Fencing ratings

[Phrogger]

Personnaly, my main problem with the rating system is the flukyness of it all...you got your A, you keep it. You could lose to 900 Cs, you could still have your A. Thats a big advantage of a ranking system like the UK has. Of course, you get the problem of fencers attending a whole bunch of comps getting way higher rankings than people who can't afford too then.

The biggest problem with the U.S. rating system may be that folks that don't get to attend lots of competitions (in more sparse geographic areas for example) won't get a rating very quickly if at all, while fencing-dense communities have ratings all over the place. But overall, I think the system works OK.

If I go to a competition and see that a guy is a "B05," I'm thinking, "he may be a B, but he hasn't re-earned his rating in two years, so I've got a shot at taking him out." If I fence a guy that's a "B07," now, I'm thinking, "this guy is in peak form, I'd better be cautious." A good fencer who is under-ranked won't stay that way, and a weak fencer who is over-ranked won't stay that way either. If you stay competitive, you will generally maintain your rank or improve, so while there is the occasional fluke, for the most part I think the ratings are fair.

High ratings may bring bragging rights, but surprising people is fun, too. I think a lot of fencers didn't see me coming when I showed up at the beginning of the year as a "U." I'm back up to a "D" after just one season back in the saddle. I think it's a fair rating. I feel I can eventually bring myself up to a "C" level if I can practice consistently, but earning a "C" will be a matter of attending competitions where I have a shot at it. Back in the early 1990's you'd be lucky if you could find even a "D" level comp in the whole state of Georgia. I had to go to nationals to get my original "C." With fencing apparently on the rise, there are lots more opportunities. I think a numeric system would dilute the significance of ratings and just make everything a lot more complicated.

[grotto]

seems to be alot of steam for little real result, in the end circumstance, and how well you fence will be the sole determination of your results. The ratings system now give a useable system to sort out pools etc for 90% of the tournaments out there. We have a usable national points system (which I really think should stay separate from the ratings) which sorts out real contenders in the national arena. I for one don't need to know I am 1,567th in the nation in epee.....

[downunder]

I may be missing something - how exactly is the British system different from our NACs?

How do your NAC's work?

[qatet]

How do your NAC's work?

Based off of the results from four national tournaments (or three, if it's an Olympic year, because then the Championships are moved from July to the weekend of the April NAC). Points for making the top 32. Obviously, more points for winning than for squeaking into the 32. Points expire in a year, so you'd better keep getting more!

Tournament format is frequently changed, but this year it's one round of pools, 75% up to DEs. Foil and epee have repechage from the 32, sabre doesn't.

[oiuyt]

Qatet- The BFA system gives points at all events. The scale of the points changes based on how strong the event is.

Relating that to a NAC, it would mean that Tucson could be worth less than Columbus. It means that Sectionals could also be worth points, although much less than either Tucson or Columbus. The little 12-person open is worth points, just very few of them, and higher-level local fencers will already have 6 higher-value results, so it likely won't count for them at all.

Go back up a few posts, follow my link, and read their description of their system.

-B

[qatet]

Go back up a few posts, follow my link, and read their description of their system.

-B

I did, but I didn't quite parse if it was awarding points at every event.

[Beloit Fencer of Old]

Anything that's not a NAC or a qualifier is a practice tournament.

Like World Cups?

[oiuyt]

I did, but I didn't quite parse if it was awarding points at every event.

I believe it awards points for everything on their National calendar. Which includes everything from largish events with all of the top people (NAC-equivalents) to little 20-person events (think local opens).

I seem to recall mention of purely regional events as well, but with how small some of the events on the national calendar are...

The same events draw the large/strong fields every year, in part because they're the events with large/strong fields (and therefore opportunity to earn points), in part because they're the events with large/strong fields (and therefore good tournaments for quality fencing). What it does allow for is more than just the top 32 to earn ANYTHING at those events, as well as the opportunity to substitute a very strong result at a mid-level tournament for a crap result at one of the biggies.

More of a spectrum of coverage from the points system, rather than just handling the top 50-or-so athletes.

-B

[downunder]

I believe it awards points for everything on their National calendar. Which includes everything from largish events with all of the top people (NAC-equivalents) to little 20-person events (think local opens).

