Wrestling Stats Site
- Thread starter matman197
- Start date
I've got to say - andrege, right? - the algorithm that you are using just seems to me to be really right on once you get to 20 matches of data or so. Yeah, guys with a limited schedule or a guy like Gantt who wrestled a weak schedule and then had his season cut short, are out of place. But if you remove those guys you get a ranking that really passes the eyeball test to me, in regards to BOTH season accomplishments and where the guys are right now. I like it better than SHP's rankings (sorry, SHP) and it compares favorably to Intermat - who I consider to have the best "human" rankings (although wrestling report would be right up there if they did their's most often). Not to bash FLO or CP, who frequents here, but I think CP's philosophy is more of a body of work ranker, and might not be a great predictor.
Would it be possible to add the points to the rankings? It would be interesting to see that at a weight class like 133 where I doubt that there is much separation between 1 and 10, compared to some other classes where there might be a more distinct pecking order.
Tremendous job.
Would it be possible to add the points to the rankings? It would be interesting to see that at a weight class like 133 where I doubt that there is much separation between 1 and 10, compared to some other classes where there might be a more distinct pecking order.
Tremendous job.
Thank you. And yes, I agree on all parts (about the # of matches, etc). That has been, BY FAR, the most difficult part of the algorithm to tweak. At the beginning of the season (say less than 5 matches), the rankings were horrible, but now they look pretty good. I tried to incorporate weighting in how much a wrestlers ranking would move up or down based on the number of matches, but there always ended up being these odd outliers that made it look bad altogether. There are still a few that don't look quite right (Clark #2 at 133), but they are awfully close.
Thanks for the kind words. Main reason I wanted to do this (instead of just averaging ALL of the rankings websites) was because I wanted to also see where redshirts, and everyone outside the top 20 would fall in the rankings. Gives fans perspective on where there up-and-comers are going to be next season (www.wrestlestats.com/Rankings/NextYearAll).
I never thought about adding the "points" to the individual rankings pages. I'm going to do some more work tonight (so the UPdate Picks for TourneyPool isn't so slow), I'll try and sneak that in....actually, I may just do it right now, give me about 15 minutes.
If people don't use this site to assist them in filling out their brackets, they don't know a good thing when they see it.
Wow - thanks. Definitely gives you an added perspective to the rankings, almost like a stratisfication, and increases my opinion of your algorithm.
133 is by far the closest in the point spread between 2 and 10 (57 points). The next closest is 157 at over 125 points and that includes 3 guys with limited data points in there, so in truth the spread would be even higher at 157 I suspect. That is my take on 133 exactly as the top 10 guys are super close and Dardanes, being undefeated, is the only one who has set himself apart.
Great stuff. But I won't be sharing it until all my crew's NCAA pools are turned in. When I win, I'll give you some credit.
Similar threads
