Yet another rating proposal...

[kapunga]

So I was poking around old posts and proposals to fix perceived ratings inflation and came up with the following idea. (Forgive me if someone else thought of it first, I didn't see it.)

It seems to me that one fact about athletic endeavor that the current rating system doesn't take into effect is that an elite level athlete requires constant training to maintain that level. An Olympic level fencer will become sub-Olympic real quick if they drop to a 2 day a week training regime. On the other hand, an unrated fencer who trains hard two days a week and has a decent level of athletic ability is almost guaranteed to eventually earn an E. Given this fact, why not make different ratings have different decay rates? For example:

• A - Must be re-earned within a year or it drops to a B.
• B - Must be re-earned within a year or it drops to a C.
• C - Must be re-earned within two years or it drops to a D.
• D - Must be re-earned within three years or it drops to an E.
• E - Must be re-earned within five years or it drops to a U.

I think this is totally fair, as an A level fencer should be expected to be able to maintain an A level or they are no longer likely fencing at an A level. On the other hand, an E has demonstrated a basic level of competency that is unlikely to leave even if the fencer stops fencing altogether for several years. Higher level fluke ratings would disappear rapidly before they have had too much chance to inflate the top end. Also, it would require no extra book-keeping, just a different heuristic for rating demotion.

[Peach]

I like the longer loss times for the lower ratings, as it avoids the "ringer" syndrome, but remember that in many weapons and parts of the country a rapid loss for the higher ratings would mean that no one in that section or division could earn a rating because there would be nothing but B fencers. I went through that in women's sabre a while ago, and until we had Division I competition and couple manufacture some A's artificially (top 8) we didn't have any A's. It's still hard for an A or B sabre fencer who doesn't do national competition to re-earn the rating (people do a good bit of traveling on the East Coast if they smell a B-rated competition). How about two years?

So I was poking around old posts and proposals to fix perceived ratings inflation and came up with the following idea. (Forgive me if someone else thought of it first, I didn't see it.)

It seems to me that one fact about athletic endeavor that the current rating system doesn't take into effect is that an elite level athlete requires constant training to maintain that level. An Olympic level fencer will become sub-Olympic real quick if they drop to a 2 day a week training regime. On the other hand, an unrated fencer who trains hard two days a week and has a decent level of athletic ability is almost guaranteed to eventually earn an E. Given this fact, why not make different ratings have different decay rates? For example:

• A - Must be re-earned within a year or it drops to a B.
• B - Must be re-earned within a year or it drops to a C.
• C - Must be re-earned within two years or it drops to a D.
• D - Must be re-earned within three years or it drops to an E.
• E - Must be re-earned within five years or it drops to a U.

I think this is totally fair, as an A level fencer should be expected to be able to maintain an A level or they are no longer an A. On the other hand, an E has demonstrated a basic level of competency that is unlikely to leave even if the fencer stops fencing altogether for several years. Higher level fluke ratings would disappear rapidly before they have had too much chance to inflate the top end. Also, it would require no extra book-keeping, just a different heuristic for rating demotion.

[downunder]

I don't think your proposal addresses the major concern that an A in one area is worth more than an A in another.

For a system designed to seed fencers in a round of pools it does a pretty good job. Perhaps individual regions could create their own ranking lists?

[kapunga]

I admit the current system does a good job seeding. However in my observation, the places where seeding gets thrown off tend to be by A's and B's with ratings two or three years old. Since there are often enough A's and B's for a combined two A's and B's per pool, people who end up in a pool with an A or a B who is out of practice and whose skills have declined end up with a decided advantage.

[keith]

I admit the current system does a good job seeding. However in my observation, the places where seeding gets thrown off tend to be by A's and B's with ratings two or three years old. Since there are often enough A's and B's for a combined two A's and B's per pool, people who end up in a pool with an A or a B who is out of practice and whose skills have declined end up with a decided advantage.

Ah the problem of 'fairness'.

In that case you need to have a seeding formula that prevents you ending up in a pool with all lefties, or folks two feet taller than you, or.....

[dberke]

I admit the current system does a good job seeding. However in my observation, the places where seeding gets thrown off tend to be by A's and B's with ratings two or three years old. Since there are often enough A's and B's for a combined two A's and B's per pool, people who end up in a pool with an A or a B who is out of practice and whose skills have declined end up with a decided advantage.

If an A's rating is older than the current year, their seeding reflects that. I would be surprised if the majority of A07s are not at least comparably strong as the current B09s.

While simple, the current system seems to do an adequate job. There will always be exceptions, but that's true with every system.

