and even #2 at 174.

If you can come up with a way to "tell the whole story" to the satisfaction of the majority I will tip my hat to you. However, it just is NOT possible and never will be as opinions on how rankings should be done are like a******s, everyone has one.

 

[pawrestlersintn]

How long until 25 and 33 see a PSU wrestler ranked? Shouldn't be to long I'm guessing.. This line up.. IF all stay healthy and goes as we expect, should be the most balanced from top to bottom.. first time in a long time that there is no glaring hole at any one weight... Is it time to start the season yet?!?!?!

Assuming things go the way we expect, we should see those guys bound up the rankings in week 4 or 5.

 

[jrd23psu]

If you can come up with a way to "tell the whole story" to the satisfaction of the majority I will tip my hat to you. However, it just is NOT possible and never will be as opinions on how rankings should be done are like a******s, everyone has one.

I don't have one. I think a computer method with a human component is probably my favorite method. I know people hated the BCS, but I think it would have been fine had we just had a 4 team playoff using that model. There are going to be flaws in an all human model, all computer model, or a hybrid, but I'd prefer a mix of both.

 

[pish69]

I find it interesting how there are some guys that were redshirts last year that are ranked high? Is that a mistake?

For example..some guys I never heard of..Paguildo at 157 and Orndoff at HWT. I believe O'Connor is 4th at his weight (I know he's good but seems strange to see him this high off redshirt year). Woodley 3 and Warner 4 at 197..

 

[RoarLions1]

I don't have one. I think a computer method with a human component is probably my favorite method. I know people hated the BCS, but I think it would have been fine had we just had a 4 team playoff using that model. There are going to be flaws in an all human model, all computer model, or a hybrid, but I'd prefer a mix of both.

You're missing the point. The algorithm (ELO) has been modified a lot to accommodate the uniqueness of college wrestling (as compared to chess). Those modifications are human-thought-developed by andegre, based on feedback from all the major fan bases. That IS a human component, though after that, the math takes over. No bias, no prejudice.

It's not better or worse than what the other ranking agencies deliver, just different. You're busting on a model because you "prefer a mix of both"...so I'll assume the WIN, Flo, IM, TOM, and Track rankings aren't to your liking either.

As an example, using one of the weights you mentioned, 141;
Last year's results --

Eierman
vs Diakomihalis W 9-6 L 6-4
vs Headlee W 12-6 W 15-5 (Maj)
vs Lee W 12-4 (Maj)
vs Alber W 12-6
vs Heil FALL
vs Perry FALL
vs Meredith L 3-2

Diakomihalis
vs Eierman L9-6 W 6-4
vs Headlee W 7-6
vs Lee W 12-7
vs Alber W 8-2
vs Heil 6-5
vs Perry 11-2 (Maj)
vs Meredith W 4-2 (SV) W 7-4

I can see this close enough to call their match-up a toss-up. Eierman beat Headlee, Lee, Heil and Perry worse than Diakomihalis did, though Yianni did beat Meredith twice. Pretty close. Wouldn't bet the house on either guy.

My thoughts on rankings, in general, is that they're fun for discussion and debate, which has value to fans as they talk throughout the season. None are great predictors of actual results though. The one feature I love best about the Wrestlestat site is the historical results for each wrestler. Use them when we pick our fantasy team at NCAA's from amongst the unseeded guys.

 

[RoarLions1]

I find it interesting how there are some guys that were redshirts last year that are ranked high? Is that a mistake?

For example..some guys I never heard of..Paguildo at 157 and Orndoff at HWT. I believe O'Connor is 4th at his weight (I know he's good but seems strange to see him this high off redshirt year). Woodley 3 and Warner 4 at 197..

Those are some great questions pish...the kind that andegre loves to get and answer. Andegre??

 

[jrd23psu]

It's not better or worse than what the other ranking agencies deliver, just different. You're busting on a model because you "prefer a mix of both"...so I'll assume the WIN, Flo, IM, TOM, and Track rankings aren't to your liking either.

With the exception of Track (they don't seem to be current) I like all of the others better. I think they are all more accurate, although WIN does have Hayes and Micah at the wrong weights. But other than that I checked the rankings for all of those and thought they were good.

