Dice roll - many repeat rolls

[Asesinoloco]

A lot of games feel like this. Dices always repeat same rolls.

And i just played a game where we saw only 2x 8 in all games. I don't remember about the first one but 2nd come in last round, after more than 30 min. So frustrating when u know the probability.
Ok sometimes 6 or 8 only come 6-7 times when the Best come 15 times but only 2 times... When we saw 20x 10 or 16x 9...

[Sparon]

The likelihood of statistically improbable events occurring is very high.

What I mean by that is each statistical event that occurs is on its own very unlikely, but the probability of SOMETHING happening that is statistically improbable is actually very high, as there are many possible combinations that could do this.

Apple had a similar issue with the shuffle feature in iTunes. People complained that it wasn't random enough. Apple responded by making it less random, and the complaints went away.

In other words, humans suck at intuitively understanding what is random and what isn't. Our brains aren't built for it.

[MrBeardy]

You could vote for the cardstack option to be added. This helps ensure a more bell-curve distribution of "dice rolls" if implemented. It's an official Catan variant so should be allowable for BGA to add it. See the bug reports: https://boardgamearena.com/bug?id=69245

[Romain672]

edit: numbers of this post look severely too high. If you assume an average Catan game is 50 rolls and not 23, you can 'divide' by 2 every probabiity, so the final probability is ~1/35 and not 1/73.

I saw 4 times successive 12. This is not random I guess. Diceroll system is broken.

Checked your last 10 catan games, I didn't find two 12 in a row, so look like it was older than that.

There is first an high chance you missremember/lie in purpose. But let's suppose not.

Let's said you have 23 rolls in average per game, and you have 110 games.
That mean in each game you can have 20 start of series of 12 possible. Multiplied by 110, that give 2200 start of series possible.

Probability to roll four 12 is 1/1 679 616.
Let's 'multiply properly' by 2200. Since google can't do it, let's multiply by 2100 instead, that give 1/800.
Then since you can do the same claim for every number, let's multiply by 11: that give 1/73.

You played no other game where you could complain about rng so let's stop there.

1/73, that's pretty much the limit where it could go either way.
Note this number is the probability for a new player who played as many games than illgongiee, to have a serie of similar numbers occuring in a row in one of his games.

But there is still two things:
- illgongiee could be missremembering/lieing: note that if illgongiee is remembering and there was like a serie of 12/12/12/7/12 instead, that would got a probability of about 1/2, and at that point, that would have absolutely no value at all.
- if there was many new players, most of them would never got a serie as unlikely (they could got a serie like OP and share it on the forum, but I would destroy it easily) and would never talk on the forum. And those who got an unlikely event would share it.

[Romain672]

For the others: please be more precise, with the games.

On another Catan post, someone else though that there was sometimes some numbers rolled a little at the first few turns of the game, and those number would have an higher chance to be rolled in a row (or close to each other) in the future.
I checked that part exclusively, and find nothing great. Even on the game (s)he shared, there was some wrong positive. There was a lot of positive too, but that's because the games were picked from a large sample of games.

Or more easily: it's very likely you are in the wrong and see paterns where there is none. Humans like seeing patterns, and are really bad at seeing very high (>80%) and very low (<20%) probabillities.


And about Sparon's post, I answer to most post of that kind on bga since some years. So goes for it and do some claims. But if the claim is weird (ie not counting doubles/similar sum of yourself/all players(/a specific player)), that should be extremely convaincing. Or then you should claim it once here, then saw it happen again in your next ~20/~100 games.

[Romain672]

And i just played a game where we saw only 2x 8 in all games. I don't remember about the first one but 2nd come in last round, after more than 30 min. So frustrating when u know the probability.
Ok sometimes 6 or 8 only come 6-7 times when the Best come 15 times but only 2 times... When we saw 20x 10 or 16x 9...

Oh I missed that one which look not precise.

Let's ask bernouilli again: https://docs.google.com/spreadsheets/d/ ... edit#gid=0
Probability 31/36.
23 rolls.
3.2% to have no 8 in any game
11.9% to have 8 once
21.1% to have 8 twice

Let's take 50 rolls to see how that change.
0.06% to have no 8.
0.5% to have 8 once
1.8% to have 8 twice

And please count two 8, else you really pick the information you want.
As you can see it's not uncommun.
And now you can imagine that for every number on the game. So you can 'multiply properly' that probability by 11. Your 1.8% became 18.1% if you assume you could do the same claim for every number. And this for a single game.

edit: I find your game: https://boardgamearena.com/gamereview?table=309796790
61 rolls, guess I really underestimate that post on my previous one.
1/156 (0.63%) to have 8 twice or less.
Multiplied by 11, that give 1/14.6 (6.8%).
But this is a game picked in purpose.

You did 406 games. That should have happen to you ~27 times already, even if most of those times, that should be about a number you didn't cared.

So it's 1/14.6 to have any number occuring as low as what happen in the game, or something less likely than that in probability (for instance, rolling four times a 2 or a 12 accross a 61 rolls game has around the same probability)

[muntzer]

Please refer to my new thread: Board Game Arena, Catan, virtual dice and randomness - a short trial

viewtopic.php?t=27679

[muntzer]

Also, funny but true - my first game back had 100 rolls. At one point, FIVE sevens came up in a row.

The odds of that happening in a game with 100 rolls = 96 x (6/36) ^ 5 = 1.235% = it should happen once every 81 games.

[Romain672]

Also, funny but true - my first game back had 100 rolls. At one point, FIVE sevens came up in a row.

The odds of that happening in a game with 100 rolls = 96 x (6/36) ^ 5 = 1.235% = it should happen once every 81 games.

The thing is again, you can do the same claim for all 11 numbers.
So if I take your 1.235% and what i call 'multiply properly' per 11, so doing 1-(1-0.01235)^11 which is maybe not fully exact, but should be close enough: you should, by only looking a suit of ANY similar number, find a result as likely or more unlikely than that 12.8% of the time.
1 time out of 7,8 .

But since you seem you didn't cheat on your results, 1/7,8 stay not much likely, I can affoard you that. But meh, that look weak

[muntzer]

So if I take your 1.235% and what i call 'multiply properly' per 11, so doing 1-(1-0.01235)^11 which is maybe not fully exact, but should be close enough: you should, by only looking a suit of ANY similar number, find a result as likely or more unlikely than that 12.8% of the time.
1 time out of 7,8 .

But since you seem you didn't cheat on your results, 1/7,8 stay not much likely, I can affoard you that. But meh, that look weak

Your maths is completely flawed. The chances of five in a row is VERY different for every total rolled. 5 sevens in a row is by far the most likely.

P(five 12s in a row) = 96 * (1 / 36) ^ 5 = one in 629,856 or 0.0001588%

You can't just multiply the odds for 7s by 11 as if they are all identical odds.

For someone who loves contradicting people, I would have expected you would have a grasp on basic probability theory.