99% loosing with Green

Sort:
Indipendenza

(a general consideration: it is clear that a game where the average chances are SIGNIFICANTLY different from the random probability, here 25%, can't be acceptable and something has to be done: the obvious solution would simply be to give handicap to red and yellow, whose amount is rather easy to calculate: to analyse the points that were in average reached by all 4 colours in 2000+ games, and to counterbalance the average lack of points for G and B. Typically G would begin maybe with 4 points and B with 2.5 (for example!).

hest1805
Arseny_Vasily skrev:
hest1805 wrote:
Arseny_Vasily skrev:

after 1.r h3 and 1.y i12 green has to defend against mate on the first move)

I think it won't end well)

Green goes l10 or k11 then l8, and what's the problem? 

it was a fun fact, no more

the problem is that by simply rearranging the king and queen, the "color" problem cannot be solved, the initiative is again in the hands of R/Y. need something more powerful)

I wouldn't be so sure about that, this deceptively small change does actually make the opening a lot more even dynamically.

Indipendenza

OMG

Indipendenza

If the move sequence were to change from time to time, it would become a completely another game.

Manuelgzma

It could be a funny variant but nothing related to the point of the discussion.

weesab

i think its because in 4pc bg goes second so it is not that fair because ry gets headstart

 

that is why bg loses more

4Player-Pug
weesab wrote:

i think its because in 4pc bg goes second so it is not that fair because ry gets headstart

 

that is why bg loses more

false. Blue goes before yellow.

The reason (or a good educated guess) is that green loses first, since he is last, and blue is to suffer next because of the "rule" opposites always team. Then ry just plays on

weesab

yeah i agree with 4player pug

Indipendenza

Guys, this problem (because it is a problem indeed) is quite easy to address.

The way to do it is as follows: every day let's say at 0:00 UTC the system looks into the latest 100 000 games played with all players being in the Top 100 (like that, no need to fix a threshold...) and determines what would be the appropriate handicap to add to which colour in order to make the things 100% balanced. (That would require to check what was the average R, B, Y, G amount of points at which place!). And these values would be permanently publicly displayed and will be used in order to adjust. Example: R will begin with 0, B will begin with 1.9, Y will begin with 0.2 and G will begin with 3.1. Like that we'll ensure that the average proportion of wins would be the same with whatever colour.

AaronSmile
grable wrote:

@Indipendenza:

I can't specifically filter out games that got to a 3-player stage sooner, but I do have the ability to filter by game duration. Here are the same two charts (1st-place by colour, then 4th-place by colour), only looking at games lasting 200 or more plies:


@Manuelgzma:

Filtered to only games where the minimum rating for any player is 2500. There are only 625 games meeting that criteria, so the sample size here might be at issue. Again, 1st-place finishes by colour, followed by 4th-place:

The last chart there got me thinking. I mean 65% probability that B or G finish in 4th among 2500+ players? Wow. - To further explore this trend, I created a graph of BG combined loss probability versus average game rating.

Note: The last bin, 2675 - 2825 average ratings, only contained 91 games, as compared with at least 200 per bin for lower rating categories.

AaronSmile

AND WHAT ABOUT 2850-3000 games?

BossAtha

Its jst colors...

AaronSmile
grable wrote:

@Indipendenza:

I can't specifically filter out games that got to a 3-player stage sooner, but I do have the ability to filter by game duration. Here are the same two charts (1st-place by colour, then 4th-place by colour), only looking at games lasting 200 or more plies:


@Manuelgzma:

Filtered to only games where the minimum rating for any player is 2500. There are only 625 games meeting that criteria, so the sample size here might be at issue. Again, 1st-place finishes by colour, followed by 4th-place:

The last chart there got me thinking. I mean 65% probability that B or G finish in 4th among 2500+ players? Wow. - To further explore this trend, I created a graph of BG combined loss probability versus average game rating.

Note: The last bin, 2675 - 2825 average ratings, only contained 91 games, as compared with at least 200 per bin for lower rating categories.

AaronSmile

What about a bin with 2825-2975? Probably no sample size