WrestleStat Week 3 Dual Comparisons

andegre · Nov 15, 2018

rmg78 · Nov 15, 2018

So Ohio St. Will have back ups in at 184 and 197 bc Martin and Moore are at U23's. Curious how long of the season Moore might miss as he is wrestling 97 kg over there and then has to come down 16 more lb's to make 197.

RoarLions1 · Nov 15, 2018

andegre said:
11/18 #16 Arizona State vs #2 Ohio State https://www.wrestlestat.com/compare/dual/4/arizona-state/55/ohio-state

Read on the tOSU board that some thought Anthony Valencia is redshirting. Thought he had surgery, and was not yet cleared...but that he will wrestle this year. Near positive he already had a RS year. We're 4 weeks from wrestling the Sun Devils...both Valencia's should be going...or did I miss something?

pish69 · Nov 15, 2018

RoarLions1 said:
Read on the tOSU board that some thought Anthony Valencia is redshirting. Thought he had surgery, and was not yet cleared...but that he will wrestle this year. Near positive he already had a RS year. We're 4 weeks from wrestling the Sun Devils...both Valencia's should be going...or did I miss something?

The way I understand it is Anthony took his Olympic redshirt when he was true freshman and he still has his regular redshirt available. Sounds like he wants to go this year but coaches want him to redshirt this year..guess we'll see who wins, the coach or the wrestler

tomsc · Nov 15, 2018

pish69 said:
The way I understand it is Anthony took his Olympic redshirt when he was true freshman and he still has his regular redshirt available. Sounds like he wants to go this year but coaches want him to redshirt this year..guess we'll see who wins, the coach or the wrestler

I remember Anthony did really well at the US Open one year. Did he qualify for for an Olympic RS by coming in 3rd at the US Open (or WTT's)? Unusual for a true FR to qualify for an olympic RS!

tomsc · Nov 15, 2018

Wrestlestat--I really appreciate your putting out these dual comparisons each week. Very helpful to see potential match-ups.

I have a question. I know that these are all mathematical projections, but I'm trying to figure out what part of the equation gave rise to this prediction: Minn's #7 McKee is picked over #5 OSU's Brock 11-7, even though Brock is 4-0 on HTH and has a 23-7 record to McKee's 11-23 against common opponents. All of those would seem to make Brock a clear favorite. Is there a component of the equation that would slant it to McKee? Just curious.

GogglesPaizano · Nov 15, 2018

rmg78 said:
So Ohio St. Will have back ups in at 184 and 197 bc Martin and Moore are at U23's. Curious how long of the season Moore might miss as he is wrestling 97 kg over there and then has to come down 16 more lb's to make 197.

Good question but it doesn't mean Moore was wrestling 'at' 97kg. That said, a 15 lb cut is about right for tOSU with a guy that size. I love that strategy every March, they peak in the quarters and then....

That is one of several consistent things about Cael - you don't have to look like Hercurock to wrestle for him as he understands the concept of wrestling strength vs weight lifting strength. Likewise with few exceptions he doesn't manage the whole team down a weight (Berge and maybe VJ out of necessity). In the end it opens the door to the kids having fun, don't get burnt out, operate near peak efficiency .... in what by all accounts is a pretty grueling sport.

They are all great coaches but Cael is more of a visionary, strategist and leader as opposed to a micro manager as we see each at several of the other elite programs.

I am shocked that more blue chip elite recruits don't see this one more clearly. I know I know money, 9.9, yada yada...

horraybeer · Nov 15, 2018

tomsc said:
I remember Anthony did really well at the US Open one year. Did he qualify for for an Olympic RS by coming in 3rd at the US Open (or WTT's)? Unusual for a true FR to qualify for an olympic RS!

He finished 4th at trials while still in HS, so that qualified him for an OlympicRS.

CholleyVandine · Nov 15, 2018

RoarLions1 said:
Read on the tOSU board that some thought Anthony Valencia is redshirting. Thought he had surgery, and was not yet cleared...but that he will wrestle this year. Near positive he already had a RS year. We're 4 weeks from wrestling the Sun Devils...both Valencia's should be going...or did I miss something?

He Tweeted recently that he's ready to go and that no trainer or coach can tell him how his body feels. Just a day or two ago, he Tweeted that he's itching to start competing again.

pish69 · Nov 15, 2018

tomsc said:
I remember Anthony did really well at the US Open one year. Did he qualify for for an Olympic RS by coming in 3rd at the US Open (or WTT's)? Unusual for a true FR to qualify for an olympic RS!

Almost positive that was the case

andegre · Nov 16, 2018

tomsc said:
Wrestlestat--I really appreciate your putting out these dual comparisons each week. Very helpful to see potential match-ups.

I have a question. I know that these are all mathematical projections, but I'm trying to figure out what part of the equation gave rise to this prediction: Minn's #7 McKee is picked over #5 OSU's Brock 11-7, even though Brock is 4-0 on HTH and has a 23-7 record to McKee's 11-23 against common opponents. All of those would seem to make Brock a clear favorite. Is there a component of the equation that would slant it to McKee? Just curious.

