PDA

View Full Version : ACF September 2007 Ratings - predictions



Rincewind
08-08-2007, 09:12 AM
People might like to calculate their rating adjustment using their favourite method and post it here and see how close we are.

I used my rating calculator (http://www.bjcox.com/modules.php?name=Glicko_Calc) and come up with...

1797 which is +22

This is based on 15 games and assumes the final game in my local club's grading swiss is completed and submitted for rating in time. I'm not planning to play any more games which might make it into this rating period.

Spiny Norman
08-08-2007, 09:57 AM
I have no real idea what's going to happen with mine. Main reason is that I had an unexpected loss to a new and unrated player during the period. I also lost a game against a higher-rated player where I missed a forced win. Other than that, my results were pretty good, so I expect (assuming a ?? for the new player) that I will probably lose 10-15 points and be around 1550

Garrett
08-08-2007, 11:13 AM
I think mine will go down, probably to about 1730, which will be a coincidence as that is my house number on Main Rd, Slumpsville.

Basil
08-08-2007, 11:22 AM
I predict everyone's rating will be an accurate reflection of their performance!

The beautiful game of football acknowledges that upsets and unfair results will happen - but the standings at the end of the season are a fair and accurate reflection of the strength of the play over the long haul.

Spiny Norman
08-08-2007, 11:26 AM
I predict everyone's rating will be an accurate reflection of their performance!
Pardon me gunner, but your slip is showing ... :P

Basil
08-08-2007, 11:39 AM
Pardon me gunner, but your slip is showing ... :P
That's actually very good indeed for a Victorian!

Ian Rout
08-08-2007, 12:07 PM
Based on the NSW Open (not so good) and the ANU Open (better), which I think are my only tournaments this time, the Barry calculator says 1868 (= +9) and the back of the envelope says 1869, so it's probably in that area. My gut feeling is it's more likely to be slightly lower.

CameronD
08-08-2007, 01:42 PM
Current rating 1274!

Played 12 games (W6 D3 L3)

True performance ratings
Flood Cup (5/7) 1537
Qld Reserve (2.5/5) 1267

Rating calculator 1342!! (+68)
expected score system 1339 (+65)

My rating always jumps up more than expected, so I'll guess at 1346 with 2 strong wins from 2 games owed for the December list against unrated and 1439 (WFM).

Kevin Bonham
08-08-2007, 03:26 PM
I used Barry's rating calculator and entered guess ratings with ??s for the three unrated players I played. I also tweaked the ratings of a couple of players up or down on the basis of very strong or weak performances in a tournament going in this period - strong or weak enough that the interim rating will significantly differ from their published rating.

For instance Kevin Hendrey (1295) has just scored 9.5/13 coming =2nd in the Burnie Club Champs despite being seeded 9th. Thus it's likely he will be treated as if rated quite a bit higher than 1295.

Anyway on the basis of all this my projected result is a loss of 19 points from 1984 to 1965.

The ratings calculator tends to be slightly less dynamic than the actual changes in my rating so I'll predict 1962. Think that drops me below firegoat again so I'll have some incentive to get my act together and stop playing so much rubbish.

Basil
08-08-2007, 08:07 PM
-- Winning Liddums
-- 2 x losses to Regina Edwards (1330) including 1 x whoppee cushion explosion

Yada yada yada ...

+ 30 to about 1515

Played 16: W:12 D:0 L:4

The two losses against Regina cost 29 points! and the loss to the Bxf7 thingo at Liddums cost another 14.

Look it's quite obvious my real strength is about 1900. Can you just slide that in please Bill? Thanks heaps.

And thanks again for the facility, Baz.

Axiom
08-08-2007, 08:33 PM
-- Winning Liddums
-- 2 x losses to Regina Edwards (1330) including 1 x whoppee cushion explosion

Yada yada yada ...

+ 30 to about 1515

Played 16: W:12 D:0 L:4

The two losses against Regina cost 29 points! and the loss to the Bxf7 thingo at Liddums cost another 14.

Look it's quite obvious my real strength is about 1900. Can you just slide that in please Bill? Thanks heaps.

And thanks again for the facility, Baz.:lol: look ! i'll only say,to be fair - if youre not 1700 in real strength right now ! ,then i'll eat my hat!

ER
08-08-2007, 09:44 PM
my turn for the slump horror camp!
I managed to get my rating to 1576, then went down to 1564! In this coming ratings I will be very happy to be included in the 1350 - 1400 bracket. No excuses whatsoever, but I think that, in my case, losing kgs has something to do with losing brain cells as well! Cheers, good luck and cagles!

Southpaw Jim
08-08-2007, 10:19 PM
Using Barry's calculator, I should probably go up about 100-150, depending on how unrated players are treated by the ACF. The lower end of this estimate range accords with my own Glicko approximation. Might even make the 'Most Improved' list! :)

Can Bill or someone explain how results against unrated players are treated? In our club champs we had 3 unrated players. One I know had a USCF rating of 2041, so I've used that as a proxy in the calculator, another I'm guessing is about 1500. The third, whom I beat, is substantially lower - I'd estimate between 1000-1200.

