When Did US Fencing Begin Using Ratings?

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

  1. Peach

    Peach Podium

    Joined:
    Feb 10, 2001
    Messages:
    163
    Likes Received:
    766
    I'd argue it rewards one-time results too much.

    But your statement reminds me of the guy at the club who used to complain about how unfair that I had a higher rating than he did because I was a woman, and then argued with the referee for half an hour after I beat him in an open.

    At one point long ago, women were seeded into events at one classification lower than their letter, which was pretty ridiculous. I earned a B in a women's event and a B in a mixed veteran event; which one should count more?

    I tend to chime in on these threads to say that while the seeding method isn't perfect, it generally produces consistent results, which is the purpose of seeding. And a more accurate method might be, as K O'N remarks, lend itself to gaming. I wouldn't mind being able to fence Div II again myself.
     
  2. DangerMouse

    DangerMouse Podium

    Joined:
    Mar 13, 2003
    Messages:
    3,184
    Likes Received:
    196
    K O'N and I tend to fall into the same camp with this argument. My big issue is the assertion that seeding pools relatively evenly is the same as predicting outcomes. They really are not.

    If you want to seed pools so everyone has about the same difficulty level, then pool seeding should only be based on prior results of 5 touch bouts to be most accurate. Some fencers win a lot in pools, but then are easy to figure out when fencing to 15 in DE rounds.

    If you want to predict final results, you should really only be looking at DE results. This is what our current system does, but it's also what every proposed system I have seen does as well.

    While the "fairness" of seeding tends to be the central argument when ratings threads come up, it really seems to be driven by an emotional response to people who "have ratings they don't deserve." I think this really is driven by jealousy and not because the weak A fencers cause a real seeding issue. So what if one pool has an "easy A" at the top of it? There are some pretty high level fencers that I beat consistently because my style works well against theirs. There are also some lower level fencers that beat me consistently because of the reverse. That's the nature of combat sports. That's also what I love about competition.
     
  3. Montoya

    Montoya DE Bracket

    Joined:
    Aug 7, 2004
    Messages:
    676
    Likes Received:
    51
    In Veterans we have the opposite problem - people who are seeded lower than they "deserve" which is actually a disservice to those who are rated above them. For example a Vet 50 who has an A15 but no veteran points will seed below the top 8 - as well as all other A fencers with points.
     
  4. Montoya

    Montoya DE Bracket

    Joined:
    Aug 7, 2004
    Messages:
    676
    Likes Received:
    51
    And the problem is?

    My last two ratings have been earned in Vet events. I fenced three D1 events this season. My final standing in these competitions was 14- 40 places higher than my initial seeding.
     
  5. dridge

    dridge DE Bracket

    Joined:
    Sep 3, 2006
    Messages:
    206
    Likes Received:
    32
    I thought that the goal of seeding is to prevent the strongest fencers from facing one another early in the competition. For instance, in pools. When there are many strong fencers (such as during an Div 1 NAC), then points distinguish the stronger from the weaker fencers. This approach has important limitations, but the current classification system seems pretty robust for most purposes. It breaks down when competitions include many fencers with similar results.
     
  6. dberke

    dberke Podium

    Joined:
    Jan 17, 2003
    Messages:
    1,516
    Likes Received:
    243
    I'd like to point out that even the FIE isn't too concerned with "evenly balanced" pools. World Cups and Grand Prix are seeded only based on current world rankings, nothing else. So if a fencer stops competing for a year, they will have no ranking and thus be seeded near the bottom when they re-enter competition, regardless of skill.

    At the saber Grand Prix in NYC back in December, Tim Morehouse was on the list of USA fencers to add to the event had there been slots available. (The host country is allowed to enter additional fencers to fill out any pools of 6 to make them pools of 7.) It turned out that the number of entrants was exactly divisible by 7, so no additional USA fencers were added. Had there been some open slots allowing Tim to fence, he would have been placed in a pool of 6 as the lowest seed (since had no rank for the 2013-2014 season.) Fair? Maybe not. Does the FIE care? Nope.

