WrestleStat Rankings - Week 2 2019
- Thread starter andegre
- Start date
Like I said before, using last year wrestlestat numbers to support this year'wrestlestat
Okay, fine ... if I were making a recommendation though, I would put more weight on good wins (6 AAs and 10 Top-20 for Pletcher vs 2 AAs and 5 Top-20 for DeSanto) as opposed to bonus points. I might also factor in placement at the previous year's NCAA tournament.
Maybe that's just me.
[QUOTE="jack66, post: 3801794, member: 502"]Like I said before, using last year wrestlestat numbers to support this year'wrestlestat
Okay, fine ... if I were making a recommendation though, I would put more weight on good wins (6 AAs and 10 Top-20 for Pletcher vs 2 AAs and 5 Top-20 for DeSanto) as opposed to bonus points. I might also factor in placement at the previous year's NCAA tournament.
Maybe that's just me.[/QUOTE]
Not sure what you're getting at. The way Wrestlestat is built, it is a career ranking, not a here and now ranking. So, it should build on last year's results. It builds on every result.
Okay, fine ... if I were making a recommendation though, I would put more weight on good wins (6 AAs and 10 Top-20 for Pletcher vs 2 AAs and 5 Top-20 for DeSanto) as opposed to bonus points. I might also factor in placement at the previous year's NCAA tournament.
Maybe that's just me.[/QUOTE]
Not sure what you're getting at. The way Wrestlestat is built, it is a career ranking, not a here and now ranking. So, it should build on last year's results. It builds on every result.
Kasparov coming across the board to take Fischer feet to back.
In short, Pletcher is an anomaly with the large number of close victories across a range of wrestlers’ rankings. It’s a small blind spot that is a small price to pay for the algorithm being better for the majority of predictions.
Last edited:
I will be looking forward to your published ranking system and your detailed explanation of every decision made by you. I am certain that the countless hours you invest in developing your superior system will be fully accepted by all viewers as the Gold standard end-all ranking system. I wait with breathless anticipation.
I believe Fischer is already on his back. Too Soon?
So the wrestlestat algorithm is now perfected, and all discussion should cease immediately, or face your wrath?
I think Mike gave me the best answer, Pletcher is perhaps an anomaly because of his style. But, I don't think my question was out of line. certainly a case can be made for Pletcher.
Sorry if I insulted you sensibilities.
Exactly. Pletcher IS an anomaly. Most guys that beat top 5 guys are majoring, at a minimum, guys in the 30-50 range, pinning or teching guys lower than that. DeSanto is also an anomaly in the opposite direction, putting up huge numbers against guys not that far below him because of pace, while not beating AAs (with the exception of an ailing Micic - which was a tech I believe, which I'm sure is impressive to the computer). Put the two together and voila! Well said, small price to pay to get an algorithm that works on 95% of field.
If you ever do decide to try to get through them all, interested to hear what you a/o Jack66 think about the 174 rankings.
People are coming back at you because many of us reallly appreciate all the work Andegre and Obrats have done with a wrestlestats. I was like you at one point and pointed out some what I would call headscratchers.
The more I listened to the explanations and started to understand the model, I was amazed and really appreciative of the work these guys put in.
IMO wrestlestats is the best wrestling technology ever. Before wrestlestats it was very hard to find match results, common opponents, etc. Look at all the information you can get at your fingertips so quickly and appreciate the fact that a couple guys with no monetary (other than donations) have given us fans. It’s absolutely amazing.
Then take the fact that Andegre has answered questions, explained the process, tweaked the algorithm (I believe he said his ultimate goal is to be #1 when you rank the rankers at the end of the year. All of his hard work to help grow the community and keep everyone informed.
It’s absolutely an incredible tool and I am so appreciative.
So yeah when you come and start picking a few ranked guys early in season and “knock” the system, people are going to be defensive.
Trust me. I was in your shoes at one point. Do yourself a favor. Go thru as many wrestlestats threads you can find and read all of Andegres explanations.
And then after that come and question whatever you want to in a civil manner without the snarky comments and you’ll get any answer you want.
For example I questioned Andegre about why some redshirt freshman were ranked high. He explained it perfectly.
