When Did US Fencing Begin Using Ratings?

Discussion in 'Fencing Discussion' started by sdubinsky, Mar 2, 2015.

  1. K O'N

    K O'N Podium

    Joined:
    Aug 14, 2006
    Messages:
    3,593
    Likes Received:
    458
    "Be more accurate" is not a very well-defined metric. Be more accurate in what way? What's the metric?

    How unkind! And I thought we were buddies. Now I'm sad.

    Since you're referring to my comments, I'm moved to wonder why you think I don't understand how an Elo system works. It's not very complicated.

    In a statistical sense you can predict the number of heads from coin flips, though. It's a simple binomial calculation. You could similarly predict the results of a tournament based on Elo scores, though I'm not sure why you'd want to. The point of a rating system is to seed pools evenly, not to predict who'll win the event.

    That's what an Elo system does, but for the most part I don't think it's what a fencing rating system is required to do. For a rating system, the measure of effectiveness should be how even the pools are.

    I'm all for unusual tournament formats, but for serious events we're likely for the foreseeable future to use pools to DEs, to mirror what international events do. So evenness of pools is important. Predicting individual bout results is less important. The two are related, but if you're establishing a metric, even pools should be high in your mind.

    Since the poster pointed to an article that described exactly such a system, I think the poster can be assumed to have been aware of such a solution.

    I agree. That's what the article I pointed to laid out. If you talk to chess people I think you'll find they don't like this, both because it's new and because you in essence now have two ratings, your two year high and your current rating, and they think it's complicated to keep track of compared to a single number. But yes, a high water mark for Div II and similar events seems to be the best solution to tanking bouts on purpose.

    Also, the same problem exists currently. A strong C can avoid getting a B before SN in order to fence in Div II. And strong C's do so. So in terms of comparison to the current system, I think that point is a wash.

    I'm sorry if you're feeling attacked about this, as your tone seems to indicate. I don't intend to do so. I'm just a guy on the internet anyway, I like math and I'm interested in how rules affect gameplay. I think large-scale changes like this should be really carefully examined for bad effects. If prying open any possible bad effects of an Elo system feels like an attack on you personally I'm sorry, that wasn't intended.

    An Elo system with a two year floor, or something similar, seems to me to be a perfectly workable system. I think I said that before, and I still think that. I'm not at all sure, though, that it's any better than the system we have for seeding pools. To show it is you'd have to establish a metric that measured something like the variance of the pool results in DEs or something similar.

    I think you'll find that an Elo system does only slightly better than the current system, if it does measurably better at all. So suppose I'm right about that; pools to DEs is not very sensitive to initial seeding anyway. If going to an Elo system would not improve seeding much or at all, then you have to ask if the trouble and added complication and the required computation is worth the non-seeding benefits (ie, measurement of progress, etc). That's a social question, I guess you'd take a survey or something.
     
  2. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    A couple seasons ago the FIE's 379th ranked senior women's epee fencer (out of 571 with rankings) was someone who never fenced an FIE epee event. Or much epee at all.

    She's a Panamanian saber fencer.

    Someone at her federation kept entering her name into FIE senior women's events ... apparently without paying much attention to which weapon.
     
  3. Purple Fencer

    Purple Fencer Podium

    Joined:
    Jan 19, 2001
    Messages:
    15,876
    Likes Received:
    510
    That's the kind of guy who deserves a single-light touch against him every time....douche-canoe.
     
  4. oiuyt

    oiuyt Podium

    Joined:
    Apr 26, 2000
    Messages:
    10,240
    Likes Received:
    974
    o.83.1.d

    B
     
  5. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    The question was ...

    A good reply would include:

    1) "more accurate" than what?

    2) How do you propose to measure "the seeding of competitions"

    That doesn't answer the "more accurate than what?" question.