    This pursuit of "perfectly seeded pools" is a red herring. Our current system works fairly well in this regard. National points are used to seed fencers in Div I, and this follows the FIE model - it's based on current point standing. (I'll point out that our system of letter classifications would seed Tim better than the FIE - even if he didn't have current senior points, he'd still be seeded fairly high since he has an A classification.) If we don't feel that seeding by national points is sufficient (i.e. too many A's without points), then we should award points to more fencers to get more granularity. A similar approach could be taken with other events - for example, if we awarded points in Div II and III, then we could seed those events with points. That would be better than the current approach where all C15s are randomly seeded before all C14s, then C13s, etc.

    Dan
     
  7. Zebra

    Zebra DE Bracket

    Joined:
    Dec 25, 2013
    Messages:
    771
    Likes Received:
    181
    You're right, Peach. The first rated fencer I ever beat in competition was a C, not an E. I'm pretty convinced it was because she had trouble with lefties, not because I'm any good (though I had enough point control that particular bout to hit what she left open). I could easily envision a situation where one of those upsets in a C1 would result in my getting a rating that doesn't reflect my actual (in)ability.
     
  8. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    I'd like to point out that the FIE isn't too concerned with pools, let alone "evenly balanced" ones, given that they exempt the top 16 from them entirely :)
     
  9. Privateer

    Privateer DE Bracket

    Joined:
    Feb 8, 2009
    Messages:
    951
    Likes Received:
    106
    interesting discussion about ratings over beers yesterday (very large, strong IPAs, so forgive me if this sounds half-baked), and while the purpose of ratings is to effectively seed pools, I think it has taken on more than that.

    The USFA has decided that you need 6 U's to have an E1 event. One of those six U's earns an E. Right or wrong, that's the formula and who are we to argue that 1 E doesn't 'deserve' their rating.

    That established, our beer-soaked hypothesis is the number of U's far outnumbers that 1:5 ratio of E's:U's. I tried crunching the member list to figure it out but it was beyond me because of the lack of data (ie - I didn't see a way to filter out 1 weapon's worth of U's). Clearly someone smarter than me can figure it out.

    If our hypothesis proves correct, that tells me ratings are not doing what they're supposed to do and therefore not seeding correctly.

    Many people use ratings to measure their progress as a fencer, right or wrong that's the only measure we have right now. The lack of measurable progress, also a hypothesis, leads many fencers, kids, and adults alike, to quit or just not compete. That's the real problem we face, IMO.
     
  10. wwalkerjr

    wwalkerjr Made the Cut

    Joined:
    May 19, 2008
    Messages:
    233
    Likes Received:
    24
    My pool results will show that I don't always get the easy pools (for me).
     
  11. jkormann

    jkormann Podium

    Joined:
    Nov 8, 2002
    Messages:
    2,733
    Likes Received:
    148
    When the ratings system was created, weren't events run under the older multi-pool system? I've heard opinions that format actually determined who was the better fencer in the event. Which could explain the logic why 6-Us could promote one to an E-rating.
     
  12. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    You need better beer.
     
  13. Privateer

    Privateer DE Bracket

    Joined:
    Feb 8, 2009
    Messages:
    951
    Likes Received:
    106
    blasphemer!
     
  14. David Sierra

    David Sierra DE Bracket

    Joined:
    May 14, 2013
    Messages:
    714
    Likes Received:
    61
    Personally, I could give a rats a$$ about the way the FIE does something. They've proven time and time again to screw up by the numbers that I tend to look at their decisions as examples of the way things should NOT be done.
     
  15. mfp

    mfp Podium

    Joined:
    May 10, 2002
    Messages:
    1,828
    Likes Received:
    221
    What's your "MEASURABLE" bit for your "GOAL: the seeding of competitions ... be more accurate"?
     
  16. Ancientepee

    Ancientepee DE Bracket

    Joined:
    Jun 12, 2010
    Messages:
    951
    Likes Received:
    181
    The last one.