 

[andegre]

Those are some great questions pish...the kind that andegre loves to get and answer. Andegre??

Sure do

One of the tweaks to make it work better for wrestling since there are fewer games/matches in a career is to have an equilibrium factor. This value has been set to attempt at getting the "super freshman" as close to their "correct" ranking within their first X matches. The math has been analyzed [by my algorithm guy] and determined that equilibrium match count is 14. So the first 14 matches of a wrestlers career are weighted MUCH more heavily than matches 15+. It works great for a Spencer Lee, Nick Lee, Yianni, etc, but it does NOT work well for the middle-of-the-road wrestlers, that have just mediocre wins [and very few losses] in those first 14 matches. That's why we are seeing Woodley #4 @ 197, Warner #5 (they've actually split between the two of them also), etc.

So right now, if we were to just tweak the equilibrium to bring Woodley and Warner down, then it would ALSO pull down Yianni, then even more people would question the rankings.

We're trying to identify a happy-medium for scenarios like these. Like others have said in this thread, no human, nor computer [ranking] system is going to be perfect, but I'm trying like HELL to get WrestleStat's as close as possible.

 

[nerfstate]

Sure do

One of the tweaks to make it work better for wrestling since there are fewer games/matches in a career is to have an equilibrium factor. This value has been set to attempt at getting the "super freshman" as close to their "correct" ranking within their first X matches. The math has been analyzed [by my algorithm guy] and determined that equilibrium match count is 14. So the first 14 matches of a wrestlers career are weighted MUCH more heavily than matches 15+. It works great for a Spencer Lee, Nick Lee, Yianni, etc, but it does NOT work well for the middle-of-the-road wrestlers, that have just mediocre wins [and very few losses] in those first 14 matches. That's why we are seeing Woodley #4 @ 197, Warner #5 (they've actually split between the two of them also), etc.

So right now, if we were to just tweak the equilibrium to bring Woodley and Warner down, then it would ALSO pull down Yianni, then even more people would question the rankings.

We're trying to identify a happy-medium for scenarios like these. Like others have said in this thread, no human, nor computer [ranking] system is going to be perfect, but I'm trying like HELL to get WrestleStat's as close as possible.

Your hard work is very much appreciated!

 

[Dogwelder]

... When the algorithm gets it this wrong I think it's evidence that we shouldn't be relying on it for practical use. ...

In other words, do NOT use wrestlestat ranking algorithm in the International Space Station life support system!

 

[Dogwelder]

What are the red and green superscript numbers in the wrestler rankings? (e.g., for 125, 3^2(green) Bresser, Ronnie ...)

 

[andegre]

What are the red and green superscript numbers in the wrestler rankings? (e.g., for 125, 3^2(green) Bresser, Ronnie ...)

That indicates how many spots the wrestler moved up or down since the previous weeks [or seasons] ranking. Red means they moved down (got worse), and green means they moved up.

For preseason rankings, if there's a value, it's because there was a senior above them, or a wrestler above them switched weights/quit (since they obviously haven't had any matches since the last set of rankings were released).

 

[James P. Whitters III]

I love wrestlestat! It is the best thing to come along for a keyboard warrior to look at upcoming duals. I love that you are constantly improving it.

 

[Dogwelder]

That indicates how many spots the wrestler moved up or down since the previous weeks [or seasons] ranking. Red means they moved down (got worse), and green means they moved up.

For preseason rankings, if there's a value, it's because there was a senior above them, or a wrestler above them switched weights/quit (since they obviously haven't had any matches since the last set of rankings were released).

Thanks andegre. I wonder whether the conventional up and down arrows might be better than green/red, given that I would have understood arrows, and given that the NIH says as much as 8 percent of men with Northern European ancestry are red/green colorblind. What do you have against Iceland?

 

[Chickenman Testa]

Thanks andegre. I wonder whether the conventional up and down arrows might be better than green/red, given that I would have understood arrows, and given that the NIH says as much as 8 percent of men with Northern European ancestry are red/green colorblind. What do you have against Iceland?