Fencing ratings

[mrbiggs]

While I do agree that the rating system probably won't be implemented without some sort of fencing results and events database, the USFA should have one of those anyway. If this is what pushes them to do it, so be it.

I'd just like to point out that within hours of this post, the USFA created a website to do just that. Coincidence? I doubt it.

[keith]

I'd just like to point out that within hours of this post, the USFA created a website to do just that. Coincidence? I doubt it.

so who wins the superbowl next year?

[DieterS]

Not wikipedia. I posted links to some articles that involved a more detailed statistical analysis of different competition formats.

I waded through all the posts and found the articles, which I appreciated. Thanks!

However; many people see the advantage of a more granular system as a way of tracking their individual progress.

Yes. From what I've seen, the letter system is fairly insensitive and non-linear in the amount of improvement that it takes to achieve each letter rating. For example, you have to be fencing at a B-level a long time before you are lucky enough to be in a tournament that has the right number and strength fencers finishing in the right places for you to "earn" your B. To earn your "A" is actually easier--you just have to beat a bunch of A's.

So a suitable seeding system is not necessarily the same as a system which provides individuals with the required detail as to there standing in the total population of fencers.

I don't see why not, as long as the seeding system is reasonably granular (but not too much--otherwise the fluctuations will simply be "noise").

The current system is good enough for seeding

I'm not sure what's "good enough." Or perhaps one can argue that precise seeding doesn't make that much of a difference for the majority of fencers: half will get knocked out in the first round anyway, another 10-20% will make the finals anyway, and nobody cares about the remaining 30-40% that don't make the finals.

If you want an ELO style system you have to acknowledge that the accuracy of the rankings is determined by the information yielded by the tournament format.

This is where I part company with George Masin's proposal. I believe that Elo ratings should be affected only by steel-to-steel results in 15-touch bouts, not on "expected placement" in the final tournament standings (too theoretical, too often falsified in practice, too dependent on tournament format).

The bottom seed of the 64 getting blown out in the first DE by the top seed provides no information as to whether than fencer was
better or worse than the fencer who sits in the 33rd position.

I completely agree--it's useless for sorting and wastes everyone's time!

The format used in chess is informative - if you get blown out in the first round you then get a series of matches against progressively weaker opponents. If you win in the third or fourth round you start getting matches against what are (more likely to be) equal strength players. So lots of useful information on competitive performance is gathered.

I really like the participation and equal matchup aspect of the Swiss system. I believe there are better sorting systems that make these matchups occur earlier in a tournament, but as I've posted elsewhere, the USFA has made them illegal to experiment with and has not responded to repeated requests for permitting them on a local level.

Every one else will end up in an ill defined numerical smudge which tells them less than the current letter system.

. . . assuming that the current DE system is used and Elo adjustments are made on the basis of tournament placement. But as you said, there are other tournament formats available that do a much better job at sorting.

Dieter

[eac]

1. Everybody so far has agreed with the idea that the fencing Elo system should not consider the 33rd place to have done any better than the 64th place. The only difference between people here is that the Elo-boosters say that this is true and that Elo is still good, while the anti-Elos say that this is true and that therefore Elo will not work. With occasional exceptions in Ds or E's, the letter system also does not pay attention to pool results. This is because most people (aside from Peter) seem to agree that the major noticeable result is the DE round you go out in, not the place within that DE round. I think an Elo system that ignores differences between people in the same DE round will still provide a significant improvement, for all the reasons stated earlier, and will have no numerical smudge related to 33rd vs. 64th issues, since those comparisons are ignored. Therefore, everybody should stop arguing about the 33rd vs. 64th place problem.

2.

Then there is the not small matter of getting an elo system that most are happy with. That is the first thing to work on, get that one solved an the rest become worth looking into.

You seem more hopeful than you did earlier in the thread. Does this mean you actually think there might be an elo system that most people are happy with?

3. KD5MDK: Excellent assessment of what is required. Do you think there's any chance that the USFA might actually commandeer FRED rather than making of its own? That would be much smarter, I think.

[Inquartata]

Any system I would support would allow you to verify your rating with two minutes, a piece of paper, and a hand calculator. The Elo system as present in chess, for example, follows this rule. The Masin proposal does as well. (He even provides a table so you don't need a calculator).

And I still don't want to be made to do it. I do not want to feel that I have to check and verify my position after every bloody tournament, just to make sure someone else hasn't screwed up. I don't CARE how "easy" it is to do so, it is an unnecessary imposition on my time and efforts and one which can be completely avoided by the simple expedient of not changing a system that works perfectly well as it is.