Dan

[AllezCat]

Another reason that a flat 2-year expiry would make sense is that it matches the amount of time it takes to age from one youth/cadet age group to the next. This would allow younger fencers to earn ratings without a "fluke" rating hanging around for two whole age groups.

[yearsofwisdom]

but we do also see A07s that may have stopped taking it seriously and suddenly come back top 8 to earn their A09 again. I think its fine the way it is now. There will always be flaws and exceptions for different people

[telkanuru]

Get points, who cares about ratings?

[epeeinacup666]

Get points, who cares about ratings?

Bingo!...National points should be the true criteria for seeding and ranking before any tournament in any area of the country,period...

That said,I want to also say that for me personally,renewing my A-rating is the most important thing to do at the beginning of EVERY year...even above Nat'l points or qualifying for Div-1 or SNs...Sure,ratings are mainly for tournament seeding purposes at best,but when one get the A-rating and lose it 4 years later because you were too busy telling everyone you're an 'A',I swore that when I get that 'A' again I would never let go of it...

I got my 'A' again back in 2001 and have renewed every it year since...that's my motivation for this crazy thing called fencing.

...

[yearsofwisdom]

how do u exactly get points?? can someone explain this??

[dberke]

Bingo!...National points should be the true criteria for seeding and ranking before any tournament in any area of the country,period...

If the tournament organizers are doing a complete job, they'll take points into account. They don't have to, but it is an option for them if they are aware of it.

Dan

[occasionalfencer]

How do you exactly get points?? Can someone explain this??

Briefly, you have to make it at least to the top 32 of a national competition. It is more restricted for certain weapons. There are different kinds of national points- if you place in a Cadet competition, you get Cadet points. Talk to your coach about this, but if you are not going to National competitions, and winning some DE bouts, or meet managing, you don't need to know any of the details yet.

[fdad]

Briefly, you have to make it at least to the top 32 of a national competition. It is more restricted for certain weapons. There are different kinds of national points- if you place in a Cadet competition, you get Cadet points. Talk to your coach about this, but if you are not going to National competitions, and winning some DE bouts, or meet managing, you don't need to know any of the details yet.

IF points are to be used instead of ratings, then I think the points system needs some minor revisions. Rather than giving points to only top 32/40%, points should be given further down the table. Also, perhaps the ROC points can also be taken into account. Maybe sections and divisions can offer points earning events.

The benefits of points based system (with more points earning events) include: seeding based on multiple results (typical) rather than best ever result, finer granularity (not all As are equal), encouragement to fence in more events (most top women's foil fencers do not fence in events other than points earning events).

Obviously the system would need to be automated and integrated with FRED to avoid mistakes and overworking the national office.

[pillow]

If the tournament organizers are doing a complete job, they'll take points into account. They don't have to, but it is an option for them if they are aware of it.

Dan

I agree with Dan. The current system is fine. Combination of rating and points work fine for NACs. Local events do just as well. So there may be some inflation of ratings here and there--anyone think it has stopped the best fencers from rising to the top at NACs or SN? It hasn't.

[Sean Butler]

If anyone has ever played chess competitively they are probably familiar with the Elo rating system. My experience with it has shown that the regional inflation/deflation problems are not too bad. As long as there is a reasonable number of people from each area (proportional to its population) who compete in national or international tournaments, the ratings tend to correct themselves.

Has anyone ever proposed the use of an Elo system for fencing?

Searching revealed this 11 page thread: Fencing ratings

I understand that there are some problems with the Elo system in that people don't like to go backwards after a bad showing, and that this creates a negative incentive to actually participate when the odds are good that you will lose points. Maybe people are happy with inaccurate ratings so long as they can't lose them from playing?

[nahouw]

Has anyone ever proposed the use of an Elo system for fencing?

Yes. Geroge Masin, oh, about 12-13 years ago now...

[TooLoftheDeviL]

Somewhere in that mess is a lot of the information for the system I created as part of my undergraduate research for my senior project in mathematics.

I am confident an ELO inspired system would be 'better,' but how much better and it justification for the cost of a change would take considerable amounts of testing.

Notable 'incompletitions' in my system included fine-tuning an update coefficient for "better" wins (5-0 vs 5-4) and appropriate rating decay over time.

[Inquartata]

has anyone ever proposed the use of an elo system for fencing?

Aaaaaaaaaaaaaaaaaaaaaaaaaaarrrrrrrrrrrrrrrrrrrrrrr gggggggggggggggggggghhhhhhhhhhhh!!!!!

[KidLazy]

Elo, but calculating new rating using DE results only, updated after each round of DE.