PDA

View Full Version : Barry's Glicko Calculators (was Question For Barry Cox)



bobby1972
23-01-2006, 02:08 PM
When Are You Going To Update Your Calculator With The Latest List?this Site I Belive Is One Of The Important Ones That Players Look At

Rincewind
23-01-2006, 02:09 PM
When Are You Going To Update Your Calculator With The Latest List?this Site I Belive Is One Of The Important Ones That Players Look At

Yes, I usually wait for someone to ask. Sorry, have been a little busy but not so busy as I can;t find 10 minutes to do this job. I'll try to address it tonight.

bobby1972
23-01-2006, 03:52 PM
thats great i for one thank you.i believe that every one is interested in this site its just that they dont know about it ,maybe the should be a link to it within this forum ,also during a tournament its good to know how you a going so you know how hard to go in the rest of the games

PHAT
23-01-2006, 05:12 PM
One fly in the ointment is that inspite of its good design, the calculator has some inaccuracy. c ompared to what Glicko 2 spits out. It stems from the fact that Bill Gletsos refuses to give anyone the super secret RD values and the excellerater sub routine. Can anyone say exactly why The Dog won't give out this info? No. Sounds fishy? Yep, sounds fishy. I wonder if there is fiddling going on. We shall never know because The Dog keeps it all secret. Power, secrets, lies by ommission, backstabbing - all in a days work.

Rincewind
23-01-2006, 06:06 PM
thats great i for one thank you.i believe that every one is interested in this site its just that they dont know about it ,maybe the should be a link to it within this forum ,also during a tournament its good to know how you a going so you know how hard to go in the rest of the games

Now updated. Enjoy!

http://www.bjcox.com/modules.php?name=Glicko_Calc

EGOR
24-01-2006, 12:20 PM
Now updated. Enjoy!

http://www.bjcox.com/modules.php?name=Glicko_Calc
I'm one person who didn't know about this, and I am very glad to.:D
Thanks.

Altecman
25-01-2006, 08:33 AM
Sweet, i thought my rating would die after the Australian Juniors :(

But i got all my rating points back in the Australia Day weekender :owned:

lol my rating stayed the same after 30 games :P

jenni
25-01-2006, 11:26 AM
Sweet, i thought my rating would die after the Australian Juniors :(

But i got all my rating points back in the Australia Day weekender :owned:

lol my rating stayed the same after 30 games :P

Playing the Aus Juniors tends to have a bad effect on ratings!! Fortunately a few adult tournaments usually restores the status quo! A few more will see you go up - however you might want to avoid the Gold Coast open.....

Garvinator
25-01-2006, 02:31 PM
Playing the Aus Juniors tends to have a bad effect on ratings!! Fortunately a few adult tournaments usually restores the status quo! A few more will see you go up - however you might want to avoid the Gold Coast open.....
and even worse, The Gold Coast Tin Cup ;)

Garvinator
14-03-2006, 12:08 AM
is your site down again Barry?

Rincewind
14-03-2006, 06:19 AM
is your site down again Barry?

Still.

I upgraded some hardware on my home network the weekend before last and haven't reconfigured Apache on my web/mail machine. Was hoping to on the weekend but it didn't happen. Sorry.

Garvinator
16-03-2006, 03:15 AM
Still.

I upgraded some hardware on my home network the weekend before last and haven't reconfigured Apache on my web/mail machine. Was hoping to on the weekend but it didn't happen. Sorry.
when is it likely to happen?

Rincewind
16-03-2006, 09:36 AM
when is it likely to happen?

Sometime this year.

Rincewind
25-03-2006, 02:01 PM
Well http://www.bjcox.com/modules.php?name=Glicko_Calc is working again. Now I just need to get the ratings from March loaded...

bobby1972
27-03-2006, 10:57 AM
thanks for the site ,come on people you all use it its the only site in town after a turny .thanks again barry

Rincewind
27-03-2006, 11:24 AM
thanks for the site ,come on people you all use it its the only site in town after a turny .thanks again barry

np. Also note that the ratings have been updated to March '06.

bobby1972
19-04-2006, 09:50 AM
will the site be up ?.i think after canberra many people will like to use it ,thanks

Rincewind
19-04-2006, 10:01 AM
will the site be up ?.i think after canberra many people will like to use it ,thanks

Sorry dude. It is broken and bad. It will take some fixing. No ETA.

antichrist
20-04-2006, 08:08 AM
.................................................. ..........................

