Results 1 to 10 of 10

Thread: The Stats Corner: A 5-Year Look at Sochi Medal Favorites, Ladies

  1. #1
    On the Ice
    Join Date
    Nov 2013
    Posts
    33

    The Stats Corner: A 5-Year Look at Sochi Medal Favorites, Ladies

    Hi everyone! I'm quite new to posting on this forum, but have been reading for a season or two. This is my very first thread, so please don't be too harsh on me.

    Figure skating statistics is something I like doing in my spare time (or when I'm playing USFSA's Fantasy Figure Skating). I'm sure some of you have heard of the Yale statistics professor who analyzed the (flawed) ISU scoring system during the 2006 Torino Olympics. Obviously, I'm nowhere near being like him, but I think I can still make some pretty cool stuff using Excel. Hopefully you'll think so too!

    There has been a lot of discussion about predictions for medals in Sochi, so I made this graph, which I hope some people may find helpful going into Olympics mode. I simply compiled all of the ISU competitions in the past 5 years for the major medal contenders and graphed them by date on the horizontal (x-axis). (Didn't do any stats tests this time.)

    If you like this kind of data visualization and find it useful, please let me know and I will be sure to post more!

    *** Click here to see the graph. ***

  2. #2
    Custom Title chapis's Avatar
    Join Date
    Oct 2009
    Location
    México
    Posts
    575
    All ladies are more less in the same place but Mao is in everywhere , so I predict she can take the gold medal or 9th place .

  3. #3
    ♥ My Mirai's Golden Pond ♥ Sasha'sSpins's Avatar
    Join Date
    Apr 2009
    Posts
    729
    Welcome Miss October! I think a 5 year look at the medal favorites for Sochi is quite interesting! I'm checking it out right now.

  4. #4
    Custom Title
    Join Date
    Feb 2010
    Posts
    648
    Quote Originally Posted by Miss October View Post
    Hi everyone! I'm quite new to posting on this forum, but have been reading for a season or two. This is my very first thread, so please don't be too harsh on me.

    Figure skating statistics is something I like doing in my spare time (or when I'm playing USFSA's Fantasy Figure Skating). I'm sure some of you have heard of the Yale statistics professor who analyzed the (flawed) ISU scoring system during the 2006 Torino Olympics. Obviously, I'm nowhere near being like him, but I think I can still make some pretty cool stuff using Excel. Hopefully you'll think so too!

    There has been a lot of discussion about predictions for medals in Sochi, so I made this graph, which I hope some people may find helpful going into Olympics mode. I simply compiled all of the ISU competitions in the past 5 years for the major medal contenders and graphed them by date on the horizontal (x-axis). (Didn't do any stats tests this time.)

    If you like this kind of data visualization and find it useful, please let me know and I will be sure to post more!

    *** Click here to see the graph. ***
    Hi, Miss October. Very cool and interesting stuff.

    The scatterplot showing only raw data is, of course, somewhat difficult to visually interpret.

    A question: do you have the capability/programming to generate probability distribution curves (sometimes known as histograms, or bell curves) by data-fitting?

    If you can generate histograms for each individual skater using their five-year performance histories, and overlay them on one graph, that would be a rough, first-order approximation of relative ranking and implied odds.

    If we could generate standard deviations (sigmas), we would get an additional sense of how significant a performance differential there were between performances as well as between skaters.

    Those are just some of my random thoughts. I'm very interested in hearing what sort of things were you planning on doing with the data.

  5. #5
    Custom Title
    Join Date
    Feb 2010
    Posts
    648
    Quote Originally Posted by chapis View Post
    All ladies are more less in the same place but Mao is in everywhere , so I predict she can take the gold medal or 9th place .
    Simply eyeballing the data, I would say that Yuna represents a huge statistical outlier.

  6. #6
    she takes the audience on her journey of emotions Layfan's Avatar
    Join Date
    Nov 2009
    Location
    Mexico City
    Posts
    3,630
    Cool stuff, Miss October! thanks.

    Quote Originally Posted by Robeye View Post
    Simply eyeballing the data, I would say that Yuna represents a huge statistical outlier.
    She's in a galaxy far, far away However, she has no dot this season.

