Board Game Arena

Backgammon dices are absolutly unreal

rab1 wrote: ↑11 November 2022, 11:44 Nonsense I’m afraid- the games are always rigged in favour of the player that has lost the most games recently- Algorithms are shocking. They have to change to random

THATs what I say!!
Algorithms in most cases prefers the player with the bigger amount of points.
In most cases, they are the player who has winnig more games, recently.
This kind of Algorithms , seems to be used especially at Arena games.

I'm pretty sure, here were used different settings on Algorithms for different kind of games : Arena game, Turnier game, simple game

yrrepnadohr wrote: ↑29 January 2023, 11:04

rab1 wrote: ↑11 November 2022, 11:44 [...] the games are always rigged in favour of the player that has lost the most games recently [...]

THATs what I say!!
Algorithms in most cases prefers the player with the bigger amount of points.
In most cases, they are the player who has winnig more games, recently.
[...]

"Most" being "around 50%" for 2-player games

Jellby wrote: ↑29 January 2023, 12:08 "Most" being "around 50%" for 2-player games

hahaha

My Englisch is not the best. My understandig of "most" (in German = meistens) , means mor for me : clearly a lot more than 50 % (over 80 - 90 %) is meistens (most)

yrrepnadohr wrote: ↑29 January 2023, 15:08

Jellby wrote: ↑29 January 2023, 12:08 "Most" being "around 50%" for 2-player games

hahaha

My Englisch is not the best. My understandig of "most" (in German = meistens) , means mor for me : clearly a lot more than 50 % (over 80 - 90 %) is meistens (most)

So why do you say "most" when it's around 50%?

Silene wrote: ↑29 January 2023, 15:16

yrrepnadohr wrote: ↑29 January 2023, 15:08

Jellby wrote: ↑29 January 2023, 12:08 "Most" being "around 50%" for 2-player games

hahaha

My Englisch is not the best. My understandig of "most" (in German = meistens) , means mor for me : clearly a lot more than 50 % (over 80 - 90 %) is meistens (most)

So why do you say "most" when it's around 50%?

show me the statistics for "your" around 50%

yrrepnadohr, please note that you have to play a PR of around 5 to 7 to be able to consistently win 65% on BGA. This is average tournament level, nothing special. Far from any proficiency, easy to reach with reading some books and playing some tournaments. I will look at your last 3-5 games but I assume that you do not play at this level at the moment (I have looked at your last 3 games and your average PR of 16.68 supports my claim).

PR ... Performance Rating according to XG

It is not the case that better players are luckier, but they play more according to theory (i.e. follow the basic concepts of holding games, priming games, back games, not just race games as a beginner would do) and thus they optimize their chances to win even when they get bad rolls. If you don´t see these optimal moves than you are missing chances to win - even if they were there.

If you are in a miserable position and you play the only move that will allow you a slim 1/36 of winning then this is great. You will win 1 out of 36 games instead of 0 out of 36 games and your opponents will probably be upset every 36 games since you turned around a "lost" game with an incredible roll. Or on the other hand, if you are in a clear winning position and you make a move that wins 95% of the time instead of the optimal move that would lead to winning 98% of the time then this will be a huge (double) blunder! In most cases you won´t notice your wrong play since you win anyways, but you will lose more then you need to and you will lose more of those "won" positions. You might blame the dice but in such a case it was your move that made it possible.

My advice: Get deeper into the game of Backgammon, it is absolutely worth it. Read one or two books about basic strategies and you will see your winning percentage increase. Then other players may accuse you of being lucky. But your luck will have stayed the same, you will just get more wins.

Your most recent games:
#341492004 (you lose against a player with 379 Elo):
Your PR = 19.26 (... Casual player according to XG), 9 Errors, 5 Blunders
Opp PR = 3.89 (... World Class according to XG), 5 Errors, 0 Blunders
Your luck rating: -0.360 ... i.e. your opponent was playing *much* better but he also had (a little bit) more luck

#341497811 (you win against a player with 301 Elo):
Your PR = 14.25 (... Intermediate Player), 5 Errors, 3 Blunders
Opp PR = 7.90 (... Advanced Player), 4 Errors, 2 Blunders
Your luck rating: +1.223 ... i.e. your opponent was playing clearly better but you were very lucky

#341487967 (you lose against a player with 210 Elo):
Your PR = 16.53 (... Intermediate Player), 7 Errors, 3 Blunders
Opp PR = 14.81 (... Intermediate Player), 5 Errors, 4 Blunders
Your luck rating: -1.092 ... i.e. you played both with a very high PR but your opponent was very lucky

My analysis after 3 games: You played against 2 very strong players (Elo Top 100 on BGA) and against a weaker player (Elo similar to yours) and you had one win and one loss against the top players and a loss against the weaker player. All three players played better than you, the top players much better. The most luck was given to you in game 2 and to your opponent in game 3. Not seeing a favor for the higher ranked players in your last 3 games - but I have analyzed only 3 games, and will stop here. These 3 games rather suggest that the players with lower ELO have more luck but of course that would even out if I watched more games. Neither stronger nor weaker players are favored in BGA (both theories pop up every now and then).

