Wincraps and other Computer RNG vs Real Casino Randies

[heavy]

Wincraps has multiple RNG's built into it. You can switch between programs. Also, you can "skew" the rolls toward specific numbers if you want. Or you can take yoru live casino rolls and enter them into a roll file and use those.

[freak]

I use them all and they seem to be off how close are they to real random rolls. Anyone here have real random rolls, I wish someone would put craps system tester rolls on here, I test the systems there and they work

I use a program called "Aw Craps" on my iPhone. It has a roll tracking / roll statistics feature that allows me to see the history. The stats after my 55,604 rolls so far shows near perfect distribution:



I don't know how the RNG actually works in these programs. It's possible that a program could be a "preset" sequence of random rolls...say 2500 rolls that repeat over and over. That would guarantee a statistically correct result but might allow someone to pick up on the pattern and exploit it over time. I do notice certain things in the Aw Craps program that tend to repeat. A PSO is often followed by a second PSO. Set the point, roll a 4 the next roll is likely to be 10. I don't know if this is actually true, but it seems like these things repeat more than normal on Aw Craps. My "hunches" in Aw Craps are often correct. If I go from Aw Craps to a live table and apply those same hunches they typically fail. More than likely this is just my brain trying to find patterns in the randomness and simply noticing when a pattern appears. If the RNG is actually generating two independent numbers between 1 and 6, then it will fairly quickly settle into the statistical norm. I remember checking this when I only had 200 rolls or so and it was noticeably off. But now it's dead on. The longer you go the closer it gets.

The distribution of live casino rolls that I have recorded over the past year in Roll Tracker isn't perfect but it's close:



2 343 2.75%
3 690 5.53%
4 1030 8.26%
5 1403 11.25%
6 1772 14.21%
7 2162 17.34%
8 1804 14.47%
9 1303 10.45%
10 966 7.75%
11 682 5.47%
12 316 2.53%

Looks like I should probably bet more fives.

[Maddog]

2 343 2.75%
3 690 5.53%
4 1030 8.26%
5 1403 11.25%
6 1772 14.21%
7 2162 17.34%
8 1804 14.47%
9 1303 10.45%
10 966 7.75%
11 682 5.47%
12 316 2.53%

Looks like I should probably bet more fives.

Not unless you can further reduce the seven, or further increase the appearance of the 5. If my math is correct, as it stands you are still negative EV on the 5.

... I think the RNG on the Wincraps game has some kind of counter code to integrily mimic the negative expectation of the game

No, it does not.

...it is too negative and evasive of the bets

an opinion based on what research?

...When I go to a real craps game its alot more forgiving on randies.

Lucky randies. Are real craps games as forgiving on Alberts and Zeldas as well?
------------
http://en.wikipedia.org/wiki/Random_number_generation
http://www.random.org/

[shunkaha]

The long and short of it is, a computer can't generate truly random numbers but then it doesn't exactly pick from a long list that recycles either [Maddog's wiki link explains it in fairly decent detail but suffice it to say they aren't exactly random but they aren't in a pattern you could generally pick up either].

As for random rollers having better or worse results than a RNG... yes, sometimes they will, sometimes they won't. If you watch the same random roller several different times, his/her results will sometimes conform to the mean and sometimes be better or worse. If you have a consistent expectation that a person meets when they toss the dice, by definition they aren't a random roller or they wouldn't meet those expectations with the exception of the mean ratio of rolls to every 7 over time being 6 +/- a tiny fraction.

As for Wincraps, I've had sessions in which I could shred $20k - $70k+ testing certain things just for fun, then I could turn around and do the exact same thing on back to back sessions and turn them into $50k+... just like at the table some nights you could bet whatever you like and win and other nights not so much. Given time Wincraps should replicate both those times that nothing goes right and those times you can do no wrong... you just happen to be in the valley instead of on the peak is all.

[freak]

2 343 2.75%
3 690 5.53%
4 1030 8.26%
5 1403 11.25%
6 1772 14.21%
7 2162 17.34%
8 1804 14.47%
9 1303 10.45%
10 966 7.75%
11 682 5.47%
12 316 2.53%

Looks like I should probably bet more fives.

Not unless you can further reduce the seven, or further increase the appearance of the 5. If my math is correct, as it stands you are still negative EV on the 5.

Ok then, looks like I should lay more nines.

[Mad Professor]

2 343 2.75%
3 690 5.53%
4 1030 8.26%
5 1403 11.25%
6 1772 14.21%
7 2162 17.34%
8 1804 14.47%
9 1303 10.45%
10 966 7.75%
11 682 5.47%
12 316 2.53%

Hi Freak,

I ran your roll-stats through my Veg-a-Matic (otherwise known as the DiceTool Plus component of BoneTracker).

Here is what they showed:

Wager………..….Occurrence.….Appearance-Rate…..Player Edge-per-Bet

Buy-2 …..………..….343 …………….…2.75%.....................-7.87%

Buy-3 …………..…….690 …………….…5.53%.....................-7.00%

Buy-4 ………..…....1030 ……………...8.26%.....................-6.93%

Place-5 …………....1403 ……………..11.25%....................-5.56%

Place-6 ……………..1772 ………….….14.21%....................-2.41%

7 ………………….…..2162 ……………..17.34%

Place-8 …………....1804 ……………..14.47%....................-1.44%

Place-9 …………....1303 ……………..10.45%....................-9.75%

Buy-10 ………..…….966 ……………....7.75%.....................-8.57%

Buy-11 ……………...682 ………………..5.47%.....................-7.77%

Buy-12 ……….......316 ……………....2.53%.....................+6.43%


MP

[believer]

But when I use Wincraps its like the program sniffs out what I am doing and makes me lose, I get all these weird rolls that set me off

I agree. Random.org works better and seems to be far more accurate.