Catan Algo is broken.

[nono_uby]

Same. I just had the most frustrating game I ever had.
Almost only the same 2 numbers would happen, letting 1 of the player completely CRUSHING everybody else. It definitely felt outrageously unfair. Sick of it ..

Here the image, 6 was drawn 25 times !!!!
While numbers like 7 or 8 would happen 6 to 8 times.
https://ibb.co/Kpt51GSt

What is this ?!?!??

[Jontia]

Here the image, 6 was drawn 25 times !!!!
While numbers like 7 or 8 would happen 6 to 8 times.
https://ibb.co/Kpt51GSt

What is this ?!?!??

Those are percentages, not the number of rolls.
And yes it's whacky, but it's one game. Odd things happen.

Checking online, the probability of getting exactly 10 rolls of 6 in 40 rolls is about 2.5%. so 2 or 3 games out of every 100 will get that 25 in the 6 row.

That's not very unusual.

[euklid314]

Just to add to Jontias correct arguments:

The game in question had 64 moves, i.e. 6 was rolled 16 times (25%), 7 was rolled 5 times (8%), 8 was rolled 4 times (6%).

The number 6 having rolled 16 (or more) times in such a game of 64 moves has a probability of 1,26%. Thus, every 80 games it should happen.

Since you will notice also if any other number has rolled far more often than expected you will see this in a lot of games. And this is expected if you take random rolls! It is extremely unlikely (almost impossible) that all numbers are rolled according to their expected value if you have a small sample size of 64 rolls. There will always be outliers. Just try it out with physical dice - roll them 64 times and do a statistic!

Just view it like that: The expected value of a 6 is 5/36 = 13,9%. In a game of 64 rolls this amounts to approx. 9 times that a 6s should appear. But since the rolls are all independent of each other, the 6s dont "know" that they should appear 9 times. In one game they will appear 4 times, in others 0 times, in others 20 times. All these cases are less rare than you think. Only on average (i.e. over hundreds of games) the number of 6s will be approximately 13,9%.

[nono_uby]

Thanks guys for the explanation (especially euklid314, I appreciate the time you took to break it down into precise statistics, it definitely ease the frustration to see things mathematically).
It's the first time I dive into game statistics, apologies for the mix up between percentages and number of occurences.

[Limitless_90]

Just wanted to add. It’s really broken. I had a game with 14 12s… my opponent won because he had 2 cities on the 12. Make it make sense. I feel like Catan’s algo pre chooses who it favors. I also had an instance where on two back to back arena games the same person won 2x. He got first pick both times, his numbers constantly got rolled, and he got all the 7s so i was always blocked.

[Slartibartfst]

Just wanted to add. It’s really broken. I had a game with 14 12s… my opponent won because he had 2 cities on the 12. Make it make sense. I feel like Catan’s algo pre chooses who it favors. I also had an instance where on two back to back arena games the same person won 2x. He got first pick both times, his numbers constantly got rolled, and he got all the 7s so i was always blocked.

you've played over 1,300 games on here. if you did not see some pretty crazy anomalies statistics wise in that time, THAT would be an issue.

[Romain672]

Just wanted to add. It’s really broken. I had a game with 14 12s… my opponent won because he had 2 cities on the 12. Make it make sense. I feel like Catan’s algo pre chooses who it favors. I also had an instance where on two back to back arena games the same person won 2x. He got first pick both times, his numbers constantly got rolled, and he got all the 7s so i was always blocked.

I find it https://boardgamearena.com/gamereview?table=679791526
So here is your numbers of
2 3 4 5 6 7 8 9 10 11 12 rolls (total 73)
2 5 3 4 6 17 10 5 7 4 10

So again that's %, so you only rolled ten 12s.
Rolling those or more has a probability of 0,0034%. 1 chance out of 29133.

But again, you can do that reasonning for any number, so let's make that probability 11 times more likely. We got 1/2650.
You played 1300 games, so I think the probability of that happening once for you was 30%.

There is still one more biais which is to stop that streak and tell that something is wrong at any moment you like, but that's fine.

And by the way that the kind of game which, if you got like 3 more of them in your future games, then this would be extremely weird and could require investigation. But having 1, or even 2 if you are (un)lucky is totally expected.

.

I will finish by adding something really strong, if you take a game of anyone which already talked by the past in the forum (and which is dated after their first post in the forum), that will have much more weight.
If 1000 people play a Catan game every day, 10 of them already talked by the past on the forum, and 1 of them report the weirdest game of the day, then you will find very low probability.
But once you goes from that set of 1000 players to a set of 10, then that reduces already drastiquely the amount of valid game to analyse already.

And Limitless_90 never talked by the past on the forum, so is part of that group of '1000 players'

[DivergentRalph]

Same. I just had the most frustrating game I ever had.
Almost only the same 2 numbers would happen, letting 1 of the player completely CRUSHING everybody else. It definitely felt outrageously unfair. Sick of it ..

Here the image, 6 was drawn 25 times !!!!
While numbers like 7 or 8 would happen 6 to 8 times.
https://ibb.co/Kpt51GSt

What is this ?!?!??

Using binomial distribution, the probability that many 6s (or more) would be rolled in your 100 rolls game is 0.00218

[DivergentRalph]

Just wanted to add. It’s really broken. I had a game with 14 12s… my opponent won because he had 2 cities on the 12. Make it make sense. I feel like Catan’s algo pre chooses who it favors. I also had an instance where on two back to back arena games the same person won 2x. He got first pick both times, his numbers constantly got rolled, and he got all the 7s so i was always blocked.

I find it https://boardgamearena.com/gamereview?table=679791526
So here is your numbers of
2 3 4 5 6 7 8 9 10 11 12 rolls (total 73)
2 5 3 4 6 17 10 5 7 4 10

So again that's %, so you only rolled ten 12s.
Rolling those or more has a probability of 0,0034%. 1 chance out of 29133.

But again, you can do that reasonning for any number, so let's make that probability 11 times more likely. We got 1/2650.
You played 1300 games, so I think the probability of that happening once for you was 30%.

There is still one more biais which is to stop that streak and tell that something is wrong at any moment you like, but that's fine.

And by the way that the kind of game which, if you got like 3 more of them in your future games, then this would be extremely weird and could require investigation. But having 1, or even 2 if you are (un)lucky is totally expected.

.

I will finish by adding something really strong, if you take a game of anyone which already talked by the past in the forum (and which is dated after their first post in the forum), that will have much more weight.
If 1000 people play a Catan game every day, 10 of them already talked by the past on the forum, and 1 of them report the weirdest game of the day, then you will find very low probability.
But once you goes from that set of 1000 players to a set of 10, then that reduces already drastiquely the amount of valid game to analyse already.

And Limitless_90 never talked by the past on the forum, so is part of that group of '1000 players'

Using binomial distribution, the probability of that many 12s is 0.08 so not that low

[Romain672]

Using binomial distribution, the probability of that many 12s is 0.08 so not that low

No? I double checked, I stand by my numbers.

What do you mean by 'using binomial distribution'?
I use my doc: https://docs.google.com/spreadsheets/d/ ... 1846324187