Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
ELO and changes Underway at SimplyBG 2023
The ELOs at SimplyBG are currently undergoing reengineering. These changes are not changes to the algorithm in any way. What is happening mostly is the additional support to ELOs for increased granularity based on both player and match types being played.

We have always supported ELO ratings on matches as described on this page:

Attached see recent chatter from the chat board in the Openplay Parlour  related to ELO discussion.

Please feel free to comment on this post.  You're welcome to subscribe to it, as doing so will send you notices whenever updates are made.

Update May-01-2023:
The Player stats showing ELOs has been updated to show ALL your ELOs and a bar-configuration-tool to graph 4 ELOs at a time.

Update May-05-2023:
The presentation of the ELOs for players is currently being worked on. In the coming days and weeks, you will be able to view all ELOs for any player.


Attached Files Thumbnail(s)
Click Image for Full View.

.pdf   openplay_elo_chatter_april_2023.pdf (Size: 1.93 MB / Downloads: 26)
Add me to the list of those who are bewildered and befuddled by the changes but ours is not to reason why, ours is but to do and try ...
I think there is a fundamental flaw in being so reluctant to use a single ELO rating for each of the main variations of the game.

There is a strong correlation between the ability of a player over both length of match and (probably slightly weaker) over the use of a timer. 

So somebody like me, for example, who plays a 3 or 5 point match very rarely, and whose 9 point matches are all in the context of one tournament, will apear to have very different elos. However anybody who wanted to know whether I was any good would be crazy to look at anything other than my 7 point elo, because that is based on a vastly greater sample size.

Come to think of it I could, if I cared enough, use this to massage my elo. I could set up a match card with my lowest rated variation, thus giving the impression that I am, say, a 1400 player, then decline to play that length and, when invited to play, offer my (more realistic) 1600-rated length!
Jeremy good points.  Here's some responses to your thoughts:

"use a single ELO rating for each of the main variations of the game." ... this is not the case. The ELOs are not managed this way. They are managed into classes of match types however, that is true.  And there is overlap. Meaning, when you play and finish a match, more than one of your ELOs is updated for all the ELOs that fit the match specifications.  So if you played a 3pt match with timer, both the 3pt match ELO without timer and the ELO for 3pt ELO with timer are updated.  Also, the ELO in the class "any match with or without timer of any match length" is also updated. The later being the parallel ELO of the single-ELO system. So this ELO is still there, defined in a very specific match class.

In a single ELO system, a player who plays nothing but 3 point matches will not be identified differently than one who plays 9 point matches. They might have the same ELO and very different skills. This is what is broken.  Your 3rd paragraph is something we agree with and have no conflict with for this matter. We want more granularity of skill aligned with the types of matches played.

These "classes of match" types is something we can post and share and is a good question if someone wanted to see the granularity of these ELOs and the classses defined. No one debating these ELO changes has gotten to that point of inquisitiveness yet.  We are certainly open to refining those classes and adding places to see them at the site associated with the players.

On your final statement about manipulating the ELOs to appear worse or better to trick someone is a good thought. In time, the display of all your ELOs will be available to everyone to see. Also, the ELOs for match offers are now always shown. And those ELOs are specific to the match your offering. So likely the ELO deceit attempt would be exposed and fail.  So we feel we're on the right path here.  Certainly the previous single-ELO system was easy to play tricks with those are slowly going away now.

The main focus right now is updating the site so everyone can see these multiple ELOs in the proper context as aligned with matches.  This is being worked on now and for the coming weeks.
Question about the ELOs I see in my 'Play Statistics'. I see an ELO of 1677 for 'Match-Play of any length with or/wo cube w/Crawford rule'. I see an ELO of 1543 for 'Match-Play of any length with or/wo cube with or/wo timer'. Lastly, I see an ELO of 1542 for 'Match-Play w/Crawford rule'. Apart from a couple of money games, and one with a timer, every match I have ever played has been with cube and Crawford rule. So why the large disparity between these categories of match? The two categories with the lower ELO do not appear to exclude any of my many matches, so I don't understand why they show a much lower ELO than the first category.
Lazy quiet evening thoughts:
While I read comments on ELOs I'm not too bothered by the subject. Occasionally try to understand what the experts discuss but mostly prefer to laugh about it all.

First came across it when I joined here and thought, 'well, it seems to make good sense' and it definitely did, but I didn't much try to analyse how it worked. Trusted it showed a combination of each player's ability, experience and luck! of course, over a certain span of time/# of games.

I started at 1500, like every other starter, dived to 1200 within 2 weeks, then creeped up to 1400 over 2 years and happily(?) live there now.

But lately I am surprised to invite someone who shows a low 1500's on the Openplay parlour and once the match starts they appear as mid 1600's. Does seem odd, though I'm sure there's proper science behind this. Meanwhile, I only see one number for me and it seems to move normally as it always has.

There must be a percentage of us fans that aren't so worried about being able to see ELOs of different types of game because 98.5% of the time we play the same game, in my case 7pt.
Replying to "lordogmore":

The disparity amounts to time. Many like yourself may see a higher ELO like this for the "any match, any length, etc" match specification.  It represents the ELO you've had prior to the beginning of 2023.  When the expansion to multiple ELOs occurred and was implemented near the end of 2022.  Good attempts were made to try to initialize the expanded ELOs using this, the only ELO available for players. In many cases, the expanded ELOs were defaulted to 1500 tho.

From around 2010 to about 2019, nine years, there we subtle shifts in how ELOs were being stored. In these nine years, many had been able to gain ELO points leveraging weaknesses in the allowances of ELO based played.  One of the main ways was to use a "ramp up" policy that allowed for an opponent to have a slightly advantaged formula for experiences 400 or less. Account shifting had been happening as well as some blatant "play versus oneself" using multiple accounts, always winning on a selected account.   Over the years, these subtle storage and the ramp up and multiple account play has been locked down.

Around the end of 2018, and 2019 during the boom playing times associated with the worldwide Covid lockdown, is when these ELO bloating techniques were being noticed were beginning to get locked down.

Because of the history of ELO management here, you could argue the lower ELO initiated in 2023 is more precise and accurate for players since it's more difficult to find ways to manipulate or circumvent the integrity of these skill ratings.  Even tho you may not have noticed or you weren't able to see these ELOs, they were being managed along with the pre-2023 ELO.  At this time, those newly expanded ELOs are now about 6 months in computation.

There are other manners that have been used by some looking only to grow their ELO. For example the practice of seeking out and playing those who didn't understand rated play, short matches with cube. Sort of an act of prey upon those who didn't know or who didn't care about ELO. Maybe call it ELO stealing if you will.

The management of ELOs is still undergoing adjustments as we monitor the system as a whole. Protecting a well earned and deserved ELO is the primary focus.
(04-29-2023, 06:18 AM)j.pickard933 Wrote: There is a strong correlation between the ability of a player over both length of match and (probably slightly weaker) over the use of a timer. 

I've been playing BG for many years but I don't see a correlation between match length and skill. Sure, in a 3 point match a short lucky streak with the dice can win. But if a strong player played a weak player 100 3 point matches wouldn't the strong player win the most?
And I've lost 7 point matches to weak players so....9 is too long for me..

Forum Jump:

Users browsing this thread: 1 Guest(s)