As to the epee only argument, even though rating inflation is most present in epee, this will not be true forever.

You keep on saying this as though its truth is self-evident. It is not. I see no indication that it is likely to become so. There are reasons why epee is the way it is, and they have do do with things peculiar to epee. Absent evidence that sabre and foil are inexorably headed the same way---and by evidence I do NOT mean additional declarations of certainty by you, or further references to the crystal ball you seem to think you have---there is no compelling reason to accept this prediction of the future of yours...

There is no reason at all to believe that sabre and foil ratings will not inflate too much as well.

Nor is there any to believe that they will...that's the flaw in all unsupported prognostications, the one the prognosticators usually neglect to mention.

Plus, you seem to be saying that epee can't be fixed until sabre is broken as well. Do epeeists have to live with a broken system until you're unhappy? I don't get it.

For the umpteenth time, I do not buy the assertion that epee is "broken", or even unduly battered.

However, granting for the sake of argument that it were, I simply turn your question around and ask why I should have to live with a distasteful change to my weapon merely to satisfy a desideratum of yours? Why inconvenience two weapons for the possible improvement of one? And I say "possible" because there is no guarantee that improvement is what will result. There is the Law of Unintended Consequences stillto reckon.

And don't suggest an Elo system be implemented just for epee. The USFA would never do it. It's all weapons or none.

I vote "none".

In the part of FoilyDeath's post that you excerpted above, he was responding to the many previous statements (some of them yours, of course) regarding math difficulty. And his answer is valid in that regard: There's no reason why you and your kinfolk's fear of mathy stuff should hold back the rest of the sport.

No, his argument is a fully formed one, and I attack it on those grounds.

Let me try to frame it in the form of a syllogism for the sake of clarity:

The sport would be advanced by an Elo system.
Dislike of math is not a sufficiently good reason to slow said advancement.
Therefore, an Elo system should be adopted.

Now, there are several problems with the argument, but the main one is the first premise, which he assumes is correct, that an Elo system would constitute an advancement of the sport. The validity of this premise cannot just be assumed in this fashion...

Now, IF the move to an Elo-like system would be beneficial is another matter entirely. See, you attacked FoilyDeath on the issue of whether the change would be good, when his comment was specifically about math. And then you assailed his logic for it.

No, he made two separate assertions, one explicit and the other hidden in an assumption. The one depends on the other. They are part of the same argument for a conclusion. They cannot be separated.

For gods sake! Why is the main problem with ELO always "its too hard"/Me no gud at maths! FRED could do it all for you, no questions asked. If you ever want to go check it yourself, it takes 5 mins and a piece of paper.

I'll keep on saying it until you understand it: The issue is not whether the verification process is easy or difficult. It is not whether it "could be done" thus and so. It is that there is no need for it to be done at all, and that we thus should not have to do it at all.

For the third time: However "easy" cleaning toilets may be, it is still unpleasant. If you have a maid named FRED who can do it for you, it is STILL unpleasant. And if you still have to make sure that the maid did the job properly, it's still unpleasant for you...

It's "easy" for everyone to switch to visor masks. That doesn't mean that no one should object to being forced to switch to a visor mask. This is just another change which some are seeking to force on everyone in order to satisfy their own personal beliefs about what is "good for the sport". In this respect they are little different than Rene Roch.

[FoilyDeath]

Its also not "pleasant" having flawed ratings. IF having to do a bit of work every time I believe my rating is flawed is all I have to do, then I will be more than happy to deal with it. If the ELO system doesn't actually help, well, we're right back where we started. I'm not saying we replace the whole system overnight, I'm saying we keep an ELO type system in the background, and see how it turns out. Simple, barely expensive than the simple we have in place.

[Inquartata]

I'm happy with the system we have. We don't have to do anything with it. It adheres to the KISS principle. It suffices.

I see no reason to experiment with change for the sake of change.

What exactly do you think such a change will improve in your fencing life?

Will it make you a better fencer somehow? Do you think it'll mean you get seeded higher and thus get an easier time of it in competitions? Do you think it will speed tournaments up? What?

On a cost-benefit basis I just don't see the point.

[Sciurus-Rex]

I do not want to feel that I have to check and verify my position after every bloody tournament, just to make sure someone else hasn't screwed up. I don't CARE how "easy" it is to do so, it is an unnecessary imposition on my time and efforts ...

No one is forcing you to feel that way, Inq. It's your own fear of the unknown and dislike of change that's pushing your buttons.

Let me try to frame it in the form of a syllogism for the sake of clarity:

The sport would be advanced by an Elo system.
Dislike of math is not a sufficiently good reason to slow said advancement.
Therefore, an Elo system should be adopted.

Now, there are several problems with the argument...

The main problem with the argument is that you constructed it yourself just so you could use the word "syllogism."

FoilyDeath never laid out the restrictive logical construction you posted above. He and others have repeatedly dealt with several points, many of which have nothing to do with each other. And each issue has come to its own conclusion, "There, THAT argument has been answered." Which is to say: If that concern is your only concern, and it's been adequately set aside, then you don't have any more reasons to fret. ... In no way did he say that ONLY because the math is simpler than you fear should Elo be supported. You're implying that yourself.

Before that, though, you began your reconstruction with the statement, "The sport would be advanced by an Elo system."

You're suggesting that the major premise has sprung fully-formed and untested from the void; and you're assuming the premise if false on its face regardless. Those are both faulty assumptions. The statement itself is a conclusion which many here are satisfied as tested valid or at least potentially true.

But on the strength of your own conviction, you've declared that to be a blatantly false premise, on which all other minor premises are inadequately hinged. ... Inq, if that's your basis, then, bloody well YES, every argument is going to look like a trick question from Aristotle's own mouth. And that belief allows you to come up with all sorts of nasty syllogisms of your own to serve your purpose of posting another declamation.

You have repeatedly put together something that looks more like this:

I don't like Elo.
Whatever you say goes directly against that conviction.
Therefore, you're wrong.

And then you have the audacity to argue minor premises based on a shaky major premise and blame the other guy for it. ... You do it very well, and it's nicely camoflaged in pretty sentences, but its still disengenous does your big brain a disservice.

[Inquartata]

No one is forcing you to feel that way, Inq. It's your own fear of the unknown and dislike of change that's pushing your buttons.

Right...and no one "forces" groups to observe elections, lest there be voter fraud, either. And yet, do you think that voter fraud goes on, and that maybe the observation is a good thing, and necessary?

The main problem with the argument is that you constructed it yourself just so you could use the word "syllogism."

It's a good word. Here's another: Enthymeme.

The whole argument is in fact here:

"I don't see why lazyness and ineptity at maths should slow down the entire sport."

Implicit in the last phrase is the assumption that failing to adopt an Elo system ( because of the reasons in the first phrase ) will be harmful---failing to do so will "slow down the sport". Meaning that an Elo system will "speed up" the sport. No Elo bad, Elo good. This is what's known as a suppressed or unstated premise.

Well, actually it IS stated, but apparently it's well camouflaged, because you cannot seem to see it.

Anyway, it's there, and it's an integral premise of the argument. To simply accept it and address only the other premise is to give the argument force it does not deserve.

You're suggesting that the major premise has sprung fully-formed and untested from the void; and you're assuming the premise if false on its face regardless.

Nope. The premise is there in FD's argument: "slow down the entire sport". His words.

The premise may be true; but it has not been demonstrated to be so. FD is merely assuming that it is true, and asking us to accept that it is true. I cannot do so without proof...

Those are both faulty assumptions. The statement itself is a conclusion which many here are satisfied as tested valid or at least potentially true.

Sounds more like faith to me. It hasn't been tested, so accepting that it "tested valid" can hardly be anything else. And "potentially true" is not the same thing as true. For a conclusion to be true its premises must be so as well. ALL of them. True, not "potentially true". Still less hopefully true.

But on the strength of your own conviction, you've declared that to be a blatantly false premise,

Reread my posts. I said that it has not been shown to be true.

Now, I believe it to be false---just as you believe it to be true. But the burden of proof lies with he who advances an argument. It is up to the proponents of Elo to demonstrate that it will advance "the sport". Demonstrate, not assume.

I don't like Elo.
Whatever you say goes directly against that conviction.
Therefore, you're wrong.

I despair of your grasp of the basic principles of logic, then.


If anything, my counterargument would be something like:

The current system is adequate for seeding and skill-level tracking.
Adopting a new system would cost more than it was worth and might not work out as desired.
Therefore, there we should retain the present system.

[eac]

K people. At this point, Inq is not just slowing down fencing, he's slowing down the thread. Recommendation: stop wasting time talking about syllogisms. Many a thread has been mired and smothered in an illogical duel to the death between Inq and some hapless soul who fell into the trap. Let's not do that here.