Is it like on FICS, where you get a provisional rating of 1500 and a really high RD? Or does ACF try and guess an individual initial rating?

WhiteElephant
09-08-2007, 12:18 AM
I think that, in my case, losing kgs has something to do with losing brain cells as well! Cheers, good luck and cagles!

Funny you should say that, hk, I find that when I put on weight I play better chess and when I lose some, I have more trouble concentrating. Weird....

Metro
09-08-2007, 01:32 AM
Funny you should say that, hk, I find that when I put on weight I play better chess and when I lose some, I have more trouble concentrating. Weird....

:hmm: More fuel in the tank?

Trent Parker
09-08-2007, 09:14 AM
I think I only played the NSW Open and the ANU open this period... Oh plus the grade matches. which i played poorly in both the ANU and the Grade matches.

Ok lets see. Here goeth nothing!

My rating after the NSW Open would be 1560 so its all backwards from there....

Hey Ya! Jeez! The Calculation after the ANU dropped me to 1505

Grade matches didnt do me any better down to 1456

slumpsville visit for me this period -90

Basil
09-08-2007, 09:24 AM
slumpsville visit for me this period -90
I have lived that pain. My thoughts are with you.

Trent Parker
09-08-2007, 09:32 AM
My mate Lance Chiddy went into ANU with a rating of 980 and has an expected rating coming out of 907. He finished on 3/7. However I feel tha his rating may not go that low. One of his games was against a rapidly improving junior: Anton Smirnov. This loss cost him 44 ratings points according to Barry's Calculator

Spiny Norman
09-08-2007, 11:20 AM
Hey Barry, has your calculator been updated with the v2.1 (v2.2?) latest mods to Glicko2 as implemented by Bill et al fairly recently? If not, then some of our calculations might conceivably be a bit dodgy.

Phil Bourke
09-08-2007, 11:32 AM
I fell in to a burning ring of fire
I went down,down,down
and the flames went higher.
And it burns,burns,burns
the ring of fire
the ring of fire.

Watto
09-08-2007, 11:49 AM
After picking up about 60 points from the City of Melbourne :) (one round a week, nice long time limit), I had two shocking weekenders (Lidums and Geelong) where my clock skills were shown to be non-existent and my tactics unravelled promptly lost about 90 points. So my new rating will be about 1155 is my guess. See you in Slumpsville TCN...

Rincewind
09-08-2007, 04:30 PM
Hey Barry, has your calculator been updated with the v2.1 (v2.2?) latest mods to Glicko2 as implemented by Bill et al fairly recently? If not, then some of our calculations might conceivably be a bit dodgy.


The short answer is no the modifications have not been made and so yes, that does impact on predictability of the estimator. However ther reasons for this vary. For the long answer, read on...


Firstly, the RD I would like to use in the actual RD value stored in the ratings master file. However, this is not published. All that is published is the RD indicator (!!, !, etc) and so from the outset the calculator has limits on its accuracy based on the rating information which is publicly available.

Secondly, the ratings estimator is a Glicko 1 implementation. The reason for this is that the major difference between G1 and G2 is the introduction of the volatility. However, the ACF doesn't even publish a volatility indicator and therefore the volatility is basically an imponderable for those of us on the outside of the ACF master file and therefore it was not worth my while writing a Glicko 2 implementation (which I could do easily, and have done so in an Excel spreadsheet) without that information.

Finally, some of the subsequent changes you allude to in your message is a tweeking of the system which means to perform a totally accurate calculation I would need to know all the results of not only you, but also all of those of all your opponents. Basically I would need all results as they occurred in real-time. This is practically unimplementable.

So the calculator remains as is stands today. A handy tool to perform rating estimates but not 100% accurate. Exercises such that that in this thread help to provide an idea as to how good it is at predicting rating adjustments. Of course people tweaking ratings and so forth, means the results are an estimate of the whole process including their tweaking, but it is still a worthwhile exercise.

There are a couple of scenarios which would cause me to update the calculator. If the ACF was to publish the RD number instead of RD indicator, then I would probably leave the page as a Glicko-1 implementation but add the code to the database so that more accurate results are (hopefully) achievable.

If the volatility factor, or even a volatility indicator, was published. I would probably implement a Glicko-2 page to make use of this information.

Ideally, both the numerical RD and volatility would be published and then a better calculator could be written which I feel must be more accurate than what is possible without this data. However, the latest changes to the rating calculations could not be implemented in a predictive estimator as the actual rating adjustment will actually depend on all the results of a rating period (including all game of your opponents) and therefore quite likely depend on games not yet played at the time the estimate calculation is performed.

I hope this answers your question, albeit in probably a longer form than you may have been expecting.

Desmond
09-08-2007, 10:01 PM
I think my rating will go up about 20 points or so.

MichaelBaron
09-08-2007, 11:58 PM
My mate Lance Chiddy went into ANU with a rating of 980 and has an expected rating coming out of 907. He finished on 3/7. However I feel tha his rating may not go that low. One of his games was against a rapidly improving junior: Anton Smirnov. This loss cost him 44 ratings points according to Barry's Calculator

This is simply funny :). 900 rated players should worry more about learning to play chess than rating. Guys, rating is not everything! If you play well, your rating will gradually go up! Does it matter if you are 907 or 980? What matters is that every tournament should be a stepping stone towards becoming a better player.