I’m not trying to slap your wrists, just giving you the feedback on why it seems some are coming at you a little bit.
My wrath? Over wrestling rankings? Certainly you can't be serious? Clearly it is not perfect......but clearly the operator is working to make it better as fast as possible. I commend that. You had 4 not short posts discussing the same subject. None were positive. You said what you would do in your ranking system......clearly implying yours would be an improvement.
I simply said I will be looking forward to seeing it. I gathered you would have the time.....why not make that contribution? It seems a better use of time than focusing on a specific instance to criticize. As far as insulting my sensibilities.....I have none....so that ain't happening.
Since we are using chess references I always found that any discussion, criticism, bragging or whatever could always be solved with 3 words....."Set 'Em Up". I guess what I am saying concerning the criticisms/suggestions in your post is....."Set 'EM Up" Let's see what you got. I (WE) look forward to it.
Last edited:
Well Fischer certainly isn't tipping his King. We are a sick group
Dang, really felt like I was tracking this conversation then BAM, you throw in a reference to Electric Light Orchestra and I'm befuddled again. That's some strange magic, brother.
Got my tickets for ELO at PPG Paints next year. Wish I had NCAA tickets for PPG Paints next year
Got my tickets for ELO at PPG Paints next year. Wish I had NCAA tickets for PPG Paints next year
Haha. Jeff Lynne still rockin the stage at age 70. Maybe he’ll start out with “I’m Alive” just to reassure the audience.
I noticed that one too. but we're talking 1 and 1A, so it's not hard to understand.
Heck, I enjoy reading wrestlestat myself ... especially with dual comparisons. I simply asked a question on a ranking I thought was out of wack, then offered results I felt supported my opinion. You would think I insulted the Pope.
Pish69 gave me a reasoned response, I will heed his advice.
The ELO system was designed for chess and doesn't take into count margin of victory at all. Strength of opponent is the main factor when evaluating wins and losses. It's a methodical system that needs a good amount of data to show it's value. In the chess world, you need 25 games before your rating is not considered provisional.
Last edited:
I need some feedback from Andegre, but I'm pretty sure that he built a factor into his version of Elo, that accounts for margin of victory.
Till it mattered the most. DeSanto thumped Mueller 16-8 (MD) at the NCAA Championships. Not kryptonite any longer.
Hey fellas [and ladies].
I'll chime-in here to clarify some of the assumptions/conclusions that have been made so far....I don't believe there are two many unanswered, so I hope I hit all of them.
Margin of victory IS included (and WinType in a way). Here's what I mean....If Miklus beats Warner 3 - 2. Neither of them will move much since the margin of victory is only 1, and both of their ELO ratings are VERY close (1404.96 vs 1404). (By the way, you can see all ELO ratings by going to a wrestlers ranking history page.....), Now let's say that Warner majors Miklus 14-2, the margin of victory is now 12. And lastly, for FALLS, those just count as a 20-0 regardless of which period, etc)
For the people that have researched ELO, we are modifying the K value here (whereas in regular ELO, I believe K is a constant 32).
Here's a subsection of the code for the k value:
Another difference is the fact that we have placed a higher emphasis on a wrestlers first 15 matches. Since wrestlers don't have NEARLY as many matches as Chess players, we needed a way to move them up faster. My "algorithm guy" did the research (he's a retired Engineer (and statistics or mathematician) and found that the equilibrium number of matches to be 14 (maybe 15, don't remember). So those first 14/15 matches are weight much more heavily than the rest.
Win % does NOT factor in at all. It's just 2 wrestlers with their associated ELO ratings, and then determine new value based on W/L and margin of victory.
WrestleStat Rankings are a career ranking, so it uses the prior seasons final ranking as the seed/base as the following years ranking.
I scoured this thread and I believe I hit all of the questions and assumptions. If I missed any, please let me know and I'll get them answered.
I'll chime-in here to clarify some of the assumptions/conclusions that have been made so far....I don't believe there are two many unanswered, so I hope I hit all of them.