[Inquartata]

Yea, brothers-ah! Cast ASIIIIDE the Devil's tool-ah, called logic, and just BELIEVE! Faith is all that ye need-ah! Hallelujah, and St. Elo shall set us faREE!

[academe]

Sounds more like faith to me. It hasn't been tested, so accepting that it "tested valid" can hardly be anything else. And "potentially true" is not the same thing as true. For a conclusion to be true its premises must be so as well. ALL of them. True, not "potentially true". Still less hopefully true.

I knew it, ELO is just another Wedge issue by the Intelligent Design folks to undermine evolution...er, wait...that's another thread.

I agree with Inq. (An end of times sign for certain, but alas, when I think he's right...) I don't see a driving need to institute anything more complex then what is in place now. I haven't seen too many people held back by the rating system, and I haven't seen too many people hurt by having to fence a "soft" or "weak" A. In fact, I'd argue that at big events, you are more likely to benefit from having a soft A or B in your pool (making your path to DEs a bit easier) if you are a decent fencer that finds you are stuck in a tough division/section and your rating is a C or even a D.

The bottom line to me has always been that you have to win consistently to get anywhere in this sport whether it be the top ten on any points list or top dog regionally. It may be due to having some bit of luck--no sport is without having some form of luck involved. But I'd argue that training and learning how to improve by competing is the key to being prepared to have a good day on those days when the seeding gods are smiling on you.

This will take you a lot farther than endlessly talking about how an artificial rating system is holding you back.

[KD5MDK]

I'd just like to point out that within hours of this post, the USFA created a website to do just that. Coincidence? I doubt it.

Actually, the USFA posted a membership list and ratings list. It does not appear to have any interest on that page in tournament results.

[PeterGustafsson]

Hi!

I'm happy with the system we have. We don't have to do anything with it. It adheres to the KISS principle.

Well, I do not think of myself as an example of the 2nd S, and if you are implying that - I have an issue with that.

It suffices.

Yes, it suffices perfectly well in what you want it to do. NOT with what I want it to do, though.

I see no reason to experiment with change for the sake of change.

I guess that us who want to tinker just have to outvote you, then!

What exactly do you think such a change will improve in your fencing life? Will it make you a better fencer somehow?

Definitely not.

Do you think it'll mean you get seeded higher and thus get an easier time of it in competitions?

I would be seeded higher in 50% of the competitions, and lower in the other half. Not a motivation for me personally.

Do you think it will speed tournaments up?

No, but it will not slow them down either, since all processing will be done after the competition ends.

What?

I would enable me to see whether I am #355 in Swedish ME, rather than #356. Also, if I go from #356 to #355 when the new ratings are published, that would be an ego boost.

On a cost-benefit basis I just don't see the point.

I would be perfectly willing to take an increase of 50% in the competition licence, if a granular and reasonably accurate rating system - which extends down to the DFL fencer in Sweden - were instituted with those monies.

You might want to do other things with the money you spend on fencing, but given that I am not national team material there is nothing the SvFF does not do that I can reasonably expect from them - except a good rating system.


Have a nice time!

Peter Gustafsson

[Sciurus-Rex]

I despair of your grasp of the basic principles of logic.

Right back atcha.

[downunder]

You might want to do other things with the money you spend on fencing, but given that I am not national team material there is nothing the SvFF does not do that I can reasonably expect from them - except a good rating system.

You're arguments are useless in this thread because you're using your experiances in Sweden to argue change to the Americans.

THh reason why a numeric rating system or comprensive ranking would be catastrophic in the US is the same as why a letter based rating would be inappropriate in Sweden.

[eac]

Wait, huh? Numerical ranking is *catastrophic* now?

[downunder]

Wait, huh? Numerical ranking is *catastrophic* now?

a global ranking across the US updated regularly accounting for different strengths and weakness in areas and competitions i feel would be catastrophic.

How many currently rated fencers do you have?

[eac]

18,000 rated members of the USFA. The US Chess Federation does exactly that with 100,000. Could you explain further why this would be catastrophic?

[pokey]

How many currently rated fencers do you have?

Judging by mrbiggs' and JEC's compilations here and here, roughly about 1800 in sabre, 3200 in epee, and 3400 in foil with ratings above Unclassified. As eac pointed out, this is out of a total of about 18,000 registered fencers.

(numbers rounded since it was indicated there may be errors)