    Julia looks pretty good here.

  7. #7
    On the Ice
    Join Date
    Nov 2013
    Posts
    33
    Quote Originally Posted by Robeye View Post
    A question: do you have the capability/programming to generate probability distribution curves (sometimes known as histograms, or bell curves) by data-fitting?
    Wow, thanks for the question! I'm just working with Excel and Access right now. I don't have any of the fancier stats programs that require programming. I suppose I could fit data to a line, but I don't know what I would plot against the score...time? I could also just do a bell curve with the mean and SD of these data for each skater, but I don't know how good that would be for prediction, as figure skaters improve throughout the season and also develop and fade as their careers progress. I made a model for Fantasy Figure Skating during last year's season using rank order between competitions, but it didn't work so well...

    Quote Originally Posted by Robeye View Post
    Those are just some of my random thoughts. I'm very interested in hearing what sort of things were you planning on doing with the data.
    Oh, nothing really. It was just fun and something to share with other figure skating fans on this forum. I wasn't sure what I could contribute to the discussions, so I decided to make something. I think looking at data would be relatively objective and helps me stay informed about how these skaters are doing going into the Olympics!

  8. #8
    Six Point Zero Krislite's Avatar
    Join Date
    Sep 2010
    Location
    Yunaverse
    Posts
    1,519
    What I like about your graph is that the increasing/inflation trend is very evident once you take out Yuna's scores from 2009-2010. I imagine the trend would be even more pronounced if you graph only PCS over the five-year period (subtract TES from total).

  9. #9
    Custom Title
    Join Date
    Feb 2010
    Posts
    648
    Quote Originally Posted by Miss October View Post
    Wow, thanks for the question! I'm just working with Excel and Access right now. I don't have any of the fancier stats programs that require programming. I suppose I could fit data to a line, but I don't know what I would plot against the score...time? I could also just do a bell curve with the mean and SD of these data for each skater, but I don't know how good that would be for prediction, as figure skaters improve throughout the season and also develop and fade as their careers progress. I made a model for Fantasy Figure Skating during last year's season using rank order between competitions, but it didn't work so well...
    I agree with you that figure skating is a tough nut for the predictions game. Some of the reasons I think so:

    1) The sport is structurally ill-suited for predictions. I wrote a post a while ago on the topic: http://www.goldenskate.com/forum/sho...-Skating/page2

    2) Ladies skaters in particular have senior careers that are relatively brief, and therefore do not yield a lot of data points compared to athletes in many other sports. This is exacerbated by the fact that there are relatively few competitive events, and it doesn't take much (an injury, puberty, semi-retirement, technical reworking, etc.) to introduce thorny irregularities or issues into the data set.

    3) Again focusing on ladies, the prime of their careers generally tend to coincide with their teen years. As you allude, this is a time when women go through particularly dramatic changes in both physical and mental characteristics. As we have seen time and again, within a period of six months, the capabilities of a skater can be transformed beyond recognition (both for good and for ill).

    4) Some people misunderstand the nature of statistical inference and "prediction". It is not like a supernaturalist foretelling that something will definitely happen. To the extent that statisticians make predictions, they are usually of the "Monte Carlo" variety: if an event were simulated using a given quantitative model 1,000 times, then skater X would win 823 times, skater Y 144 times, skater Z...etc.

    That being said, I think that statistical analysis is much more accurate and relevant when:

    -The key players are longtime veterans, who have achieved physical and performance maturity for an extended period.

    -The performance differentials are consistently significant (i.e. standard deviations are high).

    This is why I speak with relatively more conviction about the prospects for Yuna and, to a lesser extent, Mao.

  10. #10
    Custom Title
    Join Date
    Feb 2010
    Posts
    648
    Quote Originally Posted by Layfan View Post
    Cool stuff, Miss October! thanks.

    She's in a galaxy far, far away However, she has no dot this season.
    It looks like Miss October is not counting nationals or Senior B. If those were included, and Yuna's Korean Nationals score of almost 228 was discounted some amount for the nationals bonus, she would still bestride the field like a colossus, as they say (220+, easy). Jes' sayin'.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •