|
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Casey Talbott added to this discussion on March 28, 2014
|
|
Discussion Topic: NCAA statistical scoring analysis
Mark Niemann added to this discussion on March 28, 2014
In the comments section is written something I was wondering... I'd love to see how much different - or if at all different (thank you, Dan Strope) - the quarters or semis, or even the conso-semis would be.
But it is nice to see. Especially in pie graph form.
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Ryan Mitchell added to this discussion on March 29, 2014
I tried to post a comment there, but it wouldn't take (maybe that is a sign that I'm way off base). Here is my rather lengthy thought.
Chris, I appreciate the effort you put into this and anything that gives wrestling coverage and fosters discussion about how to improve our sport is a great thing.
That said, I had a big problem with your conclusions because they were not really based on a thorough analysis of the available data.
For example, you state in the above response that you were surprised how often that the riding time effected the outcome (of matches). In truth, the riding time rarely effects the outcome of a championship match. There is almost NO causal relationship between a wrestler having 1:00+ of riding time and winning a match. There is a positive correlation, but I think it is more a function of the winning wrestler having been superior from the neutral position. One theory of why such a positive correlation exists between RT & wins is that with the 1st period being 3 minutes, he who gets a takedown in the 1st period, which ALWAYS starts from neutral will more often than not, gain a significant RT advantage going into the final 2 periods.
You also showed 2 pie charts, the 1st of which showed that the wrestler who scored the first takedown won 85% of matches. Following that chart, you said “Let’s move onto a much more important statistic.” That chart showed who had First Score (of any kind) and the wrestler with first score won 70% of the time. Now, correct me if I’m wrong, but isn’t 85% significantly higher than 70%? Even if you score first, but you don’t score by TAKEDOWN first, your chances of winning drop by 18%. This also supports the idea that improving from neutral both on the attack and on defense significantly increases your chances of winning
I’m a big fan of Dumb & Dumber and I thought you might have totally redeemed yourself, but even that missed the mark. If 30% of wrestlers who do not score first go on to win the match, that is 3 champions every year who overcome giving up first points (though NOT first takedown), which makes scoring first obviously desirable, but not “Imperative.” More like 300,000 in 1,000,000.
All of this is mostly moot because as others have stated, trying to draw meaningful conclusions from such a small data set and from such a small pool of wrestlers that generally took varied paths to the finals is statistically unreasonable, call it the Small Sample Size problem.
Further, when trying to make a study like this helpful to the wrestling community, including only NCAA championship matches eliminates almost everyone from the applicable group. The farther away from the NCAA finals you get, the greater the talent gap and the less meaningful these conclusions become. The subtitle itself “What does it really take to win a D1 NCAA Wrestling Championship " would be more accurately stated as “How Some People Have Won an NCAA Championship Match."
I feel that bad conclusions presented as truth to a group not used to having advanced data at its fingertips (i.e. the wrestling community) provides a false path that can dangerously lead people farther away from where they want to go. Again, I appreciate your effort and your intentions and I found the data interesting, but I don’t know that there is much here that is helpful in coaching future champions.
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Mark Niemann added to this discussion on March 29, 2014
Ryan Mitchell stated: "...Further, when trying to make a study like this helpful to the wrestling community, including only NCAA championship matches eliminates almost everyone from the applicable group. The farther away from the NCAA finals you get, the greater the talent gap and the less meaningful these conclusions become."
I think these two sentences are the issue. You could use the eventual champ's previous four matches, along with the finals match, but more than likely, the data will become skewed as you work backwards. That is why I'd like to see data from all of the final AA matches. That would be four matches per weight class, and, for the most part, would be of daily equal competitors.
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Luke Moore added to this discussion on March 30, 2014
But isn't the point to always want to emulate the best? If the national finalists are hitting singles and stand ups, the the guys that are not in the national finals hitting granbys and under arm spins, which technique should be thought of as the best?
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Mark Niemann added to this discussion on March 30, 2014
Quote from Luke Moore's post:
|
"But isn't the point to always want to emulate the best? If the national finalists are hitting singles and stand ups, the the guys that are not in the national finals hitting granbys and under arm spins, which technique should be thought of as the best?"
|
"No." - 4x sectional qualifier
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Hank Kornblut added to this discussion on March 30, 2014
It's an interesting read but the finals alone are too small a sample size to draw conclusions. When Zeke Jones wanted to prepare his team for the World Championships (I think 2013 but am not sure), he looked at every single match that was wrestled the previous year by every wrestler (not just Americans). He analyzed who scored, when and how and at what time in the period. He was able to use this data to offer a lot of useful info to his wrestlers.
I suspect the finals are fairly similar to the event as a whole but that some differences would show up if all the matches were similarly categorized.
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Hank Kornblut added to this discussion on March 30, 2014
It's an interesting read but the finals alone are too small a sample size to draw conclusions. When Zeke Jones wanted to prepare his team for the World Championships (I think 2013 but am not sure), he looked at every single match that was wrestled the previous year by every wrestler (not just Americans). He analyzed who scored, when and how and at what time in the period. He was able to use this data to offer a lot of useful info to his wrestlers.
I suspect the finals are fairly similar to the event as a whole but that some differences would show up if all the matches were similarly categorized.
|
|
|
Discussion Topic: NCAA statistical scoring analysis
Mike Hojnacki added to this discussion on March 30, 2014
|
|
Discussion Topic: NCAA statistical scoring analysis
Rex Holman added to this discussion on March 31, 2014
IMO the goal is to win. While emulation is a template, the reason someone wins is that they are stronger at a position at a critical point in the match and yes data gets skewed by mismatches. In this is a very interesting subject because the same kind of data was gleaned from the World Championships back in the 70s, obviously different in styles and while data does not lie, the interpretation can.
|
|
|
|
|
|
|