Wrestlestat Algo is broke

[Pinnacle]

I checked the dual comps for PSU vs Rutgers and then, since I'm already there, I check PSU vs. Nebby. That algorithm(sp?) has Kasak beating Ridge Lovett. I like Kasak as much as the next guy, but what? They must have gotten into Connabon's stash. 

[1032004]

They even put the Penn State magic into the algorithm

[Idaho]

6 minutes ago, Pinnacle said:

I checked the dual comps for PSU vs Rutgers and then, since I'm already there, I check PSU vs. Nebby. That algorithm(sp?) has Kasak beating Ridge Lovett. I like Kasak as much as the next guy, but what? They must have gotten into Connabon's stash. 

Penn State wins the first 5 matches by the same score - 6-5. 

[Pinnacle]

39 minutes ago, Idaho said:

Penn State wins the first 5 matches by the same score - 6-5. 

Now that is just plain mailing it in.

[Truzzcat]

they have Soldano over Bernie which im guessing is entirely because of the weight of a pin in the algo and soldano pinned rogotzke who pinned bernie despite soldano being 4-5 against common opponents and bernie being 10-3.

[1032004]

I don’t get how their dual comparison predicts upsets in general, if it’s all based on an algorithm shouldn’t the higher rated wrestler always win?

[ionel]

20 minutes ago, 1032004 said:

I don’t get how their dual comparison predicts upsets in general, if it’s all based on an algorithm shouldn’t the higher rated wrestler always win?

It appears to be based on common who've they've beaten or lost to and how or score differential.  Not perfect but prob better than just current rank.

[Show_Me]

On 2/12/2024 at 9:06 AM, 1032004 said:

I don’t get how their dual comparison predicts upsets in general, if it’s all based on an algorithm shouldn’t the higher rated wrestler always win?

I have been wondering this same thing for several years now.

[ILLINIWrestlingBlog]

On 2/12/2024 at 10:06 AM, 1032004 said:

I don’t get how their dual comparison predicts upsets in general, if it’s all based on an algorithm shouldn’t the higher rated wrestler always win?

 

I've been thinking about that as well. The best I've come up with is a measure I use in predicting duals:

 

1. Assign a win percentage to all the wrestlers in a dual. 

2. Then, add up the win percentages for each team and divide by 100. Then, you end up with something, for example, like 6.3 to 3.7.

3. Assign the win to a wrestler from the 3.7 team who is closest to a tossup even though he is an underdog in his particular match. 

 

I think it makes sense statistically (as long as the percentages are close), and because there always seems to be at least one slight upset per dual, or you have a starter sitting out because of injury or illness. 

Is that what Wrestlestat.com does? I don't know.