PointControl: a new fencing rating system

Discussion in 'Fencing Discussion' started by David Ma, Sep 15, 2015.

  1. Black Widow

    Black Widow Made the Cut

    Joined:
    Feb 14, 2017
    Messages:
    66
    Likes Received:
    36
    Thank you very much for all your hard work-very impressive!
    I found this incredibly helpful in analyzing my personal fencing performance:eek:
    Seeing a visual representation of tournaments over my first year of competing confirmed my suspicion that I am incredibly consistent, but not improving. Recently changed coaching to shake things up, so I am interested to come back and see how the graph looks later in the year.
     
    SevenDad likes this.
  2. Privateer

    Privateer Podium

    Joined:
    Feb 8, 2009
    Messages:
    1,174
    Likes Received:
    152
    i asked this question once before also but never heard a response. It seems to be once a week, but sometimes the 'feed' seems to break down.
     
  3. Chauffeur

    Chauffeur DE Bracket

    Joined:
    Jan 31, 2013
    Messages:
    95
    Likes Received:
    23
    This flurry of posts prompted me to look...seems like it is going on three weeks (at least for the fencers I look at). I have also on occasion found it updated Monday after fencing Sunday. I confess ignorance as to how the "scraping" works, but I just chalk this lag up to the site's owner paying attention to stuff that actually feeds him. It's fun to look at and I'll check it out again later.
     
    Privateer likes this.
  4. abouttothunder

    abouttothunder Made the Cut

    Joined:
    Nov 8, 2012
    Messages:
    143
    Likes Received:
    5
    I have a result from March 11th that hasn't updated, although the system recognizes that I fenced on that date. The site had been pretty useful for me for about a year and a half, and it's something I'd pay an annual fee to access. I expect it will all be moot anyway in the not-too-distant future with the new tournament system being implemented by the NO.
     
  5. Mitchell

    Mitchell is a Verified Fencing ExpertMitchell hi Staff Member

    Joined:
    Feb 24, 2003
    Messages:
    3,288
    Likes Received:
    501
    Its almost certainly going to be weekly, at a minimum. Probably mid week if I had to spitball. Might be even less frequent due to the order of bouts being actually important and the fact that organizers don't always immediately upload their results. (just my speculation from doing my own points system and seeing similar issues)
     
  6. abouttothunder

    abouttothunder Made the Cut

    Joined:
    Nov 8, 2012
    Messages:
    143
    Likes Received:
    5
    I used it pretty regularly back when I competed a lot. Updates for weekend events showed up on Monday typically. It is unusual for it to be a month out of date.
     
  7. brianz

    brianz Rookie

    Joined:
    Dec 30, 2014
    Messages:
    7
    Likes Received:
    0
    Thanks David Ma for coming up with this system. I really like the approach of this system but I also see a few potential blind spots.
    • For youth fencers, a few of them young budding stars are consistently fencing up and doing well when fencing up. That's arguably the hardest thing to do because of their lack of physical strength and they need to compensate with skills. This system hasn't factor in strength of opponent, or so it seems. I'd suggest we take a page book from RPI type ranking system and leverage that, so we could normalize top young fencers as some of them should be seeded higher. This should also help to normalize the results of kids who manage to win everything in their age group but not able to fence up.
    • The same approach should be applied to Vet that did well when fencing against, let's say, an Olympic hopeful vs. someone in their own age group. There are some folks who had a good day in a Vet event and got a high rating but struggle to renew. This often throws off pool assignment in ROCs.
    • I notice the system does not differentiate a 5-4 win vs. a 5-0, but we ought to consider adding that. Going to the previous points, if a 13 years old losing 15-14 to someone on the Jr NRPS, that fencer should be rewarded.
    • I'd also like to suggest factoring in strength of event. Since we could now have a numeral rating for everyone, the strength of event can be used to determine individual performance instead of event classification. We all have seen tough events went down in rating due to bracket flukes or weak events having ratings due to having just enough rated fencers. With this numeral approach, we taking the entire rating scale away and we can all focus on doing our best one touch at a time because every touch (sorta) counts. This should also give more weight to NACs than smaller local events.
    There's also another way to tackle the rating issue. The biggest challenge in the current system, as many have pointed out, is the quality and the consistency of the rating. One of the quicker fix without completely overhauling the existing system is to simply use the frequency of finish in a given range and shorten the renewal time period to, let's say, trailing 24 months. This will get rid of the silly calendar year approach and give injured fencers time to recover and not lose the rating. For instance, if a fencers have finished in A 12 times within the past 24 months, it should be A-12 which indicates consistency, hence getting a higher seed than anyone lower than A-12. If a fencer cannot finish within the same rating in 24 months, then his/her rating will be replaced by the next highest level finish and frequency. For example, an A-1 couldn't finish in A after 24 months, and could only finish in C 3 times during that timeframe, s/he would become a C-3. This should make the seeding more fair as this fencer would not be able to sit on an A for 5 years and a B for another 5. Also, this could alleviate the need to use NRPS in all local tournaments since most people on the Div1 NRPS would presumably have a higher A than the other local As.
     
