Algorithm + starting board

[thebigfella]

There’s no way “Lucky Numbers” is actually based on pure luck. I’m convinced this game is NOT randomized. There’s streaks where I’ll win 8 games in a row and get an amazing starting board each time, then a streak where I’ll lose like 5. I feel like the game knows when I’m winning a lot, I’ll stop playing to end on a good note, then when I lose, I keep playing to get momentum back and therefore makes me lose more. It happens every time. I also feel like I’m being punished because I keep cursing in the chat which I have been moderated for (LOL — btw, cursing at the game and never at the opponent; i always make sure to show good sportsmanship to my opponent). Anyway, this game is consistently screwing me over to the point where I feel the algorithm is AI based in a way to make you keep playing.

Secondly, I wish our Elo took into account our starting board. If I’m playing against someone with a very low Elo who starts with 2, 6, 13, 20 and I start with 13, 19, 19, 20, I should not be losing 20 elo points when I manage to only lose by 1. And if I manage to win with that board, i should be getting way more than 6 elo points. Idk, it’s just not fair.

[Jellby]

There’s streaks where I’ll win 8 games in a row and get an amazing starting board each time, then a streak where I’ll lose like 5.

Yes. You're describing how luck and randomness work. What would you expect? Win 1, lose 1, win 1, lose 1, win 1, lose 1...? That would be far less random.

[ExaltedAngel]

Secondly, I wish our Elo took into account our starting board. If I’m playing against someone with a very low Elo who starts with 2, 6, 13, 20 and I start with 13, 19, 19, 20, I should not be losing 20 elo points when I manage to only lose by 1. And if I manage to win with that board, i should be getting way more than 6 elo points. Idk, it’s just not fair.

This would kinda make sense, but then why not adjust it also based on how good is the first piece you draw? And then the one your opponent draws? And then your second? And then...

You can't take luck out of the equation: it is part of the game and in the long run you will gain and lose Elo from it to the same extent. In the scenario you described you will maybe win a 20-30% of the games, while in the reversed scenario you will win close to 100% of the time. This added to the 70% winrate you have against the same opponent when board distribution is even, makes you compensate what you lose when you are unlucky.

[Romain672]

I took your 899 2p games of 2025 you have 46.9% winrate, and I counted your streaks of win and lose. Then I simulated with googlesheet some random win and lose with that same results, and here is the two results, in a random orders:

Set 1:
- 208 streaks of 1
- 98 streaks of 2
- 56 streaks of 3
- 25 streaks of 4
- 15 streaks of 5
- 6 streak of 6
- 5 streak of 7
- 1 streaks of 8
- 2 streaks of 9
- 1 streak of 10
- 1 streak of 12

Set 2:
- 222 streaks of 1
- 119 streaks of 2
- 42 streaks of 3
- 31 streaks of 4
- 15 streaks of 5
- 5 streaks of 6
- 2 streaks of 7
- 3 streaks of 8
- 1 streak of 11

Aren't those sets similars? If you only keep streaks of 8/9/10/11/12, you got 1/2/1/0/1 vs 3/0/0/1/0. Set A has much more streaks of 6&7, but that's still reasonnable.
Your games was the set 1.

[Skrybowiedzma]

I strongly suggest that you take a perfectly normal coin out of your purse and flip it 1000 times and note the results.
(I know a math teacher who does that as a homework for highscoolers and knows perfectly well who cheated - the cheaters have "neat", random-looking results. People who actually do it always have some long streaks, because that is how randomness really works).

[Meeplelowda]

I strongly suggest that you take a perfectly normal coin out of your purse and flip it 1000 times and note the results.
(I know a math teacher who does that as a homework for highscoolers and knows perfectly well who cheated - the cheaters have "neat", random-looking results. People who actually do it always have some long streaks, because that is how randomness really works).

Similarly, if you ask the average person to write down 20 "random" digits few people will have *any* repeated digits because people think repeated digits = not random. In reality, on average there are no repeated digits in a random sample of 20 only 14% of the time. I.e., 86% there should be *at least one* repeat.