    You should be aware that most of the posters making comments on this thread don't have a clue about how a numeric rating system would be implemented or how it would function but that doesn't stop them from offering their "expertise" as to the faults of such a system. An Elo-type rating system does not predict the results of a competition, it predicts the probability that one fencer will beat another fencer in an individual bout between those two fencers based on the difference between their ratings. It can no more predict the final standings than you can predict the sequence of heads and tails that will result from flipping a coin fifty times even though you know the probability of each individual flip. So the measurable of the rating system's accuracy is not determined by looking at the final standings, it's determined by looking at the final results of each and every bout and seeing how accurately the rating difference between those two fencers predicted whether the higher or lower rated fencer was the one who won. This is not easy since it requires averaging a large number of bout results to get an accurate estimate of how closely the predicted result matches the actual result.

    The other major misunderstanding regards the comments about using the fact that ratings decrease to "game" the system. They think, for example, that if the criteria for entering a rating-restricted event (e.g., Division II) is, say, a rating of 1600, that fencers just above that rating could deliberately lose bouts to drop their rating below 1600 to make them eligible to enter that event. What the posters don't realize is that other organizations using Elo-type ratings have already addressed this concern by keeping track of "floor ratings" for each competitor which are used to determine eligibility to enter an event and these "floor ratings" are based on but different from the "current ratings" used to actually seed the events. So, for example, the USFA could say that each fencer has a "floor rating" which is the highest "current rating" that that fencer has had in the previous two, three, or maybe four years. Once a fencer's rating goes above 1600, they would not be eligible to enter Division II events for the next two, three, or four years no matter how many bouts they deliberately lose. The only thing that losing the bouts would do is decrease their "current rating" and cause them to be seeded lower in the events that they are eligible to enter and there's not much benefit in doing that.

    It is true that fencers whose ratings are just below 1600 could deliberately lose bouts to prevent their rating from going above 1600 but this means that they'll be constantly seeded lower in the competitions that they do enter causing them to face stronger fencers than they otherwise would fence and so has it's own disadvantages. Also, when they do fence in the rating-restricted event, their result there will most likely increase their rating above 1600 and so their strategy will be effective for only one event.
     
  17. teacup

    teacup Podium

    Joined:
    Mar 14, 2006
    Messages:
    4,872
    Likes Received:
    170
    But doesn't the FIE give points down to 64 with fewer than 160 fencers? So while not precise, it is additional seeding criteria.

    As does USA fencing which gives points to 40% of the field for SYCs, which helps seeding since so many youth events aren't C1 events and most youths are Us.

    So why not give points to 40% of the field at all national events? Then to carry this even further, to help with seeding and simplify qualification paths, why not give national points to placement at ROCs and RJCCs, the same as national points are given at SYCs? Forget regional points which aren't used for seeding anyway and just add another layer of confusion. Are the national coaches really worried that someone will be knocked off a team by someone earning a few points at a ROC or placing 100th at a Div I national event? (If not for this, then what is the reasoning behind so many point lists, not counting points 33-64 for team and not awarding a greater percentage of the field national points?)
    Are three sets of points really necessary? (Rolling, team, regional)

    In the current system, there is no distinction between an A earned by placing top 8 in a Div I national event with over 300 fencers and an A earned at a local youth event with 15 fencers. (Though I admit A1 local youth events are few, they do exist in some areas/weapons.)
     
    Last edited: Mar 25, 2015
  18. oiuyt

    oiuyt Podium

    Joined:
    Apr 26, 2000
    Messages:
    10,240
    Likes Received:
    974
    Even beyond points, event placement still factors into FIE seeding. Take 215th in a 215-person event and at the next event you're seeded ahead of people with no results (within the 12-month window). Currently there are 311 fencers with at least 1 FIE point in senior men's epee. Yet, the FIE ranks a total of 1,131 men's epeeists. The lowest of whom appear to be those who no-showed senior worlds in Kazan, "earning" a 999-place finish there.

    B
     
  19. Bonehead

    Bonehead Podium

    Joined:
    Aug 27, 2002
    Messages:
    2,563
    Likes Received:
    137
    How exactly does that work?
     
  20. teacup

    teacup Podium

    Joined:
    Mar 14, 2006
    Messages:
    4,872
    Likes Received:
    170

Share This Page