Ridiculous

[euklid314]

Still kinda confused about the X+10, X+20 thingy and I've absolutely no idea whether or not the ELO works like this but I kinda get your logic (I guess?)

In a 2-player game the difference between a win and a loss is always exactly your K-factor.
Example:
*) If your win probability is 0.8 and your K-factor is 20 then you get +0.2 * 20 = +4 points for a win and -0.8 * 20 = -16 points for a loss. A draw lies right in the middle with -0.3 * 20 = -6 points.

*) If your win probability is 0.1 and your K-factor is again 20 then you get +0.9 * 20 = +18 points for a win and -0.1 * 20 = -2 points for a loss and 0.4 * 20 = +8 points for a draw.

In a 4-player game there is an additional factor that dampens the K-factor. I never remember the formula but a 4-player game is not exactly equivalent to 3 separate games that you play. Otherwise a win/loss in a 10-player game would lead to extreme jumps in ELO. But this aspect introduces just a "K_4"-factor for a 4-player game that is equal to your K-factor multiplied by some 4-player-game constant which is smaller than 1.

But then you have the same effect as in the 2-player scenario. The difference between 1st and 2nd place is exactly your K_4-factor, same as difference between 2nd and 3rd, and 3rd and 4th place, respectively. Because as 1st place you beat #2,#3,#4 and as 2nd place you still beat #3,#4 but now you have a loss instead of a win against one player (see 2-player scenario).

Thus the difference between 1st and 4th place is always exactly 3 times your K_4-factor - independent of the ELO values around the table.

[Jol_]

My brain exploded but I found it easier to understand like this.

1) If your win probability is 0.8 and your K-factor is 20 then you get +0.2 * 20 = +4 points for a win and -0.8 * 20 = -16 points for a loss. A draw lies right in the middle with -0.3 * 20 = -6 points.

2) If your win probability is 0.1 and your K-factor is again 20 then you get +0.9 * 20 = +18 points for a win and -0.1 * 20 = -2 points for a loss and 0.4 * 20 = +8 points for a draw.

So in 1), once you join the game you first receive a -6 point deduction. A win gives back +10 and a loss deducts -10.
In 2), once you join the game you receive a +8 point bonus. A win gives +10 and a loss deducts -10.

In both cases, the -6, +8 initial adjustments (should be the "X" you're referring to) are the "sunk cost" and have no effect in the decision making in the game. The actual payout is [+10, -10] which is a symmetrical payout. No one should be able to exploit by over-aggression since the actual payout is always symmetrical.

That being said, if you're an ELO 400 you probably wanna avoid joining a full-ring table of ELO 100 coz you've got a huge initial "sunk cost" and really have to outperform to get the ELO points back. But once you've joined the game the decision made should be consistent vs joining a table with average ELO 400 opponents - it could be [+8, -8], [+12, -12] depending on the "k_factor" thingy but you never win more points than you lose or vice versa. It's always a symmetrical payout.

[ErikLevin]

The chip EV / ELO EV still feels like a part of the reason some players here may feel the "bad beat" happens far too often. There are tricky spots that have +ve chip EV but -ve ELO EV, and if they stick to chip EV way of thinking, their ELO may underperform.

Yes, but isn't this just the same as the difference between cash games and sit 'n' go. In SNG played for money, one chip does not equal one money, and there is a difference between money EV and chip EV. On BGA, the games are SNG, but the payout is of course in Elo rather than money, so the same effect occurs.

[euklid314]

My brain exploded but I found it easier to understand like this.

1) If your win probability is 0.8 and your K-factor is 20 then you get +0.2 * 20 = +4 points for a win and -0.8 * 20 = -16 points for a loss. A draw lies right in the middle with -0.3 * 20 = -6 points.

2) If your win probability is 0.1 and your K-factor is again 20 then you get +0.9 * 20 = +18 points for a win and -0.1 * 20 = -2 points for a loss and 0.4 * 20 = +8 points for a draw.

So in 1), once you join the game you first receive a -6 point deduction. A win gives back +10 and a loss deducts -10.
In 2), once you join the game you receive a +8 point bonus. A win gives +10 and a loss deducts -10.

In both cases, the -6, +8 initial adjustments (should be the "X" you're referring to) are the "sunk cost" and have no effect in the decision making in the game. The actual payout is [+10, -10] which is a symmetrical payout. No one should be able to exploit by over-aggression since the actual payout is always symmetrical.