    Also it mentions measuring the rating system's accuracy. That's not measuring "the seeding of competitions ... [to] be more accurate"


    To be fair, many posters making comments on this thread advocating for various "Elo-type" rating systems likewise have no idea how they would be implemented or function but that doesn't stop them from offering their "expertise" as to the supposed benefits of such a system. :)
     
  6. Inquartata

    Inquartata Podium

    Joined:
    Jul 12, 2001
    Messages:
    36,189
    Likes Received:
    1,183
    Every time I hear how a system like ELO more accurately predicts whether one fencer will beat another I think of the 'Start Trek' episode 'A Taste of Armageddon'. Why fence the bout? Just record the results predicted by the system and all the fencers can stay home and see how they did according to the computer. Then the winners can report to a virtual podium...and the losers to disintegration chambers. ;)
     
  7. Bonehead

    Bonehead Podium

    Joined:
    Aug 27, 2002
    Messages:
    2,563
    Likes Received:
    137
    Actually that is fairly interesting. Classifications are based on the strength of the competition in a way that's seemingly agnostic to gender/age.

    Assuming it's not super controversial to say that Males tend to be stronger than females, and Open events tend to be stronger than junior/cadet/veteran, do the ratings held reflect this?

    i.e. I would expect fewer percentage of all the women fencers to hold an A rating than the men, and same for juniors.
     
  8. Blackwood

    Blackwood Made the Cut

    Joined:
    Mar 23, 2011
    Messages:
    356
    Likes Received:
    24
    In the current USA Fencing membership list there are 6,276 women and 13,620 men listed as competitive fencers.

    Of the women, 1.37% have an A in Sabre, 1.80% have an A in Epee, and 1.37% have an A in Foil.
    And of the men, 0.82% have an A in Sabre, 3.28% have an A in Epee, and 1.46% have an A in Foil.

    I didn't account for fencers with an A in more than one weapon, but I think an earlier post said that there are very few of them. So it seems the percentage of men with an A in at least one weapon is a little higher than the percentage of women.
     
  9. Zebra

    Zebra DE Bracket

    Joined:
    Dec 25, 2013
    Messages:
    771
    Likes Received:
    181
    Is it reasonable to say that that perceived weakness is operative only at the lowest ratings? I've been two touches away from an E in a vet unrated while getting my clock cleaned in opens; and I hear fencers grouse about how someone else has a weak E. But it's a lot harder to get a C or higher rating in one of those restricted groups. If a vet tournament or a women's gets up to a C1, there have to be some pretty good fencers in it.
     
  10. fdad

    fdad Podium

    Joined:
    Jul 9, 2008
    Messages:
    4,129
    Likes Received:
    214
    I am not sure what any of this proves, but it might be interesting to see - though the data might be hard to extract from FRED - what percentage of participants receiving rating upgrades from mixed events were male vs. female. Similarly, what percentages of fencers receiving ratings upgrades from open events were Y14, U17, U20, V40, V50, V60, V70, etc.
     
  11. Peach

    Peach Podium

    Joined:
    Feb 10, 2001
    Messages:
    163
    Likes Received:
    766
    One of the unofficial functions of classifications (or is it unofficial?) is to "strengthen" local competitions and attract fencers, as well as to boost the perceived strength of a club. I know of a number of clubs that hold events apparently scheduled (on a weeknight, with only club members signed up?) so that they can generate classifications. It's darn hard to generate high ratings that way, but you can probably get at least a handful of D's out of it.
     
  12. Peach

    Peach Podium

    Joined:
    Feb 10, 2001
    Messages:
    163
    Likes Received:
    766
    Given that there are fewer women competitors than men, and that no one has investigated how gender affects individual tendency to affect participation in tournaments of different strengths, it's probably more complicated than that.
     
  13. Zebra

    Zebra DE Bracket

    Joined:
    Dec 25, 2013
    Messages:
    771
    Likes Received:
    181
    You sure you're teaching English and not calculus?? [grin]
     
  14. K O'N

    K O'N Podium

    Joined:
    Aug 14, 2006
    Messages:
    3,593
    Likes Received:
    458
    Ok? I'm not sure what that means, exactly. Ratings are based on where you finish in a competition. Some competitions are restricted, to women or age groups for example.

    Kind of. It's complicated by the existence of NACs, which are large restricted events with ratings available. If you're serious enough to train hard enough to win a rating at an open event, you probably go to NACs at least every once in a while.

