PDA

View Full Version : Rapidly improving Adults...



The_Wise_Man
20-02-2006, 08:40 PM
We always hear about rapidly improving juniors...

What about the adults... been looking at the ACE website rating graphs to find rapidly improving adults...

Anyone know of people (non-juniors) that have improved greatly over the last 3 years?

Wise

Mischa
20-02-2006, 08:41 PM
We always hear about rapidly improving juniors...

What about the adults... been looking at the ACE website rating graphs to find rapidly improving adults...

Anyone know of people (non-juniors) that have improved greatly over the last 3 years?

Wise

Malcolm Pyke

The_Wise_Man
20-02-2006, 08:46 PM
Anthony Pickering 1307 in 03/03 to 1751 in 12/05 - a 444 point rise!

The_Wise_Man
20-02-2006, 09:00 PM
Whilst Malcolm has become quite a decent player at around 2100, his overall improvement is only 167!

Wise

Rincewind
20-02-2006, 09:04 PM
Whilst Malcolm has become quite a decent player at around 2100, his overall improvement is only 167!

Not wanting to detract from Anthony's meteoric rise which is impressive; not all rating points are equal. A 167 point rise at the pointy end of the list is at least as good as twice or three times that much in the middle.

Alan Shore
20-02-2006, 09:22 PM
Well, over 2 years ago I went approx. +400 from one tournament... think it was about 1440 - 1850.. but that's rapid. I improved normal rating about 4 yrs ago from 1350 - 1650 in a short space of time but haven't played much since then. And then I became just another ex-junior to be lost from chess.

And yeah, Pykey's improvement is worth more at that level, but not quite 3 times as much, lol.

Rincewind
20-02-2006, 09:32 PM
And yeah, Pykey's improvement is worth more at that level, but not quite 3 times as much, lol.

Do you have the statistical argument to back that up?

Alan Shore
20-02-2006, 09:42 PM
Do you have the statistical argument to back that up?

Sure, Number theory.

167 < 444.

:owned:

Mischa
20-02-2006, 09:44 PM
:)

Rincewind
20-02-2006, 09:46 PM
Sure, Number theory.

167 < 444.

:owned:

Since when was number theory a part of statistics? Most statisticians wouldn't even recognise a number line and none would care.

:D

Alan Shore
20-02-2006, 09:53 PM
Since when was number theory a part of statistics? Most statisticians wouldn't even recognise a number line and none would care.

:D

When you have an interval measurement system (of greater value than nominal or ordinal measurement) then I don't see how statisticians could 'not recognise' or 'not care' about it.

Mischa
20-02-2006, 09:54 PM
well there goes the fun here....it will turn into a maths lesson

Alan Shore
20-02-2006, 09:56 PM
well there goes the fun here....it will turn into a maths lesson

Just like you tuned out last night when Baz and I were talking maths, lol. ;)

Mischa
20-02-2006, 09:56 PM
do you blame me?

Kevin Bonham
21-02-2006, 02:05 AM
David Richards went up hundreds in a short space of time, although some of those were down to uplift and local adjustment.

Cameron Harris, a player at my club, jumped from about 1000 strength to about 1600 strength within a few months (this before he was rated), then didn't improve at all thereafter.

I know other cases of adults improving by a few hundred points but these involved inaccurate ratings to begin with. Can happen where a player returns and has a bad tournament first up.

Spiny Norman
21-02-2006, 08:04 AM
Anyone know of people (non-juniors) that have improved greatly over the last 3 years?
Moi (43 years young) +320 in 18 months ... from 1114?? (Sep 2004) to 1434! (Dec 2005). I qualify this by observing that I started that period with a ?? rating confidence level, so that has to be taken into account when considering speed of movement.

I now feel very comfortable playing against guys in the 1600-1800 bracket and no longer fear playing people ~2000. In recent weeks I have managed to beat a 1700+ and a 2100+ in non-rated games.

I aim this year to score 50% against people rated 1600-1800, consistently beat people rated under 1500, and to see if I can reach 1600 myself by December. I think my work is going to interfere though, as I am likely to have to miss playing some tournaments because of interstate travel.

Ratings are cool. Its me against 'the system' ... a private war ... makes competitive chess so much more meaningful to me than simply (for example) playing club tennis or something like that.

PHAT
21-02-2006, 08:55 AM
Anthony Pickering 1307 in 03/03 to 1751 in 12/05 - a 444 point rise!

AP is a returning player. As such, much of is rise comes from blowing out cobwebs. (Nice bloke though.)

jenni
21-02-2006, 04:44 PM
One to watch is Watto. She is current at around 1000, but I suspect is going to climb rapidly. (She does have a good a coach though...)

Sujakobi
23-02-2006, 02:27 PM
When I lived in Cleveland, Ohio there was a 40 something guy who walked into the local chess club there. Nobody paid much notice but three years later became an International Master. His first USCF rating was 1500 now it is about 2410 at least. He didn't even care too much about it. A very approachable guy.

bobby1972
23-02-2006, 02:31 PM
man thats a 1000 points thats almost mesianic .by the way see how topalov is playing he lost again then again Mexico is famous for its bad toilets:)

Rincewind
23-02-2006, 03:11 PM
When you have an interval measurement system (of greater value than nominal or ordinal measurement) then I don't see how statisticians could 'not recognise' or 'not care' about it.

Getting back to this you said...


Sure, Number theory.

167 < 444.

So you are saying that rating increases are directly comparable regardless of where they appear on the distribution curve. So an increase from (say) a rating of 200 to 400. Is more significant than an increase from 2201 to 2400. Since by your argument 199 < 200 and therefore is backed up by number theory.

Please confirm or clarify your position in this regard as I am genuinely interested in coming up with a way to compare the magnitude of rating differences which takes this into account.

Bereaved
26-02-2006, 01:11 AM
Quote:
Originally Posted by Belthasar
When you have an interval measurement system (of greater value than nominal or ordinal measurement) then I don't see how statisticians could 'not recognise' or 'not care' about it.

Getting back to this you said...


Quote:
Originally Posted by Belthasar
Sure, Number theory.

167 < 444.

So you are saying that rating increases are directly comparable regardless of where they appear on the distribution curve. So an increase from (say) a rating of 200 to 400. Is more significant than an increase from 2201 to 2400. Since by your argument 199 < 200 and therefore is backed up by number theory.

Please confirm or clarify your position in this regard as I am genuinely interested in coming up with a way to compare the magnitude of rating differences which takes this into account.

Hi everyone,

Surely a comparison of the numbers of people in the rating lists, active or otherwise, could show a distribution curve of the ratings and thus show perhaps how hard it is to change at various points in the pool?

Alternatively, one could examine how frequently players improved dramatically within certain rating ranges, or at all.

And in truth, having not much mathematical training, I will leave others to show how these things might be done, if they have any actual merit,

Take care and God Bless, Macavity

PS should number of years of play be a factor?? or non play? or ??, ?, , !, !! ?

Vlad
27-02-2006, 11:41 AM
Alternatively, one could examine how frequently players improved dramatically within certain rating ranges, or at all.


Yes, that is a very good suggestion. One can construct a database of changes for all rating groups. Then one would rank a particular increase; say 100 points, among all changes for players of 2200 strength and for players of 1000 strength and see the difference. For example, for players of 2200 strength 100 points increase could be at the 99-th percentile, while for players of 1000 strength it could be only at the 90-th percentile.

Edit: alternatively, it is easy to construct the following graph. For each rating group, say 1900-2000, one can find an average deviation in absolute terms. Then one can plot this average deviation as a function of rating. It will give a rough idea when it is harder to increase the rating.