Facts, Not Fiction

 
Page 5 of 5 FirstFirst ... 345
Results 41 to 49 of 49
  1. Collapse Details
     
    #41
    Jack Lint in forth place broke his PB by 142 points. What is special about Lint's performance is that his result his higher than his sum of PBs by 3 points (for comparison Erm was 223 points behind his sum of PBs and Harrison Williams by 638).

    Lint is a senior from Michiagn and given the fact that he doesn't have the talent of Erm and company it is likely to be his last decathlon.

    So beofore he retire to whatever is decathletes are doing after college - Well done Jack Lint.

    And no - I don't know him personally nor do I have special love to Michigan (the university or the state)
    Reply With Quote
     

  2. Collapse Details
     
    #42
    Senior Member
    Join Date
    Oct 2007
    Location
    Mired in Tar
    Posts
    2,113
    For what it's worth, for the fifteen athletes who finished the event, the average difference between the actual scores and my initial projected scores is 153 points, or 2.02%.
    Reply With Quote
     

  3. Collapse Details
     
    #43
    Senior Member
    Join Date
    Oct 2005
    Location
    Madison, WI
    Posts
    23,361
    Does Lint qualify for USATF?
    Reply With Quote
     

  4. Collapse Details
     
    #44
    Senior Member
    Join Date
    Nov 2015
    Location
    England, UK
    Posts
    1,430
    Erm goes from 16th to 7th on the Estonian all-time list. Not bad for a country of 1.3m. By contrast, the UK has 50 times the population and only Daley Thompson and Dean Macey have beaten 8352.
    Reply With Quote
     

  5. Collapse Details
     
    #45
    Quote Originally Posted by Davidokun View Post
    For what it's worth, for the fifteen athletes who finished the event, the average difference between the actual scores and my initial projected scores is 153 points, or 2.02%.
    Thanks for making the effort to track this.

    I think it was discussed last year (after Talence?) about needing some sort of factor for likely improvement (lower for older elite, higher for younger) to moderate your averaging approach.

    I think this result tends to support that but I'm still unsure how best to do it. Do you or olorin(or anybody) have any thoughts?
    Reply With Quote
     

  6. Collapse Details
     
    #46
    With regards to upside potential - to me it’s Owens for sure - then Erm - then Williams. While Williams is older, his hurdling and throwing are inconsistent enough to feel there is real upside. Then again - he was a phenom as a youth athlete so maybe he’s maxed out.

    Owens is a beast and very young and new to Dec.
    Reply With Quote
     

  7. Collapse Details
     
    #47
    Senior Member
    Join Date
    Oct 2005
    Location
    North Texas
    Posts
    3,370
    Quote Originally Posted by 26mi235 View Post
    Does Lint qualify for USATF?
    USATF auto qualifier is 7800. Could possibly qualify as one of the top 16.
    Reply With Quote
     

  8. Collapse Details
     
    #48
    Quote Originally Posted by El Toro View Post

    I think this result tends to support that but I'm still unsure how best to do it. Do you or olorin(or anybody) have any thoughts?
    I have many thoughts on the topic as I was trying to come up with a model to predict decathlon (and Heptathlon) scores few years ago (I think that I have few posts on the topic). Not to brag or anything, but I do analyze data in my non-track life, so I would like to think that I know a thing or two about it. Predicting decathlon scores is one of the hardest things that I tried to do in my life, if not the hardest. After two years of trying I can summarize my efforts in one word - failure. My complicated model was doing better, but only very slightly compared with other simple models that I tried.

    The main problem is that the distribution of scores is not normal (fat tales) and non-symmetric (more large negative movement than positive one) which create a situation that the mean is lower than the median. So basically you can do two things:
    1. Try to predict the score in each event separately but then you will have too optimistic estimation for the overall results.
    2. Try to predict the overall score but than you will have too pessimistic estimations for each of the individual events.

    I didn't try David's average system, but it is related to the second type of models. We "punish" each decathletes by choosing the average even though they are more likely to peak for an important competition. It may be (as you suggested) more pronounce for collegiate athletes as they compete more, in less important competitions, so their average (and thus predicted score) is bias downward.

    Anyway, I will stop now as I can go on forever.
    Last edited by olorin; 06-08-2019 at 01:23 AM.
    Reply With Quote
     

  9. Collapse Details
     
    #49
    Quote Originally Posted by olorin View Post
    After two years of trying I can summarize my efforts in one word - failure. My complicated model was doing better, but only very slightly compared with other simple models that I tried.
    Been there, done that, more than once.

    A wise man once said, "Compared to simple models, complicated models are often wrong but in a different way and with much greater precision!"
    Reply With Quote
     

Posting Permissions

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