    Yes, but as you said that's more due to the numbers of each. If you had five hundred women in the same weapon in an area, they'd very soon generate A's at the same rate men do. Women/Vets/Cadets do get ratings at national (restricted) events, but I think a WE or Vet or Cadet who's strong enough to get a rating at a NAC will have just as good a chance of getting it locally in a mixed event first.

    But if there were equal numbers and lots of restricted local events, yes, they (we) would no doubt get ratings in those more often than in mixed. There just never are.

    So I'm not sure if I'm arguing with you or agreeing with you.
     
  15. Bonehead

    Bonehead Podium

    Joined:
    Aug 27, 2002
    Messages:
    2,563
    Likes Received:
    137
    That's true.

    I guess you could frame the question as "Do Female/Junior rated fencers achieve different results than the average fencer of an equal rating", which would be easily testable by querying the the askfred database and performing stats.

    Then you could ask "On average is a fencer more likely to earn a rating in a restricted (female/male/junior event) than a mixed open?"

    You should be able to normalize for the fact that people attend these tournaments at different rates.
     
  16. Bonehead

    Bonehead Podium

    Joined:
    Aug 27, 2002
    Messages:
    2,563
    Likes Received:
    137
    Ratings are cross gender.

    So, if for example, I am in a region where there are 4 female A's and no male A's. And just for simplicity sake, to start, let's assume that there wasn't an abnormal case of people purposely not entering tournaments to purposely limit the possibility of Ratings. So we assume that everyone has access to tournaments, and everyone enters all the tournaments that they're allowed to enter and willfully fences to the best of their ability.

    If those girls entered a mixed tournament, they would bring their rating to the tournament and up the classification. If I entered that mixed tournament, and earned my A there, arguably it would be in relation to their Ratings, regardless of gender. If I was unable to earn a rating, the same thinking applies.

    So if you took a subset of the fencers from that mixed tournament, the A's should be the same, or have the same value, or however you want to describe it, as Mixed Gender A's. And one would expect that it's on average, no harder or easier to earn a classification in any particular subset tournament.

    I guess in concept it seems like ratings are currency, and even though there are partitioned trading areas (single gender tournaments), there is an international currency exchange (mixed gender tournaments) so it should equalize.

    At least that, seems to me, how it should work on paper.



    But I also feel like there is definitely a discrepancy in average skill (ability to win bouts) between certain groups. I would think Senior events are generally stronger than Junior Events. I would think that Male events are Generally harder than female events.

    So one would think, assuming that Ratings are reflective of ability to win bouts, and that certain categories are more likely to win on average, then those categories should have a disproportionate number of rated fencers.

    And come to think of it, if there are compounding factors in any particular subgroup - well that should be reflective in the average ability to win as well.

    i.e. if We had a event for only people named smith - and there was a super talented extended family of fencers named smith, you would expect that the averafge person named smith is more likely to win, and more likely to be rated. Certainly that doesn't say anything intrinsic about the name smith, but the question isn't one of causation - when you ask "Is a rating easier to earn in X group", you're asking "Are ratings correlated to ability to win".
     
  17. Zebra

    Zebra DE Bracket

    Joined:
    Dec 25, 2013
    Messages:
    771
    Likes Received:
    181
    Except for the absence of any decrease in rating for not performing as well as the rating would predict.
     
  18. Privateer

    Privateer DE Bracket

    Joined:
    Feb 8, 2009
    Messages:
    951
    Likes Received:
    106
    does the degradation of a rating have this effect, albeit at a slow pace?
     
  19. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    You not accounting for fencers with an A in more than one weapon here is far less of an issue than ... not knowing how many of these women and men actually fence each weapon.
     
  20. Blackwood

    Blackwood Made the Cut

    Joined:
    Mar 23, 2011
    Messages:
    356
    Likes Received:
    24
    True, but what I was getting at is that about 3.6% (1.37% + 1.80% + 1.37% - [% with an A in multiple weapons]) of competitive women fencers have earned an A in at least one weapon, and similarly about 4.5% of competitive men fencers have earned an A in at least one weapon.

    I'm not sure the numbers tell us anything though.
     

Share This Page