PDA

View Full Version : Final 2018 WrestleStat Rankings



andegre
03-20-2018, 12:15 PM
Here are the final rankings. Remember, these rankings may not necessarily directly reflect the placement of each wrestler at the NCAA tournament; it's still an accumulation of their career rankings. For example, Spencer Lee is still #2 at 125 while Tomasello is #1.

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

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

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

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

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

This offseason the rankings process is going to get modified to help accommodate some of the issues that were found this season. The main problem identified is when there are wrestlers that have a wide range of "best wins" vs "bad losses".

WrestleStat is going to receive some major upgrades this offseason, including a re-design of the entire site layout. Stay tuned to see all of the changes.

Lastly, any and all suggestions, site problems, constructive criticism is welcomed.

RadioBison
03-21-2018, 03:08 AM
This offseason the rankings process is going to get modified to help accommodate some of the issues that were found this season. The main problem identified is when there are wrestlers that have a wide range of "best wins" vs "bad losses".

WrestleStat is going to receive some major upgrades this offseason, including a re-design of the entire site layout. Stay tuned to see all of the changes.

Lastly, any and all suggestions, site problems, constructive criticism is welcomed.

For the 2nd straight year, I compared 8 different sets of rankings against the results at the NCAA tournament. Your predictions were correct in 456 out of 640 matches (71.3%), which trailed the other systems by a bit, but I think you can make it up with a few tweaks in the right direction. Intermat was the winner this time around with 74.7%, although that was out of only 586 matches since they only rank down to 20. I wonder if you go back and re-run your algorithms on previous seasons when you make a change so that you can get a feel for how well they predict going forward. I started doing that on my personal system, and I think it helped, but I haven't proven that to myself yet. Anyway, good luck with your updates, and as always, thanks for the great site.

Loose Lips
03-22-2018, 01:26 AM
Thanks for a very fun, interesting site, Good job. Predictions just show what SHOULD happen. You can't predict school stress, sleep, girlfriend, weight cut, diet, sickness, family problems or the many factors that a wrestler faces during the season. I am impressed!

andegre
03-22-2018, 01:10 PM
For the 2nd straight year, I compared 8 different sets of rankings against the results at the NCAA tournament. Your predictions were correct in 456 out of 640 matches (71.3%), which trailed the other systems by a bit, but I think you can make it up with a few tweaks in the right direction. Intermat was the winner this time around with 74.7%, although that was out of only 586 matches since they only rank down to 20. I wonder if you go back and re-run your algorithms on previous seasons when you make a change so that you can get a feel for how well they predict going forward. I started doing that on my personal system, and I think it helped, but I haven't proven that to myself yet. Anyway, good luck with your updates, and as always, thanks for the great site.

Absolutely. Someone on the mat forums provided a spreadsheet for scoring the different rankings services. WrestleStat placed last (but was 1st at 125, 2nd at 149 and 157, and 3rd at one more weight. The remaining 6 weights DRAMMATICALLY pulled down the overall average.

Now that I have THAT information (with other ranking services ranked) and the point system used to rank them, I'm going to work EXTENSIVELY this offseason to tweak/test/tweak/test etc until I can get my ranking system to the top of that list.

I'm confident that I'll be able to get my algorithm at, or VERY close to the top of that list. This is something that I should have done last season...but didn't.