Thats right. Every open event run to the BFA standards gets the same points awarded to it, then that is multiplied based on the strength of the competition. The multiple is calculated based on the number of top-50 fencers who attended the competition (who are divided into different point categories) or 25% of the fencers competing, whichever is higher.

I believe you have to place in both the top 64 and the top 75% to earn points from an event.

All the designated international competitions fit into the same system.


e.g. The winner of the British Championships Mens Epee recieved 2440 points, whereas the winner of the North West Ireland open (the lowest ranked open i could see) recieved 40.

[FoilyDeath]

Downunder has it pretty well explained. I prefer it to the US system frankly, but it also has its flaws : good fencers need to fence at least 6 comps to get their ranking up (such as international fencers having just arrived, etc) as oppposed to here where they can fence one good open, and it also means international fencers dont increase the "strength" of a comp, due to the fact that they aren't actually in the top 50.

[qatet]

The BFA system does sound like a good one. I'm absolutely against a system that discourages people from going to tournaments in any way. Doesn't look like the BFA has that problem.

ETA: Because, oh em gee, it's all about what I think is for the best.

[PeterGustafsson]

Hi!

I did, but I didn't quite parse if it was awarding points at every event.

I read the 8-page document. In brief:
1. Competitions are assigned strength factors (NIF values) based on who participates. For various competition types, there are special rules which impose bounds on the NIF values.
2. Fencers get to add the points from their 6 best competitions, but no more the 3 of those may be from events not organized under the British fencing organization. For those that have posted 6 results, their total points sum of the events that they have taken part in are extrapolated to what could have been expected if they would have competed in 6 competitions.
3. Each placement not worse than #64 gets a performance factor, which is multiplied with the NIF value in order to get the points value earned by the fencer at that placement at that competition. The shape of the placement vs.performance factor curve is constant for all competitions. For small competitions, less than 64 placements will be awarded points.
4. All competitions can award points.
5. If fencers with high ranking finish unexpectedly low in the final results list, this does not lower the rating of the competition (in contrast to the USFA system).
6. The organization maintains the tweak the system midseason at their discretion.


The extrapolation feature in #2 is one that I have not seen in any similar ranking system, and is a good idea in gauging the approximate strength of those who have not competed a lot. The curve shape in #3 is a characteristic that this system shares with most non-letter ranking systems. It is easy, but it fails to take into account that competitions vary in both top-level strength and depth of relatively strong field. Therefore, it introduces an unwanted padding sensitivity. #5 takes away some motivation for intra-club collusion. The USFA system is quite unique in having this characteristic, and other ranking systems seem to do just as well without it.

Overall, a better system than the USFA system. It is not sensitive to freakish luck/super-hot day, gives a feel for a fencers average strength without unduly rewarding competition every weekend all year around, good at resolving ties, integrates international events in a reasonable way, and is dead easy to calculate.


Have a nice time!

Peter Gustafsson

[HDG]

is dead easy to calculate.

Unfortunately, I can think of a number of division secretaries and local BC types who will disagree... Not an argument against change or against this model, but it will be an issue.

[contre-Sixte]

What am I missing? Here's a tournament that I attended last weekend. http://askfred.net/Results/results.p...df128348005ec2 Virtually every A finished at the top and every U finished at the bottom... across 3 weapons. And the B's ,C's,
D's, and E's lined up as one would have expected too. If ratings correlated any better, there would be no point of even having a competition. The current rating system does a great job of seeding, and it does a good enough job of letting one mark their own personal progress.

[nahouw]

What am I missing? .....
...... If ratings correlated any better, there would be no point of even having a competition. The current rating system does a great job of seeding, and it does a good enough job of letting one mark their own personal progress.

I think that people are looking for the ratings to mean something more than what they do -- they are just something that helps seed the pool round.

[mrbiggs]

The disadvantages of the current system are inflation, discrepancies across divisions in terms of skill, and the way that a fencer can have one good result and become very overrated.

All of these could probably be solved to a certain extent with tweaks to the current system, except for possibly the local discrepancies. However, if the Elo system were implemented, it would solve all of the problems forever in addition to making the system a bit easier to understand, if harder to calculate.