  8. jdude97

    jdude97 Podium

    Joined:
    Apr 10, 2013
    Messages:
    2,022
    Likes Received:
    265
    System I just came up with now:
    Renew your rating: +1
    Year has elapsed since last increase: -1
    Go to 0: drop down 1 rating

    Examples:
    1) Current A15 has not renewed rating in 1 year. Will be grandfathered into system and designated as A1. If he renews A on June 15, 2017, he becomes an A2. He doesn't fence again for a year. He goes back to being an A1. He doesn't renew his A any time in the next year. Becomes a B1.
    2) Current B gets their A tomorrow, becomes an A1. Wins div1A summer nationals next month, becomes an A2. Would drop down to A1 in June 2018 and would drop down to B1 in June 2019 if he does not renew in that timeframe.

    Potential concerns:
    - Roll down every year instead of every 4 years. I know under old system used to be rolled down after 2 years. I would be ok with rolling down every 1 year given the opportunity to stock up multiple top finishes (i.e. guy who fences one tournament in 2016 and got B16 but doesn't do that well for the next year is probably not as good as a guy who gets a B finish four times then has a bad season).
    - Might be hard to track the roll off dates but could record it by month/quarter
    - Would disadvantage people in fencing deserts (current system does that too anyways -- this would be more like a patch to the current system than a total fix since it really needs an overhaul to properly value high finishes in large strong events, such as national events)

    Positives:
    - More nuanced representation of skill
    - Still rewards flukes in letting someone ladder up to a given record but makes it so they only remain there for a year while rewarding consistency
    - Reduces arbitrary nature of the calendar year (i.e. doing well on Jan. 1 worth way more than doing well on Dec. 31)
    - Encourages fencers to attend as many events as they can to earn higher seedings and for bragging purposes between friends and enemies alike (as is, many fencers will say it's pointless to attend an event at which they can't earn a new rating or even renew their own rating)
     
  9. oiuyt

    oiuyt Podium

    Joined:
    Apr 26, 2000
    Messages:
    10,284
    Likes Received:
    1,044
    You _know_ this?

    An additional concern is that the fencer you describe who earns an A and then fails to renew becomes a B1, regardless of how many B-level results she has earned, while a near-equivalent fencer who did NOT earn the A may be a B23 and, presumably, be seeded higher. If both plateau at a B level, this initial difference could be maintained indefinitely.

    B
     
  10. jdude97

    jdude97 Podium

    Joined:
    Apr 10, 2013
    Messages:
    2,022
    Likes Received:
    265
    I had a stinking feeling when I said this comment so matter of factly that it was wasn't going to be true. My source on this factoid is an Olympian from the 90s. Seeing as I was not born at that time let alone fencing, I took his word for it. Is this info in fact false?

    Good point. How about this: you can't go higher than _4 and when you roll down you start at _4. So in your example, an A1 rolling down would become a B4. Additionally, your B23 would have maxed out at B4. This also preserves the current 4 year roll off scheme. A different number could be chosen if desired. Another benefit of this modification is that fencers in places with lots of strong tournaments can't rack up so many finishes as to permanently place themselves higher than any of their peers in fencing deserts.

