I believe oiuyt took "online" to mean "must have an internet connection available at the venue".
Personally, it seems like any numeric system involving local tournaments will probably need a standard update frequency, such as once a week or month, and results will be released at those intervals, rather than constantly.
Personally, I rather like the idea presented in the Task Force proposal of a National Points List that used the strength factor formula used for World Cups (with some modifications). However, it would take some calculations of results as they would come out to determine if that structure would produce results that match what we want.
The link on the comparison of ranking systems is interesting, what I think it emphasises is that the ranking system needs to be sport specific to allow a useful ranking for that specific sport/activity.
The USFA seems to have three distinct groups it needs to/wants to rank;
Top athletes for national squad, FIE attendance.
Nationally active athletes below this level - both developmental and recreational.
Local/regional competitors who might on rare occasions compete at national events when they are hosted locally.
The first group are well covered by the current national points system and the last group by the letter rankings. It is the middle group who seem to causing the problem (the problem being to many of them showing up at certain events).
Now clearly there are a hundred and one ways of increasing granularity at the middle level; from rejigging the current letter system to any of the numerical systems. However the aim of the changes in the system seems (and I may have the wrong end of the stick) to be about reducing the number of fencers eligible for the USFA run competitions, or at the very least that there is a link between the national tournament re-organisation and any 'new' rating system.
An extended points system would have a clear benefit of allowing the USFA to manage the pool of eligible fencers for each event it hosts. However excluding a portion of current national competitors without providing an alternative national outlet is an issue I think when it comes to developing the sport.
An extended points system (similar to the British system) through DIVI, II & III national events would allow that, although it depending on implementation it may also force fencers to graduate through the divisions unless some regional events were also included (qualifiers for example or 'bonus' points for the following season earned at summer nationals).
Now I think extending the points system to all events (as the British system does) would be alot of work for little benefit in the US given the difference in the number of active fencers. Not that it wouldn't be cool to see where I ranked in the whole country based on my two annual competitive outings
I believe oiuyt took "online" to mean "must have an internet connection available at the venue".
Actually I didn't. I have no problem with organizers inputing results into a national database at the conclusion of an event or the next day, etc. If the system has lags of a day or two here or there I don't see a problem.
Similarly, doing what the Brits do with a monthly update would also accomplish all of the important goals. With that it becomes significantly less important exactly when organizers file the results (with the exception being events in the last day or two of a month).
What I had been assuming was that there would be an online system for inputting the results which then feed into the ratings calculations, obviating the need for the National Office to deal with hundreds of pages of mailed or faxed hardcopy results. Or even emailed results.
-B
__________________
"Oh but you can't expect to wield supreme executive power just because some watery tart threw a sword at you!"
If only there was a fencing results and events database for the results to be entered into.
Oh, well.
Aside from that, the discussion so far seems to have come back to a revamping of the results entry system, with the method of classification being something we can work out after we've got the dependable, easy, and fast method down.
You're right. If anything requires more than a single step we should abandon all discussion about it. :eyeroll:
Having an online results system and/or online classification reports is a desireable thing whether or not there's a change in the classification system that requires such. I'm not sure anyone's arguing otherwise here.
The fact that we don't already have one slows the implementation of any change to a system that requires one. Okay, stipulated. If it were decided upon as a key step the USFA needs then it's unlikely to be a limiting factor, as I see a gradual roll-out of major changes in any event. There's no way we're going to massively rework the system in the next two years.
Which brings us back to the question of whether or not the change-over is something we want, and if so, in what form?
Quote:
Originally Posted by mrbiggs
For example, you'd get the rating of anyone you beat in a DE minus one if that's higher than your rating, and if you lose to someone two ratings lower, you drop one.
I don't know if that would work, I'm just thinking something along those lines.
I feel somewhat ambivalent about picking apart preliminary suggestions, but this one screams of problems. This completely DOESN'T have the characteristics that I described as being desireable:
Quote:
Originally Posted by oiuyt
With a system that allows for frequent (or more frequent) updating we can allow for changes to happen more easily as any errors thereby induced can be corrected in further modifications in the near future (rather than 4-8 years from now).
And for the less subtle point, with any of the numeric systems thus far proposed people shift up and down more often and should, generally, localize not too far off from their normal, expected performance. This results in fewer people being poorly classified, those that are being less poorly classified, and the errors in classification lasting less time.
Okay, errors don't last very long, your system has that going for it.
Unfortunately it has a rediculously high flukiness factor that means tons of errors* (including large numbers of significant errors) pop in frequently. People won't do a particularly good job of settling on their true value, but rather will chaotically swing back and forth across a wide range. This would completely break our current system for qualification into senior national events (Want to fence D3? Go to a couple of local events and drop bouts to U's the week before the deadline. Want to fence D1? Better not risk losing to that E having a career day or that collegiate fencer who doesn't normally compete USFA unless you have another couple of tournaments lined up to redeem yourself before the deadline.). The system would be considerably worse than what we have now and picks up whatever downside can be attributed to needing to have constant classification updates.
And that leaves aside the point that A's are impossible to get. I'll assume that further refinements of your system would include a modification to fix this gap and prevent everyone from eventually settling into being U's. I can think of several easy patches that accomplish this, so won't bother using it as a counter-argument.
-B
*error being defined as someone whose classification is noticibly far from his/her actual ability.
__________________
"Oh but you can't expect to wield supreme executive power just because some watery tart threw a sword at you!"
You're right. If anything requires more than a single step we should abandon all discussion about it. :eyeroll:
Having an online results system and/or online classification reports is a desireable thing whether or not there's a change in the classification system that requires such. I'm not sure anyone's arguing otherwise here.
The fact that we don't already have one slows the implementation of any change to a system that requires one. Okay, stipulated. If it were decided upon as a key step the USFA needs then it's unlikely to be a limiting factor, as I see a gradual roll-out of major changes in any event. There's no way we're going to massively rework the system in the next two years.
Which brings us back to the question of whether or not the change-over is something we want, and if so, in what form?
I agree.
I'm just saying that the change in system will be several years in the future. (At which point it might be reasonable to require tournaments to have internet access, for example.)
Quote:
Originally Posted by oiuyt
Okay, errors don't last very long, your system has that going for it.
Unfortunately it has a rediculously high flukiness factor that means tons of errors* (including large numbers of significant errors) pop in frequently. People won't do a particularly good job of settling on their true value, but rather will chaotically swing back and forth across a wide range. This would completely break our current system for qualification into senior national events (Want to fence D3? Go to a couple of local events and drop bouts to U's the week before the deadline. Want to fence D1? Better not risk losing to that E having a career day or that collegiate fencer who doesn't normally compete USFA unless you have another couple of tournaments lined up to redeem yourself before the deadline.). The system would be considerably worse than what we have now and picks up whatever downside can be attributed to needing to have constant classification updates.
And that leaves aside the point that A's are impossible to get. I'll assume that further refinements of your system would include a modification to fix this gap and prevent everyone from eventually settling into being U's. I can think of several easy patches that accomplish this, so won't bother using it as a counter-argument.
-B
*error being defined as someone whose classification is noticibly far from his/her actual ability.
Yeah, I just thought that up in 2 minutes as a simple possibility.
A more complex and possibly more accurate method would be PeterGustafsson's method, in this thread. (Not that it doesn't have its own problems. I wasn't trying to suggest a method that can actually be used, just an example of how the idea could be changed to accomodate the internet.)
My point is that an online database that allows for classifications to be updated more frequently does not necessitate a numerical system.
Actually I didn't. I have no problem with organizers inputing results into a national database at the conclusion of an event or the next day, etc. If the system has lags of a day or two here or there I don't see a problem.
Similarly, doing what the Brits do with a monthly update would also accomplish all of the important goals. With that it becomes significantly less important exactly when organizers file the results (with the exception being events in the last day or two of a month).
What I had been assuming was that there would be an online system for inputting the results which then feed into the ratings calculations, obviating the need for the National Office to deal with hundreds of pages of mailed or faxed hardcopy results. Or even emailed results.
