WrestleStat Rankings - Week 1 (10)
- Thread starter andegre
- Start date
I just have to say this again. These are awesome and thanks for all the time you must put into these. I just had a couple questions just for my own curiosity.
How many matches does a guy have to wrestle before he gets more realistic predictions/rankings? Thinking mainly about true freshman like Howard and Bartlett who get pinned or teched every prediction just because they have no body of work so far.
Also, I noticed Carter Starocci was predicted to beat Kemerer and Massa but is ranked below them. Is that common where in the predictions A beats B, B beats C, C beats A or would/is Starocci be predicted to beat every wrestler even though he doesn't have the highest rank?
How many matches does a guy have to wrestle before he gets more realistic predictions/rankings? Thinking mainly about true freshman like Howard and Bartlett who get pinned or teched every prediction just because they have no body of work so far.
Also, I noticed Carter Starocci was predicted to beat Kemerer and Massa but is ranked below them. Is that common where in the predictions A beats B, B beats C, C beats A or would/is Starocci be predicted to beat every wrestler even though he doesn't have the highest rank?
So my "algorithm guy" has done the data analysis for this, and found out that the "equilibrium match count" is 14 matches. That means that after approximately 14 matches, a wrestler will finally reach most accurate ranking. But, that was also assuming a full, regular season of matches. With the way the season will play out this year and the greatly reduced match count across the board for all wrestlers, there will be wrestlers that may not even reach 14 matches. So this years rankings may be skewed compared to prior seasons. That will make it harder for the WrestleStat rankings to be more accurate come tournament time, but I'm still confident that it will "perform" very well for the season.
Regarding Starocci, the prediction engine only uses the wrestlers ranking/rating as a baseline for the prediction process. When comparing wrestlers, it mainly focuses on previous individual match results to predict how many points a wrestler will score, vs how many points will be scored against them...all of that based on other matches for that wrestler vs opponents in the same rating range.
Hope that clears it up....I can surf through this message board to give some examples of this...
Here's a post I found that goes into this detail...
bwi.forums.rivals.com
WrestleStat Week 3 Dual Comparisons
Some real good matchups Friday including #5 Michigan vs #7 Lehigh, #10 Missouri vs #8 VaTech, and conference battle #11 Nebraska vs #19 Wisconsin Here's the full run-down: Friday: 11/16 #5 Michigan vs #7 Lehigh https://www.wrestlestat.com/compare/dual/40/michigan/37/lehigh 11/16 #10...
bwi.forums.rivals.com
ELO came up in The Queen's Gambit. For some reason, Mrspa asked about it. So, in my "knowledgeable voice," I said, "Well, you see, honey, in the chess world, they use an algorithm to give each player a rating..." and waxed poetic for a couple of minutes. I am not sure whether the look she gave me was that she thinks I'm a God or a nerd.
I bet we can guess how she meant it..... Same as my wife, but hey, they married us! Joke's on them!
Similar threads