The prediction engine uses a wrestlers past history of matches vs ALL opponents in the same rating range (as the wrestler you want the prediction against) to get an approximation of points [for] scored, and points [against]. Then do the exact same thing for the opponent (but the opposite) and it comes up with the projected points from both wrestlers. Then it just checks who scored more. The rankings are only utilized to determine what rating range each of the wrestlers live in.

Rating ranges are like this:

<1050
1051-1150
1151-1250
...........

RoarLions1 · Nov 16, 2018

CholleyVandine said:
He Tweeted recently that he's ready to go and that no trainer or coach can tell him how his body feels. Just a day or two ago, he Tweeted that he's itching to start competing again.

Saw that, and frankly, it was the reason for my post. It appears he wants to go...after surgery, just could not find any info beyond his tweet. No biggie, we'll know soon enough. Just hope he's healed and ready to go (even though he wrestles for another team).

andegre · Nov 16, 2018

RoarLions1 said:
Saw that, and frankly, it was the reason for my post. It appears he wants to go...just could not find any info beyond his tweet. No biggie, we'll know soon enough. Just hope he's healed and ready to go (even though he wrestles for another team).

I was also under the impression that it was a couple things:

1) Injured
2) Had "real" redshirt still available since he already took his Oly shirt as mentioned above
3) 165 was getting too difficult

And because of #3, he was going to fill in to 174 this season, which would then allow Zahid to go 184 for his last season.

Apparently the cut wasn't as bad as I thought it was because I only see him going at 165 if he goes this season...

tomsc · Nov 16, 2018

andegre said:
The prediction engine uses a wrestlers past history of matches vs ALL opponents in the same rating range (as the wrestler you want the prediction against) to get an approximation of points [for] scored, and points [against]. Then do the exact same thing for the opponent (but the opposite) and it comes up with the projected points from both wrestlers. Then it just checks who scored more. The rankings are only utilized to determine what rating range each of the wrestlers live in.

Rating ranges are like this:

<1050
1051-1150
1151-1250
...........

Thanks for getting back on this, andegre. So, despite losing all 4 HTH's McKee may have gotten more points because of his margins of victory (or loss) against wrestlers of similar ratings? Is that it? A follow-up question: Brock is ranked higher by the computer rating--so what factors into the dual match-up (McKee over Brock) that isn't in their rankings (Brock higher than McKee)?

andegre · Nov 16, 2018

You bet.

Regarding the HTH question, I am literally in the process of working on incorporating previous head-to-head results, as well as the probabiliity of each of the different win types for a wrestler into the prediction engine. For example, Spencer Lee gets a boat-load of TF and Falls, but very few MD's. Therefore, if the prediction engine predicts a match result around, for example, 5 to 10 point advantage, then the likelihood that he TF's is far greater than a MD. So I'm trying to figure out how to programmitaclly sway that prediction toward a TF.

Regarding the McKee vs Brock prediction, I'll try to answer with a data example...
Brock's ELO is ~1458, while McKee's is ~1410. Therefore, so using the range above, that would put McKee in the 1351-1450 range, while Brock is in the 1451-1550 range.

Now, let's say Brock has 8 matches versus opponents in the 1451-1550 range. Of those 8 matches, he averages 4, while his opponents average 2 points.
For McKee, let's say he has 5 matches vs opponents in the 1451-1550 range, of those, he averages 8 points while his opponents average 7 against him.

So, the math looks like this:
Brock points: 8*4 (his average points) + 5*7 (opponents average points surrendered) = 32 + 35 = 67 / 13 = 5.15
McKee points: 5*8 (his average points) + 8*2 (opponents average points surrendered) = 40 + 16 = 56 / 13 = 4.31

So, WITH THIS EXAMPLE DATA, the predicted outcome would be Brock by DEC 5-4.

Obviously that's not what the actual prediction says, but it shows how I derive that end result.

Hope that helps!

tomsc · Nov 16, 2018

andegre said:
You bet.

Regarding the HTH question, I am literally in the process of working on incorporating previous head-to-head results, as well as the probabiliity of each of the different win types for a wrestler into the prediction engine. For example, Spencer Lee gets a boat-load of TF and Falls, but very few MD's. Therefore, if the prediction engine predicts a match result around, for example, 5 to 10 point advantage, then the likelihood that he TF's is far greater than a MD. So I'm trying to figure out how to programmitaclly sway that prediction toward a TF.

Regarding the McKee vs Brock prediction, I'll try to answer with a data example...
Brock's ELO is ~1458, while McKee's is ~1410. Therefore, so using the range above, that would put McKee in the 1351-1450 range, while Brock is in the 1451-1550 range.

Now, let's say Brock has 8 matches versus opponents in the 1451-1550 range. Of those 8 matches, he averages 4, while his opponents average 2 points.
For McKee, let's say he has 5 matches vs opponents in the 1451-1550 range, of those, he averages 8 points while his opponents average 7 against him.