-B
Hm.. I appear to have miscommunicated. Because that is what I understood you to mean.
mrbiggs quoted you and replied:
Quote:
I agree that the main problem with the current system is with the rare updating. But rating updates after more tournaments requires a better system. (Most likely, an online system.)
Which I interpreted to mean that he was imagining a system that required internet access at the venue.
oiuyt then says:
Quote:
I hadn't realized it wasn't being taken as a given that any of the numeric systems talked about thus far would require an online system. I believe they all involve updates after either all or, minimally, many tournaments.
This is rather convoluted, but I think you mean that some people have suprised you by thinking that a numberic ranking system might be implemented without an online results component.
I think my reply was caused by getting confused by this statement; I had the impression you were saying "we don't need an online system if you use that to mean you must have internet access at the venue to run a tournament". Because indeed, we don't.
I hope that gets us all on the same chapter (page seems a bit much to hope for).
My actual thoughts on how a system might work will be given in a following post.
(At which point it might be reasonable to require tournaments to have internet access, for example.)
I don't see anything wrong right now (well, implemented in August 2007) to require tournament organizers to have internet access within 2 days of the start and end of a tournament. For weekend tournaments, you say the rankings from the previous Wednesday must be used (so you have all of Thursday and Friday to download them) and that tournament results must be turned in by the succeeding Tuesday. They will be verified by the National office (not a labor intensive task, just looking for anomolies and waiting for protests to be filed or Division sanctioning information to be received.) the following Wednesday (8 days later). If a tournament does not meet the deadline, then its results are delayed a week from the standings.
The only issue I see with this system is that Division officers, who are responsible for sanctioning the tournament and turning the results in (because they're much easier to hold accountable than assorted club members) might be a little slow regularly and miss the Tuesday deadlines, and that would be a disappointment. However, in the general course of things if it takes an extra week to get a tournament's results in, that shouldn't be a substantial issue. Put a two or three week extra deadline in front of National tournaments (the official results of 3 weeks ago will be the ones used for seeding) so that events close in can have their status sorted out, and all will go reasonably well.
If it helps anyone feel more confident that a numeric rating system can be divised that accomplishes all/most of the desired effects while avoiding the pitfalls:
Any number of possible calculation systems can be tested using real bouts and results data to make (reasonably) sure that the desired effects are acheived. FRED currently has about 140,000 bouts and 70,000 results to feed prospective rating calculation algorithms.
In fact, I already did it with the table tennis system, and it's pretty interesting. (According to the numbers, I'm not quite as good an epeeist as my letter rating would have you believe. )
Once we have a reasonably complete list of goals and requirements for a new system, we can develop any number of rating systems and run the last two years of FRED's data through them. Then compare side by side what comes out the other end to see which system came closest.
Also, I echo what oiuyt has already said, (i think it was him): I imagine any new system would be run in parallel with the current one for about a year before "going live".
You know peet, I asked for something exactly like that to be done in my letter to the Tournament Task Force.
Do you have a copy of the Task Force ideas? I am very interested in the suggestion of using a Points List that took into account strength factor very similar to the FIE formula, with some slight modifications. Now, the suggestion as given left what sort of point chart there should be completely up in the air. However, if you were to use the Div I point chart, what would the results look like? What are the strength factors of various FRED tournaments? Are there local events (like a Mercury Cup, for example) that hit the 2.0 cap on their own?
Do you have a copy of the Task Force ideas? I am very interested in the suggestion of using a Points List that took into account strength factor very similar to the FIE formula, with some slight modifications. Now, the suggestion as given left what sort of point chart there should be completely up in the air. However, if you were to use the Div I point chart, what would the results look like? What are the strength factors of various FRED tournaments? Are there local events (like a Mercury Cup, for example) that hit the 2.0 cap on their own?
Good questions. I'll see what I can do to answer some of them, though i can't promise it'll happen right away (day job & all that ).
Yeah, I've read the task force doc (just re-read it while posting my last post). The FIE style system should be pretty easy to implement.