Margin of victory IS included (and WinType in a way). Here's what I mean....If Miklus beats Warner 3 - 2. Neither of them will move much since the margin of victory is only 1, and both of their ELO ratings are VERY close (1404.96 vs 1404). (By the way, you can see all ELO ratings by going to a wrestlers ranking history page.....), Now let's say that Warner majors Miklus 14-2, the margin of victory is now 12. And lastly, for FALLS, those just count as a 20-0 regardless of which period, etc)
For the people that have researched ELO, we are modifying the K value here (whereas in regular ELO, I believe K is a constant 32).
Here's a subsection of the code for the k value:
if (@favoredWon = 1) begin
set @k = 2.500000 + (@scoreDiff * @probability)
end else begin
set @k = (abs(@ratingDiff) / @kDivisor) + @scoreDiff
end
set @k = 2.500000 + (@scoreDiff * @probability)
end else begin
set @k = (abs(@ratingDiff) / @kDivisor) + @scoreDiff
end
Another difference is the fact that we have placed a higher emphasis on a wrestlers first 15 matches. Since wrestlers don't have NEARLY as many matches as Chess players, we needed a way to move them up faster. My "algorithm guy" did the research (he's a retired Engineer (and statistics or mathematician) and found that the equilibrium number of matches to be 14 (maybe 15, don't remember). So those first 14/15 matches are weight much more heavily than the rest.
Win % does NOT factor in at all. It's just 2 wrestlers with their associated ELO ratings, and then determine new value based on W/L and margin of victory.
WrestleStat Rankings are a career ranking, so it uses the prior seasons final ranking as the seed/base as the following years ranking.
I scoured this thread and I believe I hit all of the questions and assumptions. If I missed any, please let me know and I'll get them answered.
This is awesome news! I'm not losing my mind nearly as quickly as I thought I was.
I'm glad the guy that owns it can't remember the equilibrium number, either. I think I usually say 13, and if I say it like I know what I'm talking about, no one questions it.
Thanks for the info, and all you do.
Thanks from me as well, andegre. Only flaw I see in anything you do wrestling-wise, is your choice of college team fanship
.
Thank you.Thanks from me as well, andegre. Only flaw I see in anything you do wrestling-wise, is your choice of college team fanship
One of the things I was worried about when I started this was fans discounting the site just because I'm an Iowa fan. I was VERY adamant about not showing any favortism toward one team/state, and I think I've accomplished that. Honestly though, I do look for more "Iowa" things (like the blog/feed service has more Iowa stuff at this point) for some parts, but I try not to. At first, there was a little bit of that (fans discounting based on my school affiliation), but it's definitely down/doesn't exist anymore. "You people" may actually send more traffic to our site than Hawk fans at this point.
Ha...thanks for sharing. I have had so many awesome experiences with Iowa fans over my lifetime, that I would never be swayed by the couple of over-the-top halfwits that I meet or read on HR. And it never occurred to me to consider your work something less than what it is, just because of who you root for. That's bizarre, and way too judgmental...but I guess some think that way.
Example...Several weeks ago, at the Iowa/PSU football game, I attended a tailgate that had some ardent Iowa fans in attendance too. The conversation quickly drifted to wrestling. Once the guys learned more about me, they wanted a photo, so they stripped down (it was cold!!) to their first layer of clothing, showing Iowa Wrestling shirts. Not only do I have that photo, I've been in contact, and if our schedules permit, we'll possibly meet up at B1G's or NCAA's. Nice guys.
People are usually pretty cordial when face to face, unless drunk. It's on boards where people become douchebags, and there's a lot more than a couple. Then throw in the Sandusky bs and it's Katie bar the door. Oh, and the slush fund, steroids, and what have you.
As far as wrestlestat goes? I cannot believe people bitch and make childish comments regarding it. You don't like it don't open the damn post. Better yet, build your own system. Anyone that puts work into the promotion of wrestling should be commended.
Surprised to see margin of victory factored in. The ELO rating system was designed for zero-sum games so that could actually result in less reliable rankings. Did Professor Elo sign off on this modification?
Last edited:
That's not how this works. That's not how any of this works.Love Andegre's product, but who knew so much was involved? I'm happy for all of you who understand it. Just one question. What does "algorithm" have to do with wrestling and/or has my thinking gone astray?
Not too concerned about any rankings. Those things will take care of themselves for the most part. That said, for the sake of discussion, I believe if RBY wins the keystone classic he will show up in some top 20 rankings.
