This game is broken.

[Nunzio Buonaventura]

I think I need to be more especific. Two players or more play a game involving luck. The match contain several rounds. Each round finishes when both have drawn a card, both have rolled their dice, taken a tile, whatever mechanism is involved.
How come and so often (data is required) during the same game one player gets more doubles, strongest cards, best tiles, over and over again or I don't know 8/10 times while the other nothing. Random events will require that this kind of things don't happen that often, but no, it happens so often. I will show you when I finish my briscola project, but taking the data manually is really annoying: all along the match one player, round in round out, gets the best cards, or long runs of cards of the same color, why? am I living a specially very rare event? why BGA cannot make the process of taking data easier to make this type of analysis? I finally decided to make my own simulator for briscola, not playable just to contrast both approaches.

[euklid314]

I think I need to be more especific. Two players or more play a game involving luck. The match contain several rounds. Each round finishes when both have drawn a card, both have rolled their dice, taken a tile, whatever mechanism is involved.
How come and so often (data is required) during the same game one player gets more doubles, strongest cards, best tiles, over and over again or I don't know 8/10 times while the other nothing. Random events will require that this kind of things don't happen that often, but no, it happens so often. I will show you when I finish my briscola project, but taking the data manually is really annoying: all along the match one player, round in round out, gets the best cards, or long runs of cards of the same color, why? am I living a specially very rare event? why BGA cannot make the process of taking data easier to make this type of analysis? I finally decided to make my own simulator for briscola, not playable just to contrast both approaches.

Please stick to Backgammon data in this forum or you are risking that your posts will not be understood. I just found out that Briscola seems to be a card (?) game on BGA but I know none of its mechanic.

For Backgammon, because you mention doubles, it is highly unlikely that both players get the same amount of doubles if you have true random events. It is highly likely that one player gets more doubles than the other - often by a decisive amount. This happens because the double events are independent of each other. They are not forced to cancel out if they are random events.

If I may change to coin flips, many believe that a longer streak of heads will be offset by a langer number of tails that must follow. This is a misconception.

It is also well known that many people expect a random series to look like 0111010001 0110100101 instead of 1011010101 0111111011 (believe it or not, the latter series was produced by throwing a coin). My example is more extreme than I hoped for but I got the long string of alternating 101010101 although my point that I wanted to prove was the opposite: that long strings of the same result are to be expected. I then got a 6-long streak of 1s which is longer than expected but is much more likely than you think.

My first example with 10 0s and 10 1s (made up, not randomly produced) never even included four 0s in a row or four 1s in a row. That would be a very rare event and together with the 10-10 distribution a very suspicious result! In 82% of all 20-series either the heads or tails will lead.

Summary: Random results produce longer series of same results and more "weird" patterns as we humans are willing to accept. Furthermore, If I continue my experiment, my next 20 throws will *not* need to produce more 0s than 1s because I had 14 1s vs. 6 0s. Why should it?

[Ceaseless]

I think I need to be more especific. Two players or more play a game involving luck. The match contain several rounds. Each round finishes when both have drawn a card, both have rolled their dice, taken a tile, whatever mechanism is involved.
How come and so often (data is required) during the same game one player gets more doubles, strongest cards, best tiles, over and over again or I don't know 8/10 times while the other nothing.

Yes, data is required. Because it turns out, a lot of the people the people claiming this stuff are just making things up. It's not even a matter of them lying or not, some people get so caught up in the moment that they completely lose sight of the bigger picture.

Really? So how long do I have to wait before I get a sharp upturn??? It's so aggravating after having hovered around 300 ELO since I got there, albeit for only about four months since I joined in Jan., and now I'm around 165 and can barely ever win. It doesn't help that the game is often pitting me against lower Arena levels so I take a big hit on a loss and a low return on a win.

Statistically I would think I would have already started winning again. This is insane!

Had to make some edits since I initially posted the wrong user here. https://boardgamearena.com/gamestats?pl ... finished=0 You said the dice were failing you as of the last week (I even stretched it a bit and gave you from May 20 to May 28), where you've only played 20 games, 1 of which was unrated. Of those 19 rated games you only won 7. A rough stretch where you fell from the mid 200s to the 190s, but not as severe as you're saying. And with that laughable sample size, hardly anything noteworthy can be pulled from it.