Of course, really the thing to do first is to get all the desired characteristics of a rating system together (as much as possible), and think about a system that will accomplish those goals.
Don't you realize that using a points list for the elite while only using a generic bucket system for the rest of us is unfair and elitest???
-m
Unfortunately, those who compete in systems which don't use our rating scheme pretty much have to conform to the requirements of that system. But then, those elite fencers also have ratings under our system.
More importantly, using a point system for the top handful of fencers does not disadvantage anyone else. It satisfies the Pareto principle, IOW.
__________________
Use the Shift key, people! Keyboard manufacturers everywhere are ineffably saddened when you ignore what they made just for you!
Given that there BEING a switchover, much less what form it will take has not been decided this seems somewhat premature.
You are more trusting than I.
Quote:
Originally Posted by seak
I attended a meeting today as part of coaches college with Michael M. and Christine Simmons and he stated that
2) They have moved the membership database over, and as part of the new database their is a field for numerical ranking, which is where they are going. Though the new format hasn't been determined yet.
I am willing to go out on a limb on the basis of this slip by MM and say that despite all reassurances that no decision has been made we will be hearing an announcement that we ARE going to a point system of some sort.
__________________
Use the Shift key, people! Keyboard manufacturers everywhere are ineffably saddened when you ignore what they made just for you!
I am willing to go out on a limb on the basis of this slip by MM and say that despite all reassurances that no decision has been made we will be hearing an announcement that we ARE going to a point system of some sort.
And there are those of us, not in the elite, never will be in the elite, and whose chances of actually coaching anyone in the elite are slim, who say, "ABOUT FREAKIN TIME!"
Not every non-elite fencer is opposed to a numeric system. Some of us are actually excited about the prospect. It would be interesting to see what the vote would look like across the membership...
__________________
"Prefiero morir de pie que vivir siempre arrodillado." - Emiliano Zapata
OK. But you still haven't addressed the fear itself, the protection of a rating.
Yes, of course it's a sad day to see your position bumped. Especially if you're right on the edge of qualification for an important team or competition qualifier.
But that's an inherent risk for any serious athlete, isn't it? And don't we want the best of the best to fight their way to the top?
right, it is a risk... but it's an unnecessary one at a local event. there's simply not enough for a top fencer to gain and way too much to potentially lose. under the current system, there's still not much to gain, but there's also nothing to lose, so there isn't a disincentive for strong fencers to attend local events. under a proposed ELO style system, there is a great disincentive.
Uh, I'm pretty sure that Brad doesn't need to be trusting to say that a decision hasn't been made... he's kinda part of the decision making process...
You're not cynical enough. I'm part of both the group tasked with making suggestions for changes and the group that has official responsibility for deciding what changes to make and whether or not to make them.
That doesn't mean I'm part of the shadow group(s) that ACTUALLY runs the USFA. It's this secret cabal ensconced in their mountain-top lair in Colordao (aka SPECTRE headquarters) that's already made the decision. :)
-B
__________________
"Oh but you can't expect to wield supreme executive power just because some watery tart threw a sword at you!"
right, it is a risk... but it's an unnecessary one at a local event. there's simply not enough for a top fencer to gain and way too much to potentially lose. under the current system, there's still not much to gain, but there's also nothing to lose, so there isn't a disincentive for strong fencers to attend local events. under a proposed ELO style system, there is a great disincentive.
Of course, the countervailing argument under an Elo or other non high- watermark system, is that strong fencers may have more of an incentive to go to smaller tournaments if there is a chance of improving their rating, which there wouldn't be under the old system.
An A06 has no particular (ratings) incentive to attend any local tournament. Similarly, where the tournament (either through lack of 15 participants or lack of of enough of the right classification fencers) is unlikely to give a higher classification to a potential competitor, there is likewise no ratings incentive to participate.
Some classification systems would allow ratings increases in circumstances that aren't available in the present system.
Whether that incentive would be enough to counteract people "protecting" their ratings by passing on tournaments is an open question.
)
).
Linear Mode