Garvinator
07-05-2006, 12:22 PM
Hey Barry,

When will your site be making a return?

Rincewind
07-05-2006, 05:53 PM
Hey Barry,

When will your site be making a return?

Sorry, not a priority at the moment. In fact I haven't even been reading the board this weekend, just too busy.

But one day...

(Actually, further to this there is a good chance I will change ISP this month end. Perhaps that will spark of enough interest ni IT related matters to get it working again).

Garvinator
07-05-2006, 06:37 PM
Sorry, not a priority at the moment. In fact I haven't even been reading the board this weekend, just too busy.
ohhhh :(

Garvinator
11-07-2006, 07:10 PM
But one day...

(Actually, further to this there is a good chance I will change ISP this month end. Perhaps that will spark of enough interest ni IT related matters to get it working again).
and how is the reinvention of your site going Baz?

Rincewind
11-07-2006, 08:07 PM
and how is the reinvention of your site going Baz?

Still pressed for time. I did change ISP but it did not spur me on reinvent the site. And I spent all last week in the sunshine state and I'm still playing catchup.

I'll let you know.

Rincewind
10-09-2006, 03:09 PM
OK. I have a site of sorts back up.

http://www.bjcox.com/

The only thing of note is the Glicko_Calc module accessible from the menu of the left hand side.

September 2006 ratings are loaded in the database.

Let me know here if there are any problems.

Garvinator
10-09-2006, 03:33 PM
:clap: :clap: :clap:

Basil
01-03-2007, 12:56 AM
This is my first time to this thread. Congrats, Barry. I like the site. Do you wish for feedback relating to same using the feedback functions on the site? (Genuine question - not sure if you check them, as the site seems to have had a re-incarnation or 3 ;))

Rincewind
02-03-2007, 12:48 AM
This is my first time to this thread. Congrats, Barry. I like the site. Do you wish for feedback relating to same using the feedback functions on the site? (Genuine question - not sure if you check them, as the site seems to have had a re-incarnation or 3 ;))

Yeah, I think it works still. Try it and I'll reply if I see it. :)

I shoul put the March 2007 ratnigs up sometime. Need a spare moment sometime..

Rincewind
02-04-2007, 01:02 AM
The site was down for a few weeks. It is back up now and the March 2007 ratings have been loaded.

Edit: Also the Dec'06 and Mar'07 ratings lists have been added to the downloads area.

Vlad
02-05-2008, 03:30 PM
I was playing with Barry's calculator and noticed some inconsistency.
Let us say a player with a rating of X (with ! next to it) played 30 games against 1400 rated opponents (all have ! next to their ratings) and drew all 30 games (or lost 15 and won 15). Surprisingly the resulted rating as a function of X is not a monotonic function. Specifically, let us say f(X) is the resulted rating after playing 30 games. The following numbers are given by the calculator:
f(800)=1397,
f(900)=1389,
f(1000)=1369,
f(1100)=1354.
That means for ratings 800-1100 the function is monotonic but the other way around. The higher is your original rating, the lower is your final rating. :wall:

Next I changed ! for the player X to be an empty space and I got
f(800)=1765.
So we have 800 player who drew 30 times a 1400 player and as a result becomes 1700 player.:clap:

Kevin Bonham
02-05-2008, 04:02 PM
That means for ratings 800-1100 the function is monotonic but the other way around. The higher is your original rating, the lower is your final rating. :wall:

That seems to bottom out at an initial rating of about 1150 and then pick up again for higher initial ratings.

I don't know if this can also happen under the current Glicko-2 but I think there is actually some kind of argument for it even if it can. That is that the greater the disparity between the original rating and the observed performance, the stronger the statistical evidence becomes that the original rating is simply total junk, as opposed to (say) a player with erratic runs of form having a good few months.

However an 1100! who gets 50% from 30 games vs 1400!s should go closer to 1400 than 1354. I expect that under G2 which is more dynamic they would go closer (maybe 1375 or so).


Next I changed ! for the player X to be an empty space and I got
f(800)=1765.
So we have 800 player who drew 30 times a 1400 player and as a result becomes 1700 player.:clap:

I had an overshoot like this under the old ELO system where because of a backlog of dozens of tournaments being processed at once in a period in which I had improved, I went to a rating that was about 60 points higher than my PR for the tournaments. However I was able to maintain that rating from that point.