And though I'm generally a supporter of the Elo system, it can actually be fairly easily abused, especially in very isolated areas such as are common in fencing. For example, Claude Bloodgood spent several years in prison playing other inmmates in chess and winning every game. This effect would be seen in very weak areas with several strong fencers. So to be fair, the Elo system doesn't completely solve that problem either; it just makes the fencers overrated rather than underrated.

[oiuyt]

And though I'm generally a supporter of the Elo system, it can actually be fairly easily abused, especially in very isolated areas such as are common in fencing. For example, Claude Bloodgood spent several years in prison playing other inmmates in chess and winning every game.

I'm not sure Bloodgood's rating is an example of "easily abused". Abused? Yes, probably. Easily? Not so much. We're talking about many thousands of games played. Note that he was also successfully engaging in master-level correspondance chess at the same time.

http://www.chessville.com/misc/Histo...FBloodgood.htm

"Within any given month, on cellblock C of the Powhatan Correctional Facility, the VAPEN chess club (with TD Bloodgood pushing g4 on board 1) would finish more rated tourneys than were played on the whole eastern seaboard that year."

Takes some extreme scenarios to set up anything resembling that. Just make sure that National Team selection isn't based off of the Elo ratings and at best all one can accomplish by setting up such an abberrant environment is qualification to the highest level tournaments -- something that can already be done under the current system at least as easily (grab a bunch of newbies, fence them against each other to generate D's and then C's, grab a bunch more and eventually you're making A's and B's.).

-B

[mrbiggs]

I'm not sure Bloodgood's rating is an example of "easily abused". Abused? Yes, probably. Easily? Not so much. We're talking about many thousands of games played. Note that he was also successfully engaging in master-level correspondance chess at the same time.

http://www.chessville.com/misc/Histo...FBloodgood.htm

"Within any given month, on cellblock C of the Powhatan Correctional Facility, the VAPEN chess club (with TD Bloodgood pushing g4 on board 1) would finish more rated tourneys than were played on the whole eastern seaboard that year."

OK, not so easily.

That was an extreme example, but although it's unlikely that someone could abuse it to that extent in fencing, it could happen a little bit. In other words, no rating system is perfect because people can always set up tournaments to boost their rating, even if it's only by a letter or two or a few hundred rating points.

[CvilleFencer]

In other words, no rating system is perfect because people can always set up tournaments to boost their rating, even if it's only by a letter or two or a few hundred rating points.

I hate to point this out as it may be confused with support for the ELO or some similar system which I would probably abhor...

mrbiggs, I think that in the case you worry about, this comes down to the EC/Division Board and by extension the USFA, doing its job. If they do, it won't happen. In our division for example the EC sets the schedule and a tourney can only be sanctioned if the EC has an officer present (usually also reffing) to ensure things are above board. We also have other safeguards in place such as having X number of rated refs for events that charge Y and so forth. All of this is done so that as a division, we don't have to worry overly much about ratings inflation, at least at the low level, or about people "gaming" the system by holding RR's with "just enough" people and so forth. I have to think that most other divisions also have some type of mechanism in place, even if it differs.

So while I welcome any criticism that can be leveled against an ELO type system, I don't think this one needs to be worried about overly much as long as there is not cooperation by the Division. If that happens, IE you have a rouge Division or one that is simply incompetent, the USFA can step in to pull charters or disavow events (even seasons IIRC) that were "gamed" to inflate ratings.

[mrbiggs]

So while I welcome any criticism that can be leveled against an ELO type system, I don't think this one needs to be worried about overly much as long as there is not cooperation by the Division. If that happens, IE you have a rouge Division or one that is simply incompetent, the USFA can step in to pull charters or disavow events (even seasons IIRC) that were "gamed" to inflate ratings.

My point is that any rating systems have certain requirements that fencers have to meet in order to advance, and that any of these conditions can be set up artificially no matter the system.

It doesn't even have to be malicious; as I pointed out earlier, a fencer in an isolated area who goes to tournaments often and wins every one will probably be fairly overrated. He probably won't be rated second in the country as per the example I posted, but he will probably be overrated nonetheless.