My MG classic blackjack sessions...

[NewOrleans]

I read a crapload of posts about microgaming classic blackjack....
here is my last session
Session 1 -150$ bankroll
bet 2 win
bet 5 lose
bet 2 win
bet 5 lose
bet 3lose
bet2 win! lol
bet 5 LOSE!
bet 25 lose dealer blackjack
bet 50 lose dealer blackjack
bet 75 lose dealer blackjack
Session 2 - 50$ bankroll
bet 2 win
bet 5 lose
bet 2 win
bet 5 lose
bet 10 lose dealer blackjack
bet 20 lose dealer blackjack
bet 2 win
bet 2 win
bet 3 lose
bet 5 dealer blackjack

LOL no more microgaming classic blackjack

[vinylweatherman]

I read a crapload of posts about microgaming classic blackjack....
here is my last session
Session 1 -150$ bankroll
bet 2 win
bet 5 lose
bet 2 win
bet 5 lose
bet 3lose
bet2 win! lol
bet 5 LOSE!
bet 25 lose dealer blackjack
bet 50 lose dealer blackjack
bet 75 lose dealer blackjack
Session 2 - 50$ bankroll
bet 2 win
bet 5 lose
bet 2 win
bet 5 lose
bet 10 lose dealer blackjack
bet 20 lose dealer blackjack
bet 2 win
bet 2 win
bet 3 lose
bet 5 dealer blackjack

LOL no more microgaming classic blackjack

MG Blackjack is high variance, even at $1 chips, $50 is not enough to keep a session going long enough for a fair sample (unless you hit an upsurge early).

Ideally, you should use Autoplay, and low bets, but a larger starting bankroll. You can program in the number of hands, and gather the stats at the end. Try 5000 hand sessions, and then cut & paste the stats, and then do another 5000. Note your balance each time. You can do this in fun mode (for fun), or real mode if you are prepared to lose money.

You can build up a long term sample of 1000's of hands this way.

[tombee]

However.. this seems VERY unbelievable

[Jufo]

MG Blackjack is high variance, even at $1 chips, $50 is not enough to keep a session going long enough for a fair sample (unless you hit an upsurge early).

The standard deviation of MG Classic BJ is close to 1. If the game really behaves streaky (wins and losses are more clustered together than what normal randomness would dictate) then such game cannot be random if it doesn't behave exactly the same way as a physical shuffled deck of cards would. In other words excessive streakiness can imply game being rigged even if the long term house edge is unaltered.

[vinylweatherman]

The standard deviation of MG Classic BJ is close to 1. If the game really behaves streaky (wins and losses are more clustered together than what normal randomness would dictate) then such game cannot be random if it doesn't behave exactly the same way as a physical shuffled deck of cards would. In other words excessive streakiness can imply game being rigged even if the long term house edge is unaltered.

MG Blackjack, of whatever flavour, is always streaky.

I did an interesting experiment at Spin Palace in 2005, playing Vegas Strip on autoplay at £1 chips. I ran a total of 80,000 hands, and from a starting balance of £3000 I ended up on £3844 - a profit of 844 units on 80,000 hands. I thought this quite exceptional, albeit unusually for MG, in the right direction

It seems I had hit some kind of "sweet spot", so I decided to try £10 chips instead of £1, and lost the entire £844 in less than half an hour.

Had I been playing at max bet £500, and achieved such a run - quite possible presumably, because it happened at £1, and MG casinos tell us the games are random - I would have won some £422,000 (844 betting units) - and quite possibly bankrupted the casino. This is what makes me think the games are not 100% random, and that factors such as bet size somehow make the games play a little differently.

Maybe I should have pressed on with the experiment, going far beyond 80,000 hands to see if the trend continued, however, I took the chance that I had somehow got the software into some kind of mode, and risked my profit of £844 on the hunch. I preserved my initial £3000 of course.

[aka23]

The original post contains 20 hands. This is not large enough a sample size to prove much of anything.

However, VW's data is more interesting. A gain of 844 over 80,000 hands on VS BJ is about a +3.5 SD win. Again not rare enough to prove anything, but it would be interesting to see if anyone could repeat an unlikely win on 1-unit bet autoplay.

[tombee]

I agree that the sample is too small to prove anything about randomness of winning vs losing hands, but isn't the chance very small anyway that you lose every hand with a bigger bet, and win every hand with the smallest bet.. even losing EVERY bet over 10 due to a BJ of the dealer. Come on.. that chance must be very small even though this is a small sample

[pubjoe]

Unlucky.

[Jufo]

I agree that the sample is too small to prove anything about randomness of winning vs losing hands, but isn't the chance very small anyway that you lose every hand with a bigger bet, and win every hand with the smallest bet.. even losing EVERY bet over 10 due to a BJ of the dealer. Come on.. that chance must be very small even though this is a small sample

The player made 9 "small bets" and 11 "big bets". He lost all the "big" bets. The probability of 11 losses in a row (taking account pushes) is:

0.47882^11 = 1 in 3297,

so yes it is unlikely.

[winbig]

The player made 9 "small bets" and 11 "big bets". He lost all the "big" bets. The probability of 11 losses in a row (taking account pushes) is:

0.47882^11 = 1 in 3297,

so yes it is unlikely.

This is a prime example of what my old stats teacher was trying to convey to our class.

You can construe the data any way you want to make it 'say' anything you want.

Moral of the story: They didn't lose 11 hands in a row, they lost 11 big bets. The longest losing streak (from the information provided) was 4 hands.

Don't twist the data around...

Side note:

With these rules (1 deck, no peek, double on 9-11, split only once, able to hit aces after splitting them, stand on all 17's), the wizard's calculator - http://wizardofodds.com/blackjack/ho...alculator.html shows an optimal return of -0.0307%, and a realistic return of 0.121%; a far cry from the 0.47882 you use...