Quantcast

Jump to content


Dreww

Member Since 23 Oct 2005
Offline Last Active Jan 25 2016 09:45 AM

#1934962 Survey of Bagatelle Outcomes

Posted by Dreww on 02 January 2016 - 03:39 AM

Well then, I'm only about four years late with this reply, but here it is!

 

rbFVrdj.png

 

From the 5488 rolls collected, the breakdown was as follows:

   1    2    3    4    5    6    7    8    9   10   11   12   13   14   15   16
2832 1332  766  269  179   67   25   10    4    1    2    1    0    0    0    0

No rolls above 13 occurred, which is to be expected when dealing with a sample size this small. This resulted in the following actual percentage, which I'll use for extrapolating the hypothetical probabilities:

       1        2        3        4        5        6        7        8        9       10       11       12 
51.6035% 24.2711% 13.9577%  4.9016%  3.2617%  1.2208%  0.4555%  0.1822%  0.0729%  0.0182%  0.0364%  0.0182%

Eyeballing this, the game likely uses the simplest infinite series possible.  The odds we're looking for 2-n would be:

       1        2        3        4        5        6        7        8
50.0000% 25.0000% 12.5000%  6.2500%  3.1250%  1.5625%  0.7813%  0.3906%
       9       10       11       12       13       14       15       16
 0.1953%  0.0977%  0.0488%  0.0244%  0.0122%  0.0061%  0.0031%  0.0015%

Our deviations from the pattern are all due to sampling error, which is a byproduct of using a finite sized dataset.  Emulating these results have a wide range of outcomes that show how much variation occurs purely due to chance:

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   14 
2777 1363  652  345  180   90   42   23   10    4    1    1 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   12   13   14 
2764 1346  671  349  179   85   51   16   12    6    4    2    2    1 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   12   13 
2723 1398  689  340  159   84   46   19   16    9    2    1    2 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   12   13   14   16 
2767 1374  656  340  167   92   43   17   15    8    2    3    1    2    1 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   12 
2752 1356  711  338  172   77   41   24   13    1    1    2 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   12   14 
2807 1299  710  318  178   94   42   10   15    8    2    4    1 

> table(sample(1:16,5488,replace=T,prob=prob))
   1    2    3    4    5    6    7    8    9   10   11   13 
2766 1343  694  329  194   83   46   17   10    3    2    1 

Chi-Square Test:

> prob<-2^(-(1:16))
> chisq.test(tab/length(cc),prob)


        Pearson's Chi-squared test


data:  tab/length(cc) and prob
X-squared = 176, df = 165, p-value = 0.2646


Warning message:
In chisq.test(tab/length(cc), prob) :
  Chi-squared approximation may be incorrect

A p-value lower than a cut-off of 1 standard deviation (p=0.05) would suggest the game does not follow this payout pattern.

 

If you paid 250 million to play Bagatelle one million times, and each jackpot ambitiously gave you 500,000 NP, you would only win back between 160 and 165 million NP.  With a jackpot payout of 1,000,000 NP, this goes up to a whopping 161 to 169 million NP.

 

TL;DR : Bagatelle is a pretty predictable game when you play it enough, and pays out 2/3 of the money you put in.

 

 

 




#1422835 Survey of Bagatelle Outcomes

Posted by Dreww on 09 March 2011 - 12:51 PM

Test for profitability in longevity of Bagatelle

The game interested me because it resembled the Plinko board from The Price Is Right. The Plinko board follows a binomial distribution of each outcome defined by 1/(2^{n-1}) where n is the number of peg-rows on the board (in the case of the Bagatelle board, this is 5 rows). This distribution of P(5) results in a 1/16th chance of any ending. The second distribution I am testing for is a bastardization of an infinite series lim n->16 of 1/(2^n) where n is one of the 16 outcomes. Following this, each reward (#1 50%, #2 25% etc.) would progressively be half the previous until 16, where to allow for 100% distribution I permitted 15 and 16 to have the same odds. Getting to the point, I am creating a data set of bagatelle outcomes to test both of the distributions I mentioned and to formulate any other potential ones. My two goals are to a) create a reliable dataset that can be simulated with low margins of error; and b) to use the sample data itself to see if profit occurs.

Wall of text aside, all you'd have to do...
Would be provide me with data! Bagatelle allows for 20 rolls a day costing 250np each (5,000NP total). If willing to contribute, post the daily outcome of your 20 rolls (in a single post!) with the # (1 through 16) and if you are lucky enough to get 12+, the item/jackpot value.

DATA:
First evaluation will come at 1000 entries, or if people care enough to post their rolls at all.

Simplified playing:
Courtesy of the amazing Mr Kway, this link will process the game for you! Feel free to copy and paste this outcome directly or post just the results if you use it.

To interpret the results, look for the 4-letter drop result that looks like LLLL - RRRR. These correspond to the following:
'

  • RRRR
  • RRRL
  • RRLR
  • RRLL
  • RLRR
  • RLRL
  • RLLR
  • RLLL
  • LRRR
  • LRRL
  • LRLR
  • LRLL
  • LLRR
  • LLRL
  • LLLR
  • LLLL

Status: Analysis on post #299




#1407734 Does anyone actually HAVE a Super Attack Pea?

Posted by Dreww on 11 February 2011 - 03:49 PM

>SD was revamped
>Everyone bitched and moaned about the revamp
>Admins say "Fine. See how you like it now" and close the site.

I was no help with that. All I did post-the "db/backup crash" was bitch about the flamboyantly orange+blue layout and logo. The least he could've done was let us use Loki's old skins, but he removed the skin options (including default), which really ticked me off. Considering he was making a nice profit on premium up until the end, my lasting guess was more legal action taken against him to the extent that costs outweighed all profit.

EDIT: Posted Image


#1403185 Funny Screenies.

Posted by Dreww on 04 February 2011 - 08:21 PM

Posted Image

Had just gotten it in a trade and was two clicks from putting it in my SDB. But to put that in perspective:

Posted Image


#1386194 HPD Results?

Posted by Dreww on 05 January 2011 - 03:46 PM

OH SERIOUSLY?
I never knew that. I always though restocking amounts were set.
Never knew they got affected by amount of restockers. That's pretty interesting and i'll pay attention next HPD

Shops have four rows of items, they can't have any more than that. If a shop stocks during this, you won't even notice because the only things that can come in are what is already stocked (the junk items). So the more spaces cleared, the more that can potentially stock. Major/minor restocks are noticeable on HPD because there will be some that drop 2-3 new items, and others that drop 10+ new items.