MichaelBaron
10-08-2007, 12:17 AM
If somebody wants rating points. I am happy to play that person in an ACF rated match. I am happy to lose every game. However, for every rating point that i donate..i would like $1 to be donated to Worldvision.

This sounds like a good arrangement to me. People in need will get some money and idiots who think that high published rating makes them chess players can get what they want.


It is obsession with ratings and titles that has led to GM and IM titles being bought and sold freely. But what is the point? Look at GM Grechikhin! the guy is taking place in an open event in Sukhum in Georgia and after 4 rounds he was on 0.5/4 against the field with an average rating of 2000 or so.

To me, if someone is a strong player, he does not need a high rating or a title
Does Hamilton need an FM title? I do not think so. Everyone knows he was and still is a real master! winning strong weekenders at that age of 67 :clap:

at one point Hacche's ACF rating was about 2035 only. Big deal! Everyone knows he is a fairly strong player capable of performing well in Master events ireespectively of his rating. At the same time there are other players who never played in Masters events and are "club players" yet their ACF rating is higher.

Anyway, the offer is here :). If you are rated 900-something and want to get to 2300 - please message me and we will find a suitable arrangement that will keep everyone (importantly for me - worldvision in particular) happy!

Bill Gletsos
10-08-2007, 12:24 AM
If somebody wants rating points. I am happy to play that person in an ACF rated match. I am happy to lose every game. However, for every rating point that i donate..i would like $1 to be donated to Worldvision.

This sounds like a good arrangement to me. People in need will get some money and idiots who think that high published rating makes them chess players can get what they want.


It is obsession with ratings and titles that has led to GM and IM titles being bought and sold freely. But what is the point? Look at GM Grechikhin! the guy is taking place in an open event in Sukhum in Georgia and after 4 rounds he was on 0.5/4 against the field with an average rating of 2000 or so.

To me, if someone is a strong player, he does not need a high rating or a title
Does Hamilton need an FM title? I do not think so. Everyone knows he was and still is a real master! winning strong weekenders at that age of 67 :clap:

at one point Hacche's ACF rating was about 2035 only. Big deal! Everyone knows he is a fairly strong player capable of performing well in Master events ireespectively of his rating. At the same time there are other players who never played in Masters events and are "club players" yet their ACF rating is higher.

Anyway, the offer is here :). If you are rated 900-something and want to get to 2300 - please message me and we will find a suitable arrangement that will keep everyone (importantly for me - worldvision in particular) happy!Knock yourself out, however such matches are not rated. :hand:

Watto
10-08-2007, 08:23 AM
Guys, rating is not everything! If you play well, your rating will gradually go up! Does it matter if you are 907 or 980? What matters is that every tournament should be a stepping stone towards becoming a better player.
Yep, I completely agree with this. The rating calculators are quite fun though.

Basil
10-08-2007, 12:02 PM
Oh dear ... bites tongue. Leaves Miko and Jeano under impression they are saying something wise ...

Watto
10-08-2007, 03:08 PM
:rolleyes: can we add supercilious git to the bio please?

Basil
10-08-2007, 03:50 PM
:rolleyes: can we add supercilious git to the bio please?
Giddy up!

Trent Parker
12-08-2007, 03:13 PM
This is simply funny :). 900 rated players should worry more about learning to play chess than rating. Guys, rating is not everything! If you play well, your rating will gradually go up! Does it matter if you are 907 or 980? What matters is that every tournament should be a stepping stone towards becoming a better player.

Hey Michael. I was just doing this calculation for another example of the glicko calc. My friend does not worry about his rating. He enjoys playing chess and thats all that matters to him. I was curious about where his rating would go so I did the calculation.

Bill Gletsos
19-08-2007, 09:43 PM
Based on the NSW Open (not so good) and the ANU Open (better), which I think are my only tournaments this time, the Barry calculator says 1868 (= +9) and the back of the envelope says 1869, so it's probably in that area. My gut feeling is it's more likely to be slightly lower.Currently your gut feel is correct. ;)

Bill Gletsos
19-08-2007, 09:44 PM
People might like to calculate their rating adjustment using their favourite method and post it here and see how close we are.

I used my rating calculator (http://www.bjcox.com/modules.php?name=Glicko_Calc) and come up with...

1797 which is +22

This is based on 15 games and assumes the final game in my local club's grading swiss is completed and submitted for rating in time. I'm not planning to play any more games which might make it into this rating period.Currently your estimate is too low. ;)

Bill Gletsos
19-08-2007, 09:48 PM
I think mine will go down, probably to about 1730, which will be a coincidence as that is my house number on Main Rd, Slumpsville.Currently you are heading further down the street. ;)

Garrett
21-08-2007, 06:57 AM
Currently you are heading further down the street. ;)

Thanks Bill

If anyone want's to visit me on Main Rd Slumpsville then please drop in.

My place is the one thats white, about 4 foot x 4 foot by 8 feet high, and has 'Coates Hire' on the door.