WrestleStat #FantasyCollegeWrestling is LIVE!

andegre

Well-Known Member
May 18, 2004
4,146
789
1
Hello Wrestling Fans!

We are excited to announce that #FCW is officially open for the 2021-2022 season!!!

The major change for this year is providing the ability for leagues to be EITHER Head-to-Head scoring, or Cumulative Scoring. Last year because of COVID and the abbreviated schedule, it did not make sense to do the traditional head-2-head format that we did for the prior season, so we developed a cumulative scoring version. Well, now that that is done, we have the ability to provide both versions to our users.

Here's our Rokfin (subscription required, but it's our FREE channel) post giving all of the details, as well as rules, and signup: https://rokfin.com/article/6020/202122-Fantasy-College-Wrestling-FCW

Here's a link to the #FCW on WrestleStat where you can also find the same information: https://www.wrestlestat.com/fcw
 

GogglesPaizano

Well-Known Member
Feb 6, 2018
2,568
3,707
1
Andrege this is not intended to be yet another tired WS bash peice, but if I may could you run an individual audit on one or two let's say interesting but questionable matchup predictions to validate the ELO analysis is indeed running correctly?

I totally get the disparities we see with me wrestlers who haven't got in their 10-20 matches to get the math to work, but for veterans we should start to see more accurate and expected predictions.

I was tooling around and looked at PSU vs Missouri and one of several that jumped out to me is it National champion now Senior Nick Lee vs #10 Allan Hart which results in a tight Nick Lee 8-7 victory.

Against common opponents Nick is 13-0, whereas Hart it s 4-4. Nick doesn't have a single 1 point victory in the group. Hart's last matches was an 8-7 victory over Chad Red who Nick last beat by 2 points and a loss to Seabass, whom Nick beat 9-3 and earier in SV.

Without my having attempted to trying to duplicate the analysis this is one of those that on face value looks a little problematic one would think maybe a 6-3 predicted outcome a bit more realistic.

To be fair I really should have used the HWT example first as there you have #14 Zach Elam (1-5 common opponents) Beating #9 Greg Kervliet (4-1). Here Kerk's last match is a 12-2 MD over Wood whom Elam lost to twice?

In these cases the guys should have had close to a large enough body of work behind them to deliver a more expected result. Surely Kerk should be favored.

Do you have the ability to test your modeling on an individual match like this to validate it is operating correctly at scale?

PS - I think WS is awesome and my hat is off to you for the work you do. I hate to be bled dry by every service seeking a significant monthly subscription, but WS for say only a couple of bucks a month would likely keep most users and reward you appropriately.
 
  • Like
Reactions: andegre and Ski

GantryZ

Well-Known Member
Jul 12, 2013
302
390
1
PS - I think WS is awesome and my hat is off to you for the work you do. I hate to be bled dry by every service seeking a significant monthly subscription, but WS for say only a couple of bucks a month would likely keep most users and reward you appropriately.

Disagree, most users won't stay if it became a paid service. Wish it isn't the case but wrestling fans have proven time and time again that they will only pay for streaming video.

That said WS has a donation page and I encourage everyone to chip in, $15 gets rid of ads:


Had been awhile myself so I just threw a little their way.
 

andegre

Well-Known Member
May 18, 2004
4,146
789
1
Andrege this is not intended to be yet another tired WS bash peice, but if I may could you run an individual audit on one or two let's say interesting but questionable matchup predictions to validate the ELO analysis is indeed running correctly?

I totally get the disparities we see with me wrestlers who haven't got in their 10-20 matches to get the math to work, but for veterans we should start to see more accurate and expected predictions.

I was tooling around and looked at PSU vs Missouri and one of several that jumped out to me is it National champion now Senior Nick Lee vs #10 Allan Hart which results in a tight Nick Lee 8-7 victory.

Against common opponents Nick is 13-0, whereas Hart it s 4-4. Nick doesn't have a single 1 point victory in the group. Hart's last matches was an 8-7 victory over Chad Red who Nick last beat by 2 points and a loss to Seabass, whom Nick beat 9-3 and earier in SV.

Without my having attempted to trying to duplicate the analysis this is one of those that on face value looks a little problematic one would think maybe a 6-3 predicted outcome a bit more realistic.

To be fair I really should have used the HWT example first as there you have #14 Zach Elam (1-5 common opponents) Beating #9 Greg Kervliet (4-1). Here Kerk's last match is a 12-2 MD over Wood whom Elam lost to twice?

In these cases the guys should have had close to a large enough body of work behind them to deliver a more expected result. Surely Kerk should be favored.

Do you have the ability to test your modeling on an individual match like this to validate it is operating correctly at scale?

PS - I think WS is awesome and my hat is off to you for the work you do. I hate to be bled dry by every service seeking a significant monthly subscription, but WS for say only a couple of bucks a month would likely keep most users and reward you appropriately.
I apologize that I'm just now seeing this....

The prediction engine is mostly uses previous results instead of the wrestlers ELO rating. Let me explain:

For Lee, it calculates the average points scored, and average points against, for the wrestlers rating range. If Hart is 1300 (just guessing), then the PE (prediction engine) will come up with those averages for all previous matches against wrestlers rated from 1250-1349.

Then, do the same thing for Hart, average points scored and average points against, for opponents in Lee's range (just guessing, 1525), so that range would be 1450-1549.

Then for Lee, it would average HIS average points for with Hart's average points AGAINST.
Then flip it around for Hart, HIS average points for along with Lee's average points against.

That's how it comes up with the predicted score...so the ELO doesn't play too much into the prediction, other than determining the range to use for average point stuff.

Hope that helps.