Deja vu #2.CaractacusPots wrote: ↑06 September 2020, 00:34 lol
and yet another game , my opening throws . . .
14
62
62
62
Another 1 in 5832 chance
Still think the RNG isn't off??
Yes you are quite right but that being the case this kind of stuff should only happen on average once in 324 times, more realistically a few in 324 times for natural spread.RicardoRix wrote: ↑06 September 2020, 11:02 Your numbers are off.
If you want the same combination of 3 throws, then the 1st throw is irrelevant and it's only the 2nd and 3rd throw that should match the 1st.
So for your 3 6-2, it's only 18 * 18 = 1 in 324.
Yes I agree no question. Unfortunately there doesn't appear to be any way to get hold of the collective game histories for any and all players in any practical sense. I'd be very happy to churn the data if someone provided say a csv file of the dice throws of every BG game played say this year.RicardoRix wrote: ↑06 September 2020, 11:02 If you really want to test the RNG then you're going to have to get a far bigger sample size for analysis. You're never going to convince anyone with hand picked 'snippets'.
The human mind is very bad at determining randomness. Your own admission that they only seem to be off confirms that no statistical analysis has been done. And if you agree that single snippets prove nothing, please stop posting them. They do nothing to help your cause.CaractacusPots wrote: ↑06 September 2020, 14:45 However in less than probably 5 or 6 games I seem to have had 2 such events and a game with 18 doubles and so on.
Just seems that these expected rare events are happening far more frequently than should happen.
Yes I agree no question.RicardoRix wrote: ↑06 September 2020, 11:02 If you really want to test the RNG then you're going to have to get a far bigger sample size for analysis. You're never going to convince anyone with hand picked 'snippets'.
Backgammon is one of the easier games to write a scraper for. It actually lists every die roll in the log, so all you have to do is load a game, select replay (which will display the log when asking which player's perspective you want to choose), and parse that page. While many of us will be happy to tell you how to do the work, it would be unreasonable to expect us to actually do the work.CaractacusPots wrote: ↑06 September 2020, 14:45 Unfortunately there doesn't appear to be any way to get hold of the collective game histories for any and all players in any practical sense.
Your maths is off because you expect 16.6% doubles every game. Random doesn't work that way.CaractacusPots wrote: ↑06 September 2020, 18:52 Ok happy to get more serious on this and look at data
I've waded through my game histories to look at the number of DOUBLES occurring against the number of dice throws in the game
Here are the first 22 games that weren't conceded by either player
Game No---No Of---Total---%
----------- Doubles -Throws
#109661286----12----50 (24%)
#109671565----4----- 44 (9%)
#109671796----8----- 55 (14.5%)
#109677539----10----41 (24%)
#109696866----9----- 35 (25%)
#109826928----3----- 42 (7%)
#109840092----12----41 (29%)
#109849486----8-----36 (22%)
#109844128----9-----64 (14%)
#109856820----5-----43 (11%)
#109859222----7-----61 (11%)
#109851074----8-----50 (16%)
#109969081----4-----46 (8.6%)
#109963403----4-----53 (7.5%)
#109968324----9-----75 (12%)
#109965336----5-----57 (8.7%)
#109963218----9-----43 (21%)
#109981226----9-----30 (30%)
#110013686----6-----71 (8.4%)
#110012369----18----69 (25%)
#110129592----5-----62 (8%)
#110129835----10----45 (22%)
There are 6 doubles among 36 possible throws so doubles should occur 1 in 6 throws on average or 16.6%
As can be seen above the number generation is nowhere near this. It is consistently either way above at 22%-30% or way below at 7%-11%
Of the 22 games above only 4 of them have doubles happening close to the expected 16% (allowing 12%-20%)
41% of games (9) have way under the number of doubles expected
41% of games (9) have way over the number of doubles expected
18% of games (4)have the correct average number of doubles
You may say 22 games is a small sample but frankly I have every confidence that if I continue looking at the next 20-30 games this pattern will continue.
Or maybe my maths is off?