So, the math looks like this:
Brock points: 8*4 (his average points) + 5*7 (opponents average points surrendered) = 32 + 35 = 67 / 13 = 5.15
McKee points: 5*8 (his average points) + 8*2 (opponents average points surrendered) = 40 + 16 = 56 / 13 = 4.31

So, WITH THIS EXAMPLE DATA, the predicted outcome would be Brock by DEC 5-4.

Obviously that's not what the actual prediction says, but it shows how I derive that end result.

Hope that helps!

Thanks for the example. Not to extend this forever, but am I correct in understanding that margin of victory/loss over people in your ratings group counts for more than the actual number of guys/ranking of who you have beaten (for the duals comparison)? If so, I would think that guys like Heil and Pletcher would do less well in duals comparison than in their actual ratings/ rankings.

andegre · Nov 16, 2018

tomsc said:
Thanks for the example. Not to extend this forever, but am I correct in understanding that margin of victory/loss over people in your ratings group counts for more than the actual number of guys/ranking of who you have beaten (for the duals comparison)? If so, I would think that guys like Heil and Pletcher would do less well in duals comparison than in their actual ratings/ rankings.

I'm not following you here...

ad1986 · Nov 16, 2018

Michigan's dual vs Lehigh will now be 5pm on Sunday (11/18)
postponed due to weather-related travel issues
4 or 5 starters will be out for Lehigh

86PSUPaul · Nov 16, 2018

ad1986 said:
Michigan's dual vs Lehigh will now be 5pm on Sunday (11/18)
postponed due to weather-related travel issues
4 or 5 starters will be out for Lehigh

I heard about the postponement. Any idea why 4-5 are out?

ad1986 · Nov 16, 2018

Schram, Scott Parker are hurt. Karam was carried off the mat last Sunday. Not sure how hurt he is. They may pull Humphreys shirt, but he will be very good.

86PSUPaul · Nov 16, 2018

Bummer, hate to see injuries!

tomsc · Nov 16, 2018

andegre said:
I'm not following you here...

Sorry. It seems for dual comparison, that score differential (in matches wrestled) counts for more than HTH results or comparative results against common opponents (McKee vs Brock as the example). If I understood your explanation correctly, that is why McKee is favored over Brock. If that's so, then people who win most of their matches, but win by close margins (like Heil and Pletcher) would get "punished" in dual comparisons. I don't know that that's true, but am guessing.

andegre · Nov 17, 2018

tomsc said:
Sorry. It seems for dual comparison, that score differential (in matches wrestled) counts for more than HTH results or comparative results against common opponents (McKee vs Brock as the example). If I understood your explanation correctly, that is why McKee is favored over Brock. If that's so, then people who win most of their matches, but win by close margins (like Heil and Pletcher) would get "punished" in dual comparisons. I don't know that that's true, but am guessing.

Oh, yes, that's exactly right. Point scorers are favored with the current system, while the ones that eek them out will be more susceptible to upset predictions.

tomsc · Nov 17, 2018

andegre said:
Oh, yes, that's exactly right. Point scorers are favored with the current system, while the ones that eek them out will be more susceptible to upset predictions.

Thanks, andregre. Now I have a better handle on interpreting the projections. Appreciate your hanging in there with me as I try to understand.

PSU James · Nov 18, 2018

I simply love wrestlestat!

GogglesPaizano · Nov 18, 2018

Hey Andregre - I just noticed Mark Hall is ranked #1 over Zahid at #2. I understand the ELO analysis but it's still a slight headacher given their match history.

I assume that despite Marky losing his most recent match (not close) our Big 10 schedule is much stronger so Mark has more wins vs quality opponents and Zahid's margin of history vs common opponents must be less on average?

andegre · Nov 19, 2018

GogglesPaizano said:
Hey Andregre - I just noticed Mark Hall is ranked #1 over Zahid at #2. I understand the ELO analysis but it's still a slight headacher given their match history.

I assume that despite Marky losing his most recent match (not close) our Big 10 schedule is much stronger so Mark has more wins vs quality opponents and Zahid's margin of history vs common opponents must be less on average?

I think it's both, with a slight favor to margin-of-victory. Without going through and doing the math, it's mostly because Hall bonuses more...the more-and-more that happens, he'll continue to hold a lead since he has beaten him once. Now if Valencia beats him again (on 12/14), I believe they are close enough that Hall would slide back to 2, but it will be RAZOR thin margin at that point.

I wish I had a version of the code where I could just insert a single match and get the new rating (NOT ranking), but the way it's written right now it re-processes for all matches in a season (for a single wrestler). That could be a future enhancement, but it would be low on the totem pole. (I just logged the enhancement...)

WrestleStat Week 3 Dual Comparisons

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Well-Known Member

Similar threads

Football Penn State remains in No. 4 in latest College Football Rankings (Week 14)