I think overshoot/undershoot protection should be applied in such cases generally. Not sure if it is at the moment or not. Perhaps there are some cases though, with fast improving juniors, where an overshoot could be more predictive of future results than the player's actual performance rating for the period.

Bill Gletsos
02-05-2008, 04:28 PM
I think overshoot/undershoot protection should be applied in such cases generally. Not sure if it is at the moment or not. No players rating can change such that their new rating for the period exceeds or drops below their performance rating for the period.

Vlad
02-05-2008, 04:34 PM
Sorry Bill, are you saying that Barry's calculator has an error?

Rincewind
02-05-2008, 04:46 PM
Sorry Bill, are you saying that Barry's calculator has an error?

My calculator does a straight Glicko-1 calculation and (as far as I am aware) works as per its design. The ACF rating system is Glicko-2 with a number of tweaks and so my calculator is only an estimate of what might happen.

Kevin Bonham
02-05-2008, 04:54 PM
No players rating can change such that their new rating for the period exceeds or drops below their performance rating for the period.

That is the sort of protection I had in mind.

Bill Gletsos
02-05-2008, 05:01 PM
Sorry Bill, are you saying that Barry's calculator has an error?Not at all. I'm just pointing out the ACF system has additional features to the standard Glicko system.
Barry's calculator does what standard Glicko-1 does. However even back when the ACF was running Glicko-1 and not Glicko-2 the ACF system the overshoot/undershoot protection in place.

Vlad
02-05-2008, 06:55 PM
Not at all. I'm just pointing out the ACF system has additional features to the standard Glicko system.
Barry's calculator does what standard Glicko-1 does. However even back when the ACF was running Glicko-1 and not Glicko-2 the ACF system the overshoot/undershoot protection in place.

Thanks for the reply. Regarding the non-monotonicity, is it different for Glicko-2?

The way I thought the system is working is that your new rating is a weighted sum of your current rating (your performances in the past) and your performance in the current period. If your current rating is higher then the new rating has to be higher as well. I do not buy the argument that if you can perform significantly better than your current rating then your former performances should not be taken into account.

Bill Gletsos
02-05-2008, 07:12 PM
Thanks for the reply. Regarding the non-monotonicity, is it different for Glicko-2?I believe when I looked at it ages years ago that it was similar to Glicko-1. BTW we also do the overshoot/undershoot protection in the current ACF Glicko-2 system.

The way I thought the system is working is that your new rating is a weighted sum of your current rating (your performances in the past) and your performance in the current period. If your current rating is higher then the new rating has to be higher as well. I do not buy the argument that if you can perform significantly better than your current rating then your former performances should not be taken into account.If your performance rating in the period is higher than your current rating then your new rating should be higher than your current rating and if your performance rating in the period is lower than your current rating then your new rating should be lower than your current rating.

Vlad
02-05-2008, 07:17 PM
If your performance rating in the period is higher than your current rating then your new rating should be higher than your current rating and if your performance rating in the period is lower than your current rating then your new rating should be lower than your current rating.

Sorry Bill, by non-monotonicity I mean the fact that
f(800)=1397,
f(900)=1389,
f(1000)=1369,
f(1100)=1354.
Your higher current rating results in lower new rating.

Bill Gletsos
02-05-2008, 07:25 PM
Sorry Bill, by non-monotonicity I mean the fact that
f(800)=1397,
f(900)=1389,
f(1000)=1369,
f(1100)=1354.
Your higher current rating results in lower new rating.Only to a certain point.

Using the settings you described in your original post then it stops getting lower at 1148 and starts increasing again.

Using Barry's Calculator you get

f(1148) = 1352
f(1200) = 1354
f(1300) = 1372
f(1350) = 1385
f(1400) = 1400

Vlad
02-05-2008, 07:46 PM
Thanks Bill. Yes, I understand what it is doing. I am just saying that I do not think that it makes sense.
f(800)=1397,
f(1100)=1354,
f(1390)=1397.

Why is 800 player able to jump as high as 1390 player after exactly the same performance?

