Elo Calculation (K factor display?) is Inconsistent

[sprockitz]

The Elo calculation seems just plain wrong to me in this case. One players total Elo variation is multiplied by 3/4 while another player is not.

Here is the game: https://boardgamearena.com/1/nippon?table=56719312 (though this also happens in a bunch of other games too)

My K factor is 20, which nets me 2.07 Elo points for beating both opponents...this however ends up being multiplied by 3/4 (presumably b/c it is a 3 player game)

Marc_D has a K factor of 30 and that nets him 13.59 Elo points...however his net gain is NOT multiplied by 3/4

So my question is why? Is this a bug or is this somehow the intention?

Edit: So is the K factor incorrect...based on games played mine would be 20 (21 games played) and Marc_D would be 40 (12 games played). So it seems like my player multiplier is not included in the K factor, which is why the total of 2.07 doesn't match the 1.55 actually gained. However Marc_D displays a K factor of 30 instead of 40 (the 3/4 multiplier is already taken into account here).

[sourisdudesert]

Hello

You can place your mouse over the black arrow corresponding to each ELO variation to have the details of the computation.

Hope this helps.

[sprockitz]

Hello

You can place your mouse over the black arrow corresponding to each ELO variation to have the details of the computation.

Hope this helps.

Yes, that's where the confusion comes in. Using the back arrow my two scores add to 2.07, but I only gain 1.55.

[Jaxxxis]

So my question is why? Is this a bug or is this somehow the intention?

Hmmm, I see you are a top player...there has been another guy recently accusing BGA of somehow manipulating games...if what you write about those miscalculation is true, there should at least be a little concern now...

Lets say that generally BGA should have interest in holding the race for top spots tight, that should ensure them more played games of course. People could lose interest in playing if the top guys are too far away in the rankings.

So, the other guy complained about algorithm manipulation, you about elo mis calculation, and both of you are top players.

SO, maybe it's time for BGA to discuss this, at least that elo miscalculation should be explained, which they didn't do in this thread.

[eoc]

So, the other guy complained about algorithm manipulation

Yeah, that was utter bullshit and you're not helping this thread's very valid case by conflating it with the other matter. Please make an effort to differentiate next time. Of course, if BGA were even slightly more open about the actual problems and bugs with their calculation (and ideally with the computations themselves), it wouldn't have come up in the first place, so that part asking them to step in is very true regardless.

Lets say that generally BGA should have interest in holding the race for top spots tight, that should ensure them more played games of course. People could lose interest in playing if the top guys are too far away in the rankings.

Again, please at least show an attempt at differentiating ELO rank calculations and how BGA chooses to utilize them (which, as I understand, is about to change significantly anyways). The best course of action in the current system if you're a high-ranked player who cares about that ranking is to rarely – but in a regular interval to avoid becoming unranked – play, and only when you're sufficiently confident to win. That's the exact opposite of what a page skewing their algorithms and computations towards "ensuring more games are played" would do. Even then, not all players are motivated to a similar degree by rankings or catching players occupying certain ranks, which means you cannot tie a possible loss of interest to that stuff. Problematic player behavior and platform bugs are two points that I suppose correlate a whole lot more to losing player engagement here.

Everybody who does care a great deal is welcome to pick a rarely-played game on BGA – there won't be much competition for the top spot – practice it offline or train somewhere else, and dash up the ranks while their initial K-factor is still huge.

[Jaxxxis]

@eoc

I am always pleased that my doc follows my mental health situation, now if my stats guy (Een) got us an answer on how Sprockitz's elo got screwed, that would be much more interesting than your comments

[eoc]

I am always pleased that my doc follows my mental health situation

Could you please stop derailing this topic with nonsense about your alleged health situation. That's a matter for you and your actual doctors and certainly not for the BGA forums. Thank you.

[Jaxxxis]

Cmon doc, don't become aggressive, maybe in our next session I will play the part of the doc to determine what's the exact condition you suffer from.

[sprockitz]

I believe the final Elo value is correct, but showing the intermediate values when they don't add up to the total (in some cases) is very misleading. I believe the multiplier is really the K value times a player multiplier which seems to be included in the K value for some (40 changed to 30), but not others (20 not changed to 15)

[Tulivuori]

I'm not sure if my conclusions are exactly correct, but that's how it roughly works: https://boardgamearena.com/forum/viewto ... 094#p48094