Ratings - FIDE ELO, USCF ELO, BCF

Historically,

I can understand why different rating systems have come into existance.  But why are they still in existance today?

Wouldn't it be a good idea to unify them?  Is there any effort underway to do this?

Personally I'd be happy with standardisation on FIDE or USCF.  I for one couldn't tell you whether a BCF of 101 was good or not without resorting to some algorythm or other.  I think there's generally more awareness around FIDE and USCF and what those grades mean. 

And please, lets NOT start to discuss CHESS.COM ratings and how they compare to the real world - this has been covered in another posting.  I'm talking about OTB ratings in this case.

What do you think?

 Pretty good idea, it is confusing i am currently rated 1737 USCF but have no idea what that compares to When it comes to BCF ratings.

http://www.exeterchessclub.org.uk/bcftable.html

Begin by dropping BCF ratings entirely ... do you know how they pronounce "lieutenant" in that country? But slightly more seriously, it doesn't help. Ratings localize no matter what system you use, so ELO in Ottawa won't equal ELO in Prague, nor even in Vancouver.

there's a conversion from bcf to uscf and fide, i don't know the formula, plus i think there was an issue with deflation for bcf ratings

there are conversions from fide to uscf for pairing purposes, also if you play in a FIDE tournament under the US flag these conversions are used. i don't recall the formula used. generally uscf is 50 to 100 pts. higher than fide rating but i know of some bemusing exceptions (i know of a 900 uscf player with an 1800 fide rating).  these ratings really shouldn't be combined, they suit their purpose - they're for different pools of players. also uscf has rating floors, fide does not, so really old players sometimes get stuck at 2000 or 2200 whereas their fide drop a bit. uscf ratings are kind of weird at the low end too, with scholastic players often rated 100-1000.

actually i will get a fide rating in two weeks in an invite tourney, i have black against 5 players 2170+, if lightning strikes, its easy to see how inflated ratings are possible when establishing a rating.

If one more person uses the acronym BCF instead of ECF I'm going to go insane...

For quick reference:

USCF = ECF*8 + 600

FIDE = ECF*8 + 650.

Simple really.

Best of luck in your games Petrosianic - I came out of provisional fide at 1957 and my USCF is currently 1910. I guess my USCF is lagging my fide since I took a few years off and then started playing in Fide rated events.

The ECF grading database:

http://grading.bcfservices.org.uk/help.php#elo

It's not so much that the formulas are different, I'm sure if you took them mathematically, for example FIDE vs USCF would be very very little difference, lets say just a few points.  This is also why conversions work in the first place, because formulas are roughly equivalent.  So why are players in FIDE vs USCF in general 100 points different?

The difference is the pool of players, and that's why unifying ratings wont ever happen.  Each pool (i.e. nation) will keep theirs as well as having an international, FIDE, rating.  This is because not everyone wants to / or is able to play internationally to maintain an international pool which is the only way to "unify" ratings.

Now, something like changing the ECF to a 4 digit number so conversion is easier, I could see that maybe.  But unification doesn't have as much to do with formulas as it does player pools.

Isn't a FIDE rating the only international rating one can have? USCF and BCF ratings are national ratings. And last time I checked there were a lot more nations than just those two.

In a practical sense, you only have an international rating if you regularly play internationally. If you go to a tournament in Seattle or Vancouver and come out with a FIDE rating, don't fool yourself into thinking you have an international rating.

I agree that it can be practical to have national ratings that at least work approximately like the FIDE ELO rating. That way, many could see easily that a 1100 tends to be among the weaker players in his club, a 1900 may be among the strongest players in his club, and a 2700 tends to be among the best players worldwide.

A few decades ago, chess in Germany changed from INGO rating to DWZ rating for such a reason, DWZ being very similar to FIDE ELO, USCF rating, etc. .

british chess fedaration 

Hi All,

Sorry for the delay in responding... I stepped out for a moment ;-)

Yep - I'm aware of the formulas for doing the conversions between USCF, FIDE, ECF (happy Nytik?).

The question is why so many different rating systems are in use? It all seems a bit nationalistic/pointless.

Imagine if there were different national standards for measuring time... we'd never really be sure who could run the fastest over a certain distance. Hey, and whilst we're at it, let's invent different distance measurements too!

In truth, it's not a massive bug bear or anything like that. I just wondered if there was a valid reason for things being as they are.

Cheers!

Wow, huge posting gap

Part of it comes down to the fact that all rating algorithms measure things in slightly different ways and some organizations may think that their way is the more accurate one. Some of it may even come down to inertia and a dislike for change; so even if there is an objectively best rating formula, statistically speaking, change might not happen anyway.

Good luck getting organizations to all settle on the exact same algorithms and rating standards.

Oh Yeah. I haven't been in jail or anything like that 

Thanks Martin, that makes sense.

Once you have a rating in an Elo-based system it influences your expected scores. In other words your rating is based, to some extent, on every game you have ever played. So, for example, two players with different initial ratings can get exactly the same results against exactly the same players (which, to any sane observer, would mean that they are equally strong) yet in the Elo system the player with the higher starting rating will have the higher rating at the end.

There are many other flaws in the Elo system, most of which have been pointed out by Jeff Sonas. I think Elo himself understood perfectly the limitations of his system. The problem is that Elo-based ratings like FIDE ratings are used as a proxy for "chess strength" when The Elo system was actually designed to predict results. It does that quite well for players whose ratings are sufficiently close, but does increasingly poorly (over or underestimating) when ratings are far apart.

It was never designed to measure "chess strength".

The cynic in me thinks that the Elo system is fudged so that the ratings of IMs and GMs cannot drop too far too quickly. That would also explain why high rated players are given a smaller K-factor.

The old 3-digit BCF rating system discarded old results (with some caveats for players that played very few games). It therefore gave a more accurate assessment of a player's CURRENT strength than ratings based on the Elo system. In the old BCF system players with the same results would get the same rating or, at least, ratings that were very close.

Unfortunately the old BCF system was also fudged (hence flawed). I do not mean the usual problems of incorporating new players and retiring older ones that are faced by all rating pools. The problem is that to prevent the occasional loss to a much higher or lower rated player affecting the rating too much players that have a rating difference of more than 40 points were treated as if they were exactly 40 points apart.

I thnk that was a mistake. If a highly rated player carelessly blunders against a much lower rated player that is an aspect of his strength and ought to lose a lot of rating points.