    Alternatively, keep track of how many times a fencer would have renewed their rating had they been one rating lower while also keeping in mind how many finishes at the previous rating they had before laddering up. For example, a fencer goes from U to A1 in May 2017, does not renew it for a year, but would have renewed a B twice in that period. In May 2018, she would become a B2, not a B1. If a fencer goes from B23 to A1 in May 2017 and does not fence for a year, she would revert to a B23 in May 2018. However, this solution requires a lot more data tracking.
     
  11. DangerMouse

    DangerMouse Podium

    Joined:
    Mar 13, 2003
    Messages:
    3,299
    Likes Received:
    244
    I don't see much difference between your new proposal and the current system except to punish fencers that have to travel to get higher level competition and, therefore, don't have as many opportunities to re-earn their classifications.
     
  12. jdude97

    jdude97 Podium

    Joined:
    Apr 10, 2013
    Messages:
    2,022
    Likes Received:
    265
    Yeah...
    Was just throwing out an off-the-top-of-the-head idea. More often than not they'll end up not being good :p
     
    DangerMouse likes this.
  13. DangerMouse

    DangerMouse Podium

    Joined:
    Mar 13, 2003
    Messages:
    3,299
    Likes Received:
    244
    I still think our current system is pretty good. It doesn't need to be super accurate because performance fluctuates all the time and people improve constantly too. I do think men's epee could use one more category of granularity just because of the size of some of the competitions, but it's really not a big deal because the people who would be in that category are going to be at the top of the results no matter where they get seeded to start.
     
    K O'N likes this.
  14. Mitchell

    Mitchell is a Verified Fencing ExpertMitchell hi Staff Member

    Joined:
    Feb 24, 2003
    Messages:
    3,288
    Likes Received:
    501
    Its moving to a good spot. National points restrictively measure the elite fencers in each category, alphanumerical ratings measure local maximum peak performance, and regional points measure participation (to some degree).

    I think if we dial in regional points better and then make better use of all 3 systems in coordination with each other, we'll have a fairly good measuring stick overall.
     
    DangerMouse likes this.
  15. fencer-dad-of-fencer

    fencer-dad-of-fencer Made the Cut

    Joined:
    Jul 4, 2013
    Messages:
    129
    Likes Received:
    11
    It doesn't look like the pointcontrol.info ratings have been updated for several months. Does anyone have information on its status?
     
  16. Zebra

    Zebra Podium

    Joined:
    Dec 25, 2013
    Messages:
    1,172
    Likes Received:
    341
    It's still up to date on the last time you competed, but the ratings haven't changed in months. If I had to guess at the cause, I'd say that some matrix or other data structure filled up, or the developer didn't anticipate having this much data, and didn't have enough cycles in a DO loop somewhere. Might be a very easy fix. David?
     
  17. abouttothunder

    abouttothunder Made the Cut

    Joined:
    Nov 8, 2012
    Messages:
    143
    Likes Received:
    5
    Yeah. A friend with fewer results has a relatively current graph, through April 2017. Mine froze as of last fall, though I have competed since then.
     
  18. RebelFencer

    RebelFencer Podium

    Joined:
    Oct 29, 2003
    Messages:
    2,658
    Likes Received:
    361
    So I'm a bit late to the party and was wondering if anyone could explain something about point control (I read the website and it didn't answer this question). I am an Epee fencer and generally do pretty well at higher rated tournaments, but somehow my Foil 'rating' on this is higher even though I only just occasionally do some local Div 3 events for fun. How does this happen?
     
  19. Privateer

    Privateer Podium

    Joined:
    Feb 8, 2009
    Messages:
    1,174
    Likes Received:
    152
    without looking, it's all based on who you beat or lose to. beating someone with a higher point control number will boost your number more than beating someone with a low number. conversely, losing to someone with a lower number will drop yours more than losing to someone with a higher rating. The amount you go up/down is all predicated on the difference between their number and your number.

    i'm pretty sure it also only pulls from askfred, so if you have event results not in askfred, it does not take those into account.
     
  20. RebelFencer

    RebelFencer Podium

    Joined:
    Oct 29, 2003
    Messages:
    2,658
    Likes Received:
    361
    I figured as much, but a 39 feels a bit high in a weapon that I have never had higher than a D in.
     

Share This Page