Rincewind
11-07-2008, 02:18 PM
June 2008 ratings have been loaded into the calculator (http://www.bjcox.com/modules.php?name=Glicko_Calc).

We now return you to your normal programme.

Rincewind
09-03-2009, 07:25 PM
After several version updates of mail and web servers as well as a couple of releases upgrade on the CMS my website is hopefully again operational.

http://www.bjcox.com/modules.php?name=Glicko_Calc

Should still work. If you have time try it out and let me know if there are issues. (PS I know the ratings are still the December 2008 ones).

Rincewind
14-02-2010, 10:15 PM
I migrated the webserver to new hardware today. Seems to work but if you notice anything funny happening, let me know.

Rincewind
20-06-2010, 06:45 PM
The website (http://www.bjcox.com/) will be going offline Wednesday 23 June, 2010 and will not be coming online again until some time in July (hopefully before the 15th). In the interim I will be moving the last few rating lists and some software downloads onto a temporary site in Google Sites.

To see what the temporary site will look like go to the temporary website (http://sites.google.com/site/bjcoxtemp/).

Rincewind
09-07-2010, 06:57 PM
OK. The new internet service has been provisioned and http://www.bjcox.com/ is back on the air. Sorry for any convenience during the downtime but these things take time.

Tony Dowden
25-08-2010, 08:28 PM
Thanks :)

But a pity it doesn't turn Glicko into Elo ;)

schess
14-03-2011, 10:45 AM
Can you please update the website and put the 2011 March ACF ratings

Rincewind
14-03-2011, 03:05 PM
Can you please update the website and put the 2011 March ACF ratings

Hi the database has been updated.

schess
02-06-2011, 05:43 PM
Hi
Can you please put 2011 june rating on barrys new site?
Thanks

Rincewind
02-06-2011, 06:13 PM
Hi. The new ratings will get there soon. I usually wait a little while as people may like to check their adjustment based on the ratings from last period. After a week or so I try to change the site for people who might want to check results from the new rating period (the one we are in now).

I've been thinking about adding a selection where people can choose which rating period to use. But it hasn't moved beyond the concept stage at present.

schess
02-06-2011, 06:38 PM
Hi,
Thanks!
Do the ACF rapid rating also work with barrys new web site?

Rincewind
02-06-2011, 09:36 PM
Hi,
Thanks!
Do the ACF rapid rating also work with barrys new web site?

They are available for download from the appropriate section there but they are not available from the Glicko_Calc module. Something that could be added along with therating period selector mentioned above.

Rincewind
05-06-2011, 05:01 PM
Hi
Can you please put 2011 june rating on barrys new site?
Thanks

The June ratings have been added (see this news item (http://www.bjcox.com/modules.php?name=News&file=article&sid=14)).

Kaitlin
05-06-2011, 06:45 PM
You have a News site :eh:

Rincewind
05-06-2011, 07:51 PM
You have a News site :eh:

Haven't you subscribed to the RSS feed yet?

schess
19-06-2011, 05:27 PM
Hi Barry,
Does Barry's Web Site work with Rapid Acf Ratings or do Rapid Ratings have a different formula?

Rincewind
19-06-2011, 07:39 PM
Hi Barry,
Does Barry's Web Site work with Rapid Acf Ratings or do Rapid Ratings have a different formula?

The Glicko calculator on my website does a Glicko-1 calculation which is a reasonable estimate to the ACF system for most people however it does not implement Glicko-2 (which is closer to the ACF system) mainly because the ACF does not publish precise RDs or any volatilities.

So that being said you can use the calculator and it should give you a reasonable estimate of rating changes. The name look up fills in the current ACF standard rating of players and their opponents from a database. But if you type in ratings and RD indicators from the Rapid list then you should have no problem estimating rapid rating changes. I believe this is also a Glicko-2 system at the ACF level but the Glicko-1 calculator should still be ok provided the system parameters used by the ACF are the same.

schess
19-06-2011, 08:01 PM
I used Barry web site last rating period i played 9 games and my rating that was determined by the calculator was 10 to 15 rating points off from my real june rating why is that?

Kevin Bonham
19-06-2011, 08:23 PM
I used Barry web site last rating period i played 9 games and my rating that was determined by the calculator was 10 to 15 rating points off from my real june rating why is that?

As Barry explains in his post above the calculator only provides an estimate because it uses a slightly different system. Differences of 10-15 points are common for that reason.

