# Thread: Maris Ratings System (was Glicko Ratings Simplified)

1. ## Maris Ratings System (was Glicko Ratings Simplified)

I hope that if the ACF ever move away from Glicko that they change to a similar rating system such as the one below:

Maris Rating System

a) similar rating changes to Glicko
b) very easy to calculate rating changes – you can simply work them out in your head
c) reduces the affect of under-rated juniors and avoids depressing deflation that has required adjustments

a) The attached excel file has 8 scenarios with comparisons for each of the 5 RD’s (!!, !, blank, ?, ??) – that is 40 comparisons with Barry Cox’s Glicko calculator (I realise that Glicko 2 is currently used). The results are similar: within 2 points for the 16 examples with reliable ratings (!! & !).

The values used for RD are:

RD RDV
!! 30
! 15
Blank 8
? 4
?? 2.5

b) The very easy to calculate method is:
Rat Per = Op Rat +/- 400
Rel RP = Rat Per - Prev Rat
Rat Ch = Rel RP / RDV (the attached excel file shows this calculation for each comparison)

For example 1500!! wins against 1400:
1400+400=1800
1800-1500=300
300/30=10

Note to prevent the loss of rating points when winning against much lower opponents the following applies: if Pre Rat - Opp Rat > 370 then Opp Rat = Pre Rat – 370

c) What has been happening?
There is a continuous flow of juniors that have significant rating gains before they stop playing. For each of these junior rating increases, much of the rating points gained cause losses to seniors (except where they are playing in junior only events).

To reduce the affect of this, the RD of juniors should be kept flexible (RD: blank, ?, ??). This enables junior ratings to change more rapidly and far less seniors would suffer from their gains. For example, a 1300 junior win against a 1500!! senior. The senior will lose 20 points. As per the attached, the junior will only gain 20 points if !!, 40 if ! but will gain 75 if RD=blank, 150 if ? and 240 if ??.

Any comments welcome on a,b, or c

2. well Rob it appears that no one wants to discuss sort of contra-Bill or post-Bill, well you brave enuf to put it up so you deserve a response.

I think a very good attempted compromise that can lead Oz chess out of its divisive camps. That is without judging the merits of the system but taking your word on it being transparent it should pass scrutiny.

Good luck with it see how it goes.

3. How do you decide a player's RD in your system? Note that in Glicko (including G2) there are not five RD classes as indicated by !!, ! etc, but rather RD is a sliding scale and each player has their own RD, for which the !!, !s etc are just published approximations.

4. Originally Posted by Kevin Bonham
How do you decide a player's RD in your system? Note that in Glicko (including G2) there are not five RD classes as indicated by !!, ! etc, but rather RD is a sliding scale and each player has their own RD, for which the !!, !s etc are just published approximations.

The attached excel file includes a worksheet "RD games" that determines the method for assigning RD's for seniors and juniors.

For the calculation of rating changes, the English Chess Federation have used a similar simplistic and transparent rating system for over 50 years.

5. Thanks. So for any adult who is blank, ! or !!, a junior with the same history is two categories further down on the reliability scale. An adult is only below blank if they have <25 games total or if they have <40 including <14 in last year. An adult who played 45 games but the last of them in 1992 would have the same RD as a junior who's been playing 60 games a year for seven years. In a predictive sense that doesn't make sense, but it might still help with deflation protection.

Note that the ACF-Glicko 2 system at present is probably non-deflationary, but compared to basic Glicko it includes volatility, some different treatment of juniors and also a small measure of anti-deflation protection. The treatment of juniors cf adults in your version is so radical I wonder if it would actually inflate.

By the way (though not really relevant to this thread) I note that in the comparative thread on the other forum, Just2Good is claiming that under Glicko as it is, if you are doing poorly it's better to take time off and let your RD build before playing again.

Actually that's only true if the tournaments on return are really strong ones (well above your actual playing strength) and that's something that can't be relied upon; there's at least as much risk they'll be really bad ones and you'll lose a really big lot of points. If you are a certain playing strength but your rating is below that strength then your best strategy under Glicko is to play as often as possible - you won't regain points at the same rate as if you take time off, but you'll regain more of them over several periods at !! than in one hit with an increased RD.

Alternatively, if a player is doing badly because of poor form it makes sense for them to take time off before returning, but that applies to all rating systems, not just or even especially to Glicko.

6. Alternatively, if a player is doing badly because of poor form it makes sense for them to take time off before returning, but that applies to all rating systems, not just or even especially to Glicko.
__________________

can they not take time off before returning? Or is it if not had a break and playing poorly to take time off?

7. I don't understand the post above in the slightest.

8. Originally Posted by Kevin Bonham
I don't understand the post above in the slightest.

With reference to a post on another thread. Perhaps one should drop the '7' and the '0'.

9. Well FG reakons that he could work out his own rating under the old system, plus you could plan tournament runs coz you knew how many points you needed every rating session to hold your rating. This sounds pretty sensible and hands on to me. To lose that capacity is something substantial.

He also reakons you can take a year off from tourney chess, and eventually your rating will be reset every 12 months anyway.

Is he correct?

And regards to my earlier post I was trying to iron out an ambugility, that you must take time off before returning, otherwise you have no break to return from.

Or in case it was me fooling myself (almost impossible) that if there had been no break then to take a break - actually that fits in with what FG says over there. FG has a valid complaint, in fact two, the second about not being able to calculate how many points he needs.

10. Originally Posted by antichrist
He also reakons you can take a year off from tourney chess, and eventually your rating will be reset every 12 months anyway.

Is he correct?
No, as far as I can tell he was pretty much echoing J2G and is wrong for the same reason. If you take time off you take the risk that you will perform poorly on return and your rating will go down further. Of course you might perform well and wipe out the losses but that cannot be guaranteed.

And I think it's good if players actually don't know exactly how to "game" their own rating. Makes the system a purer test of current playing strength rather than of people's abilities to take pragmatic draws to achieve ratings goals.

And regards to my earlier post I was trying to iron out an ambugility, that you must take time off before returning, otherwise you have no break to return from.
Go play those sorts of silly word games with starter. At best it's a case of tautology not ambiguity (correct spelling) anyway. I'm really not that interested and you are off-topic.

11. Originally Posted by Kevin Bonham
Go play those sorts of silly word games with starter. At best it's a case of tautology not ambiguity (correct spelling) anyway. I'm really not that interested and you are off-topic.
I don't even believe it is a tautology anyway, or only in the ludicrous situation where any period of time (however short) is termed a "break". A player can return to chess immediately, or after a (reasonably substantial) break. Since the discussion was talking about how to game one's RD factor, it was clear from context that the break would have to be of the order of months, if not years. But I agree with the off topic comment. If you want to look at grammatical issues can I suggest we try and tease out the problems with

Originally Posted by anticrist
Or is it if not had a break and playing poorly to take time off?

12. Originally Posted by Rincewind

With reference to a post on another thread. Perhaps one should drop the '7' and the '0'.
gold.

13. Originally Posted by antichrist
Or is it if not had a break and playing poorly to take time off?
I agree Rincewind, this has to be the most clueless comment on ChessChat yet. Are there any special prizes for this feat?

AC's usual tendency to drop off a few important nouns sometimes makes his sentences dribble with "Ambugility". Go wonder.

But I have to admit AC is entertaining. O dear, now I'm feeding the troll again