You want to see what real ranges look like? I took a look at the profiles of the top 10 at the time of this post and posted the game history of the accounts that had over 1000 games below. Some rise and falls are a mere 100-200 points, and in larger cases you'll even see players that have hit 500 eventually drop to 200. If the links give you trouble just check the profiles for the usernames posted next to them.

https://boardgamearena.com/playerstat?i ... 76&game=53 stenkoll
https://boardgamearena.com/playerstat?i ... 03&game=53 Daromar
https://boardgamearena.com/playerstat?i ... 42&game=53 domnik3
https://boardgamearena.com/playerstat?i ... 35&game=53 VIN JAUNE
https://boardgamearena.com/playerstat?i ... 12&game=53 Manisha

[Nunzio Buonaventura]

People often reduce “luck” to obvious random events, such as rolling doubles. That is too simplistic.

Luck is not only what outcome you get. It is also when you get it.

Getting the perfect roll, card, tile, or draw at the most critical moment can change the whole game. This kind of timing luck seems to be real here, and it happens conveniently often. Yet whenever the topic is discussed, many players focus only on the most visible examples and ignore the strategic context in which randomness occurs.

The problem is that most answers in these discussions are not analytical. They are usually based on intuition, personal experience, or the same repeated clichés. Very few people present actual data from BGA games (including myself).

That is why I think the more important question is: Why does BGA not make game data easier to access? This debate would also become less speculative if BGA provided more transparency about each game’s implementation: for example, read-only access to non-sensitive rule logic, or at least public documentation of randomization methods, edge-case handling, and rule interpretations. Full code access may raise legal, security, and licensing issues, but more transparency would still make the discussion more grounded.

If players could export or analyze game histories more easily, we could study luck seriously: first-player advantage, decisive random events, rating gaps, comeback patterns, variance across games, and the relationship between skill and randomness.

Without data, this debate will keep going in circles. Everyone will keep defending their intuitions, but nobody will actually measure anything.

[Romain672]

Something you should do then is before any roll is done, decide yourself if the next roll will count or not.
If yes, you add it on per example excel.
If not, you don't count it, and ask the same question next time.

If you don't cheat, and do that 100 times, and if the effect you describe exist, you should start to see patterns.
And if that's the case, you would have much more credits, especially if you can point exactly some situations where you start counting rolls.


And before doing all of that, try to write what you expect at the end. Per example if you think that the second half of the game of a player with a pseudo starting with B roll more 6 in average, you can check that with what I proposed.
You can add another column 'prediction' if you need it, like you are saying that the next roll will be 9, and then write if you are successful or not.

The thing is, getting the perfect roll/card/tile/draw should happen all the time, you have so many ways to have a luck starting. And so when that happen, you will remember it.
And if you start doing what I proposed, then you can't start at the start of a luck streak, since you don't know when the streak will start. So instead you could start after a player got a streak of 2. But after that, you should see that most of the time that streak stop there, and if you wasn't writing that you would forget it. And in rare cases the streak will goes on to 4/5. But then it only 'count for 2/3' (since the starting streak of 2 doesn't count).


And people have tried to mesure it by the past, but it rarely show anything weird. (and when it show something weird, you can be sure that there will be lots of investigations after that, per example once an admin answered, I don't know everything, but there was something weird with cancelling moves i believe (not fully sure of that)).

[BarnardsStar]

Getting the perfect roll, card, tile, or draw at the most critical moment can change the whole game. This kind of timing luck seems to be real here, and it happens conveniently often. Yet whenever the topic is discussed, many players focus only on the most visible examples and ignore the strategic context in which randomness occurs.

I did attempt to address it here and just got an “ok bud” from the OP. Backgammon would be easier than most games to rig but there would still need to be analysis of the game state in order to do so. And why would anyone want to do that?

This debate would also become less speculative if BGA provided more transparency about each game’s implementation: for example, read-only access to non-sensitive rule logic, or at least public documentation of randomization methods, edge-case handling, and rule interpretations. Full code access may raise legal, security, and licensing issues, but more transparency would still make the discussion more grounded.

Well, have you gotten yourself a BGA Studio account and gotten access to this game’s code? Anyone can do that. And documentation of the main randomization method, bga_rand() is also available.

[Ceaseless]

To be sure of hitting the target, shoot first, and call whatever you hit the target

You can make up an impressive amount of stuff with "data" when you only cherry-pick the results at the end. That's why it's so important that if you're going to claim something weird is happening to be able to make a model that can reliably predict future strangeness.