The third game shows nicely what I have said further up. You were crushing your opponent and were already bearing off checkers while he still had a checker on the bar. In this very late phase of the game you made 2 errors and one huge blunder! Thus, you greatly reduced your winning chances and then had to give your opponent a shot - which he made. This loss was very much due to your 3 errors in the end game. Playing according to theory helps to *greatly* reduce such unfortunate late losses. End game theory is relatively simple, Backgammon as a whole is not...

yrrepnadohr wrote: ↑29 January 2023, 16:06

Silene wrote: ↑29 January 2023, 15:16

yrrepnadohr wrote: ↑29 January 2023, 15:08

hahaha

My Englisch is not the best. My understandig of "most" (in German = meistens) , means mor for me : clearly a lot more than 50 % (over 80 - 90 %) is meistens (most)

So why do you say "most" when it's around 50%?

show me the statistics for "your" around 50%

On evereyone's collected statistics that I ever looked at (including yours and many other doubles-conspiracy-theorists), there is a number of doubles that is very close to 1/6.

Apart from that: a random function using the common random_int() is the easiest thing to implement for simulating a dice-roll. Developers have checked in the past whether it's used correctly and they found it is.

The main thing is that there is no meta-info taken from the game-table and fed to the dice-rolling-function and therefore the result cannot be dependent on rankings/premium-status/arena-points/new-player-status or whatever specifics people made up in the past to be favoured/disfavoured by the dice rolls.
And even if there was some backdoor to do that, it would have to be incredibly well hidden and a conspiracy among many different game-developers (because the same complaints show up in almost every luck-dependent game). And on top of that those developers would additionally have to program an AI/solver for all those games to find out what even IS a good dice roll in a specific situation, which is not as trivial as you might think it is. And then that solver would have to work very subtly to not make it too obvious and not show up in the statistics about doubles that is already collected so it would also have to hand out LESS doubles to the favoured players in other games (that you are not looking at) to not mess with the average.

On the other hand various forms of perception bias are a psychologically confirmed and a widely found cognitive error to be made. The effect is usually bigger in online play because the amount of played games is much higher than is usually is in real live and complaints are heard by many more people (you won't hear about someone's grandma having horrible rolls in her real-live games).

All-in-all: an unbiased random_int is the simplest assumption and nobody has ever shown any relevant set of data that shows irregularities. Single extreme games, short lucky series and gut-feelings do not prove anything because it is normal things that happen on a fair RNG. It would be more suspicious if such things didn't happen.

It's normal for random events to sometimes go crazy in one direction or the other. That's how randomness works.
Humans are prone to remember the noteworthy events for much longer and more intensely than all the boring results that happened between. That's how our brain works.

Romain672 wrote: ↑08 December 2022, 11:05

JBeachy wrote: ↑08 December 2022, 03:12 I just lost a game where my opponent rolled 8 doubles in a row. 11 of their final 13 turns were doubles, and 13 of their 26 total dice rolls were doubles.

Let me check since that claim would be really weird if it was true, here is all his rolls of his game (https://boardgamearena.com/gamereview?table=324587603 ):
ndnnndnnnnnnnddddddddnnddd

13doubles/26, so this was true. it has a probability of (https://docs.google.com/spreadsheets/d/ ... 1846324187 ): 0.009%. 1/11 012.

You find 8 doubles in a row, it has a probability of (8^6) 1/1 679 616.


But a single probability said nothing, so let's go deeper.
Usually when people find some weird events about double happening, they will look for their double, the opponent double, or both number of doubles.

So, let's take back that 1/11 012, I will 'multiply properly' (i will explain at the very bottom what i mean by that) by 3, and find 1/3 671 which I assume would be the probability I would keep for your game.
But not for your profile, since you did 138 games of backgammon, i could go further and 'multiply properly' by 138. And I find 1/27. So congrats, if 27 people like you played backgammon for 138 games and shared in the forum their game with the most amount of double between yourself, the opponent, or both player, you would be the one which would get the most unlikely outcome.

And let's take back the second which was 1/1 679 616.
On your game, there was 26 rolls, so there was 19 (26-8+1) starting of 8 streaks of 8 doubles possible. So let's start by 'multiply properly' by 19: 0.001131%, so 1/88 401. Then multiply properly by 3 because you could take your number of double, opponent number, or both player doubles: 0.00339%, so 1/29 467 which is for me the probability i would keep for that game.
Then for your profile, i can multiply properly by 138: 0.467%, which give 1/214. So if 214 people like you played backgammon for 138 games and shared in the forum the highest number of doubles in a row they got in a single game, you would be the highest.

(and finally about 11 doubles / 13 at the end, it's cherry picking, so I will not analyse that)


My two cents is that it's not likely, so thanks for sharing, but we would need way more testimony like yours to see something weird is happening. Many people just share some thoughs with their game, when in reality, their perception of randomness is wrong and just got slightly unlucky, or just got their one game out of one hundred the most weird.
And thanks for not lieing on your post, it's pretty rare


Finally, since people like saying my maths are wrong (mainly because of langage barrier), I'm open to discussion. But let's said you rolled 5 dices, and got 2 doubles in a row at maximum. It has a probability of 1/36. Then You can have ddnnn or nddnn or nnddn or nnndd for your two doubles in a row, which leave 4 possibilities (5-2+1). So I will multiply properly by 4: "1-(1-1/36)^4" which leave 10.7% chance. You can see that multiply by any number will never make it reach 100% since it multiply the 1-x probability with itself.

Thanks for reading

Just saw this.. pretty awesome breakdown!