K-Factor of 20

[Jonathan Bryant]

There's not much that any system can do with games it doesn't know about.

Since my first elo rating in 2011 I’ve lost about 150 rating points - at a time when my ECF grade has risen by about 5%.

This has been for three reasons:-

the systematic under-rating of juniors;
the consequent deflation of the ratings of people who play those juniors;
a huge (double figures) plus score against unrated opposition.





I’ve been saying for some time that - at the tournaments I play, at least - the elo rating system is simply broken. In the past this has been disputed. It seems more people are of a similar view these days.

[John McKenna]

Jonathan Bryant>... I’ve been saying for some time that - at the tournaments I play, at least - the elo rating system is simply broken. In the past this has been disputed...<

It's still in dispute then because, above, you pointed out yourself (and others have done so elsewhere) that the real problem -mainly a self-imposed English one - is due to the high number of unrated English players. English juniors are often underated for the same reason since the small pool of FIDE-rated players that most of them are likely to play at first often results in a very low initial rating.

The relatively slow rise of the FIDE-rated tournament in this country means that the situation will continue to persist until the vast majority of chess games played in England are FIDE-rated. The use of both the ECF-grading and FIDE-rating systems in this country is a clear indication of a halfway house situation that has been around far too long.

[MartinCarpenter]

No one is calling the system fundamentally broken, and especially not for what Elo originally designed it for It just got expanded to cover hugely more people and then tweaked to hopefully better match the changed requirements.

Junior initial ratings are often justifiably low but they then improve rather and it took a lot of games for them to catch up with that. Probably too many in principle, hence these changes.

The UK specific problems are another matter of course, and do mean FIDE grades below a given level really aren't worth anything. Elsewhere the biggest thing I can think of would be not rating evening league chess. Must make some difference. Of course, the ECF doesn't rate much evening league chess in Yorkshire

[Barry Sandercock]

If it happens that e2e4 congresses are now discontinued ( as seems likely), there will be considerably less FIDE rated games played in the UK.

[Jonathan Bryant]

No one is calling the system fundamentally broken ...

I am. At least for the tournaments in which I play. It’s almost completely useless there. It may be working fine elsewhere.

[Jonathan Bryant]

Jonathan Bryant>... I’ve been saying for some time that - at the tournaments I play, at least - the elo rating system is simply broken. In the past this has been disputed...<

It's still in dispute then because, above, you pointed out yourself (and others have done so elsewhere) that the real problem -mainly a self-imposed English one - is due to the high number of unrated English players.

I don’t believe that’s the 'real problem'. It’s part of the real problem, for sure. But only one part.

[Roger de Coverly]

I don’t believe that’s the 'real problem'. It’s part of the real problem, for sure. But only one part.

Everyone is presumably familiar with the 40 point rule in the ECF grading system. This has the effect of protecting higher graded players from the worst effects of playing much lower graded players. The converse side is that it acts as a brake on improving players reaching their true level. For juniors, the ECF effectively abolished it, as the hack of treating all juniors as new players can disregard earlier performances. FIDE though, has a 400 point rule, the effect of which, notwithstanding K=40 is that players with "old" ratings will emerge from the same performances with radically different ratings from new unrated players.

[NickFaulks]

I don’t believe that’s the 'real problem'.

So what is the real problem, and why doesn't it seem to apply elsewhere?

[Roger de Coverly]

and why doesn't it seem to apply elsewhere?

It's generally reckoned that the USCF has around five to six times the number of active players as the ECF. Yet if you list USA players below 1600, there are 85 of them as opposed to 193 ENG.

So the extension of ratings downwards has been faster in the UK than in at least one other country with a mature national rating or grading system. If you can compare rankings according to your domestic system with rankings according to the International system, you can determine which you think are more reliable.

Whilst "maintenance of quality" was a good reason not to allow games of under four hours into the standard ratings, why is it that games up to an hour can be rated and those over four hours, but not those in between?

It remains a question as to how prestigious is a FIDE rating.

According to the website
http://arena.myfide.net/

The chess world’s ultimate prize is a FIDE rating or title. Only the top 3% of chess players in the world get there, ranging from active club players to the current World Champion phenomenon,

A player with a rating of 1001 is ranked 186091, so if that's the top 3%, that estimates the total count of players to around 6 million. That's out by a factor of 100 to what some in FIDE seem to believe as being the world count of chess players.

[Paolo Casaschi]

I don’t believe that’s the 'real problem'.

So what is the real problem, and why doesn't it seem to apply elsewhere?

The real problem is obviously the stubbornness to maintain two sub-optimal ranking lists rather than having just one more accurate system.

[Jonathan Bryant]

I don’t believe that’s the 'real problem'.

So what is the real problem, and why doesn't it seem to apply elsewhere?

Well certainly not rating games against unrated players is a long-term drain on the system but the as I said earlier that’s just one facet of the problem. The under-rated rapidly improving junior is another.

I don’t know why this doesn’t apply else where (or even if it doesn’t) for the obvious reason that I don’t play 'elsewhere', but *perhaps* it’s more of an issue where I play because a larger proportion of juniors play in those tournaments. That’s just a guess, but the knock-on effect of those under-rated players deflating the ratings of other players would be more significant in that scenario.

The *real* problem - in my area, at least - is that the system has ceased to be self-correcting. Playing more games seems to enhance the deflating effect rather than reduce it.

As a result I think I have more chance of seeing 200ECF than I do getting back to my starting Elo of 2050 or thereabouts - a rating I achieved when my ECF grade was 172. Unless I play somewhere else.

[Paolo Casaschi]

Since my first elo rating in 2011 I’ve lost about 150 rating points - at a time when my ECF grade has risen by about 5%.

How many games did you get rated/graded for each system?

[MartinCarpenter]

I presume the problems with juniors are universal, or they wouldn't have made this sort of (universal?) change

Otherwise the real UK problem is that there's a load of people being given FIDE ratings based on only a few games a year at best. Can't expect any rating system to do well with that. That isn't truly a problem as such, just a slightly silly waste of time. (Below a given level those grades do absolutely nothing that ECF - with many more games - won't do better.).

[Jonathan Bryant]

Otherwise the real UK problem is that there's a load of people being given FIDE ratings based on only a few games a year at best.

This isn’t a UK a problem - in the sense that this doesn’t describe what’s going on in my local tournaments. My experience suggests the more active = the more deflated a rating is.

[Jonathan Bryant]

Since my first elo rating in 2011 I’ve lost about 150 rating points - at a time when my ECF grade has risen by about 5%.

How many games did you get rated/graded for each system?

I’d have to count back, but it would be a similar number.

[EDIT: I’ve counted. I got my first elo rating at Benasque in July 2011. Since then I’ve played about 140 rated/graded games. My quick count suggests 70+ games in elo rated tournaments - much more club chess than I imagined, but still a reasonable number of games to have played over three years. That’s played games, btw, not rated games. 20 or so of those didn’t count for rating - and guess how many of those 20 I lost]


One of the things that people who aren’t familiar with what’s actually going on in my local tournaments say is that you can’t expect games played in different circumstances (time controls etc) to end up with similar ratings.

True enough, perhaps, but this argument overlooks the fact that most of the games I play are graded AND rated. It’s not an either or situation.


I’ve more than once had a ECF grading performance for a tournament that exceeds my current grade and yet still lost Elo points from that same tournament, even though my Elo rating when you convert it is much lower than my ECF grade to start with.