Rincewind
19-06-2011, 08:58 PM
I used Barry web site last rating period i played 9 games and my rating that was determined by the calculator was 10 to 15 rating points off from my real june rating why is that?

Over nine games 10-15 point is only 1-1.5 points per game so not that bad. The way the ACF actually calculates ratings is based on Glicko-2 which has a concept of a volatility which does not exist in Glicko-1. Also the ACF uses intermediate ratings which an online calculator of this type cannot reproduce.

schess
10-09-2011, 10:08 PM
Hi Barry,
Can you please put september ratings on your website

Rincewind
10-09-2011, 10:48 PM
Hi Barry,
Can you please put september ratings on your website

All done. The Glicko_Calc application updated and the lists added to the download section of the site.

Tony Dowden
11-11-2011, 07:47 AM
The 'search' boxes for players won't close which means the calc freezes and won't proceed. (I had the same problem several days apart - but still unsure if it's just my computer or not)

Rincewind
11-11-2011, 07:53 AM
I haven't seen that problem here, using Firefox on MacOSX. I check by searching on my name and your name and both worked fine. I take it the problem is happen consistently for all searches you do?

What OS and browser are you using?

kibitzer
01-12-2011, 11:47 AM
As a matter of interest, does anyone know how accurate Barry's Glicko calculator is?
His site:
http://www.bjcox.com/modules.php?name=Glicko_Calc

Kevin Bonham
01-12-2011, 11:50 AM
As a matter of interest, does anyone know how accurate Barry's Glicko calculator is?

It uses Glicko-1 so it is somewhat different to the ACF's version of Glicko-2. In most cases the differences will be small (I have often found it to be within 10 points) but when players have large ratings changes, especially if they are inactive or juniors, there will be substantial differences. (So for instance if a player is shown as going down by 300 points, don't be surprised if it's really 240 instead.)

kibitzer
01-12-2011, 12:40 PM
Thanks Kevin! I don't suppose there's a Glicko-2 calculator online then? :)

Rincewind
01-12-2011, 01:36 PM
Thanks Kevin! I don't suppose there's a Glicko-2 calculator online then? :)

A glicko-2 calculation is reasonably easy to implement in (say) excel. I had one which I have emailed to some people on request. However I think the original copy was lost in a recent hard drive problem I experienced so if you're interested I might need to trawl my sent mail to find it or ask someone who I sent it to. (From memory I think I sent it to Vlad and possibly Bill).

I thought I had it on my webserver, but I am mistaken that is a glicko-1 excel file.

kibitzer
01-12-2011, 01:48 PM
Yes thanks Rincewind that would be great!

Kevin Bonham
01-12-2011, 09:49 PM
Note that while a G2 calculator will be more accurate than a G1 calculator, it still won't produce an exact rating in advance.

Rincewind
02-12-2011, 09:11 PM
I've decided to make the glicko2 calculator available on the website. You can download it here

Barry's Web Site - Downloads (http://www.bjcox.com/modules.php?name=Downloads&d_op=viewdownload&cid=12)

Note that the spreadsheet has a macro that runs behind the scenes. You need t ensure your Macro Security level is set to Low for this macro to be allowed to run on your computer.

Don't worry it's not a virus. It is a process in glicko2 calculation which you need to perform a root-finding process. From memory I implement this using the bisection method (http://en.wikipedia.org/wiki/Bisection_method).

kibitzer
02-12-2011, 09:30 PM
Thanks Rincewind! Where do you get these RD's though and how important are they in the final result?

Rincewind
02-12-2011, 09:50 PM
Thanks Rincewind! Where do you get these RD's though and how important are they in the final result?

You can use an indicative number depending on the !!, !, space, ? and ?? that is published with the rating. I'd use 60, 95, 140, 205 and 300, respectively. As for the sensitivity, it's probably best to have a fiddle and work that out for different scenarios yourself.

kibitzer
02-12-2011, 10:13 PM
It seems to be giving fairly accurate results, without working out all my opp's RD's yet - by sensitivity do you mean that volatility figure?
Thanks again!

Kevin Bonham
02-12-2011, 11:01 PM
Also do you know what figure you use for the rating of an unrated player please?