That being said, if you're an ELO 400 you probably wanna avoid joining a full-ring table of ELO 100 coz you've got a huge initial "sunk cost" and really have to outperform to get the ELO points back. But once you've joined the game the decision made should be consistent vs joining a table with average ELO 400 opponents - it could be [+8, -8], [+12, -12] depending on the "k_factor" thingy but you never win more points than you lose or vice versa. It's always a symmetrical payout.

Very clearly written down. Thank you.

In Poker one could see it such that you start with a "sunk cost" of, say, -40 or -25 or -10 points (for getting 4th place on the table) and then you get your K_4-factor, say +18, every time you beat one of your opponents. Payout would be [+54, +36, +18, 0] for [1st, 2nd, 3rd, 4th] position.

[Jol_]

Yes, but isn't this just the same as the difference between cash games and sit 'n' go. In SNG played for money, one chip does not equal one money, and there is a difference between money EV and chip EV. On BGA, the games are SNG, but the payout is of course in Elo rather than money, so the same effect occurs.

Yeh it's like SNG and that's why the lengthy discussion on how ELO payout works here - whether a guaranteed 2nd place vs calling the all-in is better for the ELO, esp if we face a full ring table of 100-ELO who just shove all in the very first hand.

Knowing the ELO payout we now have a very specific optimal solution towards those junky hand all-in maniacs. Assume facing the full table of all-in maniacs shoving random all-in the very first hand, and you're in the big blind deciding whether to call:

2-player table: it's just HU so a "guaranteed 2nd place" is meaningless.

3-player table: minimum required equity for an all-in call is 56%. So even if all opponents are shoving all-in with any two cards, we should only call with top 2.3% of the hands {TT+}

4-player table: minimum required equity for an all-in call is 63%. So even if all opponents are shoving all-in with any two cards, we should only call with {AA} only.

5-player table (or more): minimum required equity for an all-in call is 68%. Even the strongest hand AA shares only around 56% of equity in a 5-way pot. So our best response is to fold our AA and let the players go bust. The 44% odds of "bad beat" is enough to drag your ELO down.

-------------------------

And we have a more common situation - assume one player spams all-in every single hand from the start and no one else has dared to made a move yet. Now we need to decide when we should make a hero call:

2-player table: HU situation so the ELO becomes irrelevant. The decision is related to blinds defense so we can refer to some google some push fold charts, but let's not cover it here.

3-player table: minimum required equity for an all-in call is 59.8%, i.e. we should call with top 16% of our hands {55+,A5s+,K9s+,QJs,A8o+,KJo+}

4-player table: minimum required equity for an all-in call is 64.3%, i.e. we should call with top 6.6% of our hands {77+,ATs+,AQo+}

5-player table: minimum required equity for an all-in call is 66.7%, i.e. we should call with top 3.5% of our hands {88+,AKs}

6-player table: minimum required equity for an all-in call is 68.2%, i.e. we should call with top 3.2% of our hands {88+}


Thanks to euklid's input, by understanding the ELO payout we can now develop a specific response against them.

[Semprini72]

Too many suck outs on turns and rivers. Having the best hand pre flop or post flop gets cracked more often than it should. The site rewards loose aggressive players...the opposite of real Texas Hold Em.

This happens on many many sites when playing for micro stakes or free chips. BGA is like the absolutely lowest stakes that one can play anywhere online in the known universe. This just attracts more suckouts since there is no real risk (not even a play money bankroll, just an elo rating). Players stay in the hand far far far too long because there is no risk and when this happens hands are more apt to be decided by luck.

It actually does NOT reward loose aggressive play. It rewards loose PASSIVE players. I'll explain.

Loose aggressive players bluff a lot. Loose passive players call a lot. On BGA players call far more than anywhere else I have seen (and calling too much is the most common error in all of poker). This makes it extremely difficult to bluff which hurts the loose aggressive style a LOT.

The game looks perfectly random to me at least by my casual observation. I fully expect there to be nasty suckouts up the yinyang when J-high jam bluffs the river and gets called by K-high. To see that happen multiple times in only a few weeks here is just bonkers! It's not the deals. It's the players 100%. Trust me. ; )

[Semprini72]

What matters is not what the stakes are, rather only that the players care about the stakes. We are used to poker being for money, but nothing is stopping players from caring about their results for other reasons, for example rating and arena points. After all, no game on BGA is played for money, but most players still care about winning.

Sure, but even the players here who care about winning, their skills are very low. If you have skills, why would you play here when you could play somewhere else and make a few bucks? The most common thing that unskilled players do even when they are trying their best to win is just call way too much.