When I'm using Barry's Glicko-1 calculator I find an ELO true performance rating for the unrated player using pax's calculator (here: http://www.paxmans.net/performance_calc.php) and put that in with a ?? . I think this is probably going to give better results than leaving the unrated player out altogether.

A more advanced alternative is to find a Glicko performance rating for the unrated player and if you don't have a way to do this directly, then you can do it roughly by trial and error by finding the rating at which a player's rating would not change based on that performance. For instance I did this to estimate the rating of the then-unrated Hobart player Alex Nagy using Barry's G1 calculator and got a figure of 1647. Nagy's actual rating which was based solely on the same tournament came out as 1643.

Rincewind
03-12-2011, 08:27 AM
It seems to be giving fairly accurate results, without working out all my opp's RD's yet - by sensitivity do you mean that volatility figure?

No by sensitivity I mean how much the final number varies depending on the input. So if you enter a bunch of games with opponents having a "!" and enter 95 as the RD. What is the difference if you change all the 95's to 80's, say.

As for the volatility, the only thing to do is to pick a number and run with it. Perhaps Bill can advise a indicative value to use for most purposes. It is less important in the sense that volatility affects how much your rating varies in a single period and not the value that the system assigns to wins over your opponents. So the only volatility that's important in a single calculation is yours. (This is not true in the ACF system which uses a provisional rating system and therefore there is a feedback loop for the volatility of all opponents to affect the volatility of all others and likewise even in a vanilla glicko-2 system over time the effects of volatility will permeate throughout the rating pool).

schess
04-12-2011, 11:48 AM
Can you please add the december acf ratings to your website

Rincewind
17-12-2011, 01:58 PM
Can you please add the december acf ratings to your website

Thanks for your patience. Tennis was rained out today and the ratings have been updated. These two facts are positively correlated. ;)

ChessGuru
26-12-2011, 05:21 PM
:)

Rincewind
23-05-2012, 04:27 PM
The old website has moved to http://web.bjcox.com/ and the new website http://www.bjcox.com currently has the Glicko Calculator working but not much else. It will take a little time to get all the files in the download area moved across. Please us the old site for access to the old ratings lists, etc in the interim.

If you need a direct link to the Glicko Calculator it's: http://www.bjcox.com/?page_id=2

Rincewind
03-07-2012, 09:40 PM
The Glicko-1 Calculator has been upgraded to allow users to load and save data. This should enable users to keep adding games throughout a rating period without having to continually re-enter the results from earlier tournaments. Read more here (http://www.bjcox.com/?p=208).

Max Illingworth
04-07-2012, 12:55 AM
Is there a way to increase the number of lines for inputting data on the calculator, so that all the calculations can be made in one go? Often in a rating period I've played more ACF rated games than the calculator allows me to input.

Rincewind
04-07-2012, 01:45 AM
Is there a way to increase the number of lines for inputting data on the calculator, so that all the calculations can be made in one go? Often in a rating period I've played more ACF rated games than the calculator allows me to input.

It is arbitrarily set to thirty mainly due to HTML not effectively implementing arrays. The trouble with adding more is more repetitive code and how many lines is enough? Apart from a small number of super active juniors like Norm Greenwood, who needs more than 30 lines? :P

Rincewind
11-12-2013, 03:52 PM
I've added a new facility to my site called "ACF Rating History". It allows you to graph the ACF Masterfile rating of up to 5 players on a single set of axes for rating periods between Dec-2000 and Dec-2013. For example the last six years graph for the top five players on the current U/20 list appears as follows.

2437

So now you can plot your rating progress or benchmark yourself against up to 4 friends.

Enjoy!

Kevin Bonham
11-12-2013, 05:12 PM
So now you can plot your rating progress or benchmark yourself against up to 4 friends.

Or enemies! I think people can figure out who this one is.

2438

Rincewind
11-12-2013, 06:38 PM
Or enemies!

Indeed, cohorts of any disposition, amicable or inimical.

BTW the image generation using server side PHP scripts which hits my database. So if you want to save some output or post a graph somewhere, please save it on your client and use the static picture file. Thanks. (I've already edited the post above).

Sir Cromulent Sparkles
11-12-2013, 08:48 PM
is fumblegoat still without a club ?

Rincewind
12-12-2013, 10:34 AM
is fumblegoat still without a club ?

His rating is still wriggling so he must be getting some rated games somewhere.