DD Trouble

Just looking for insight on this and advice. I have been using hi-lo for a while now with good success overall. Some ups and downs but can t complain, I have been playing 6 or 8 deck games with das,splt to 4, h17 ns, d any 2 cards and 85% pen. My last two trips I have played some DD with the same rules and 65% pen, and have been getting hammered. I spread 10-80, and on a 5 dollar table 5-80. Am I doing something wrong or is this just variance?

Variance.

Quote: Ibeatyouraces

Variance.

^^^^^^^^^^^

Quote: Ibeatyouraces

Variance.

I keep hearing about variance in blackjack and im like huh?

thought the variance in blackjack using spread is like 4?
and I heard a good rule of thumb for bankroll is 20x your bet.
so $10 bet = $200 bankroll

to compare, a low variance vp game (ie: 9/6 JoB) is 20.
10000 hands at $2 denom ($10 per spin) is about +/- $4800.
(aka DiaD on a '$20 per tc' machine.)

guess swings in black jack are not as violent?

That is what I thought, seems tho in DD you only get one or two hands at a good count and you hit the cut card, with 3 or 4 players, also seem to loose a lot of insurance bets, maybe just me

Quote: 100xOdds

I keep hearing about variance in blackjack and im like huh?

thought the variance in blackjack using spread is like 4?
and I heard a good rule of thumb for bankroll is 20x your bet.
so $10 bet = $200 bankroll

to compare, a low variance vp game (ie: 9/6 JoB) is 20.
10000 hands at $2 denom ($10 per spin) is about +/- $4800.
(aka DiaD on a '$20 per tc' machine.)

guess swings in black jack are not as violent?

Can't really make an arms length comparison between vp and blackjack volatility in terms of just the variance number, vp can generally play 10-15x as many hands per hour than a blackjack player (maybe 12x avg?).

So if the Vp variance number is 15x higher than bj, but can play 12x as many hands/hr, the swings of the two can/should be pretty comparable if the advantage/disadvantage is also similar, say 1%.

Quote: dwight56

That is what I thought, seems tho in DD you only get one or two hands at a good count and you hit the cut card, with 3 or 4 players, also seem to loose a lot of insurance bets, maybe just me

It's variance.

1.15 is the 'standard' Standard Deviation (sqrt of variance) in BJ... of course pending rules, etc.

You should do the following:

AvgBet = X
AvgEdge = 1%
OriginalSD = 1.15*X

EV(generic) = (NumHands*AvgBet)*(AvgEdge)
SD(generic) = Sqrt(NumHands)*OriginalSD

So let's look at your example (I'm estimating spread/etc for $10-$80) and have some other assumptions (wong out at < TC -2, max bet out at TC +4, etc, etc, etc)

AvgBet = ~$22
AvgEdge = 1%
OriginalSD = 1.15*22 = 25.3

*Mathematical Disclaimer: 1.15 is a given standard in the community. For more precise numbers to your EXACT game (with rules, PEN, spread, wonging, etc) you'd need to SIM the results... However, this should be "fairly accurate" for a quick check and my point below.

First, you've only played 2 trips (SUPER SMALL SAMPLING SIZE). Secondly, let's again assume you played for something like 2 hour each trip getting 80 hands per hour (since you said there were others at the table). That would give us a total of 4 hours, and thus 320 hands.

EV(320 hands) = (320*22)*(.01) = $70.40
SD(320 hands) = Sqrt(320) * 25.3 = $452.58 ...we want 2SD though for 95% confidence in our results (1SD is 68%, and 3SD is 99%). 2SD = $905.16

So what does this mean? Well in your SHORT SAMPLING SIZE you could expect to make $70.40 +/- $905.16... ANYTHING in that range either plus or minus is just "variance" from expectation. So you could be DOWN $800 and that would be perfectly normal. You could also be UP $900 and that would be perfectly normal. The key to blackjack is increasing your sampling size in order to reduce variance and get more accurately to your expectation. I'll leave it to you as an exercise to do the following:

1) What's your EV and 2SD for 10 trips, 2 hours each, at 80 hands per hour?

2) What's your EV for 10,000 trips, 2 hours each, at 80 hands per hour?

3) What do you notice about your EV +/- SD as you increase the number of hands in each example?

I'd suggest taking a peek at the 3 articles I wrote and discuss a lot of this in that are posted here in the Articles section of the WoV site:
https://wizardofvegas.com/articles/A-to-Z-Counting-Cards-in-Blackjack/
https://wizardofvegas.com/articles/A-to-Z-Counting-Cards-In-Blackjack-2/
https://wizardofvegas.com/articles/A-to-Z-Counting-Cards-in-Blackjack-3/

It's just variance. Also the 6 to 8 decks games are not great with those rules. The DD is very good if it's 65% pen (perhaps too good?)
Romes I've been thinking about the way everyone calculates variance using the average bet, isn't it a gross underestimation of the actual variance because of the spread?
For example an extreme case is:
1000 hands at $10 and 1 hand at $10020
=> avg bet =$20
Variance = 1.15*20*sqrt(1001) = $728
But we all know the actual variance is a lot more.

Quote: tyler498

...Romes I've been thinking about the way everyone calculates variance using the average bet, isn't it a gross underestimation of the actual variance because of the spread?

Well, yes and no. Yes in the sense that your example surely identifies an issue. No in the sense that your example would literally never happen in a real life situation for someone card counting properly. The average bet based off spread. A better mouse trap would be to weight your bets with the frequencies of actually betting that amount and then get an average bet. Basically, my disclaimer that a simulation will give much better results is the case. HOWEVER, if one isn't betting $1, $1, $1, $10 million... the above formulas can certainly be used for a fair estimation and be done much quicker/simpler. In the time it would take you to just get TO a site to download simulation software (not even download it) I could crank out several EV's and SD's.

For any counter with a good spread it should do just fine (again my assumptions above were $10, $20, $40, $60, $80) for TC = 0 through TC +5, with wonging, etc. I've run my spreadsheet calculations and these formulas against simulations and the're quite on par, except for more extreme scenarios (such as 90% PEN, or a WILD betting spread of like 1-50). I would say 90%+ of the time I've double checked the mathematical formulas above with a SIM they've been rather close (and usually "underestimating" a bit - which I prefer to do with all my numbers). Basically if my EV comes out to be like $46.75, the SIM EV is usually about $50. The PEN makes the difference (both ways). So I'd sum all this word garbage up in 2 sentences:

1) For the most accurate results possible you should absolutely SIM your situation. For anyone who wants to take the game quite seriously and make any kind of serious money at it, I'd recommend getting the simulation software.

2) For the do it yourself-ers who enjoy working through the math and are okay with a "good" estimation that's quick and dirty, these will do just fine. Aka I saw this thread and was able to put that post together in 5 minutes... and unlike a SIM where you plug and chug I feel the formulas (combined with the spreadsheets, frequencies, etc) give one a deeper understanding of how all of the math pieces to the puzzle go together in the game.

Quote: Romes

Well, yes and no. Yes in the sense that your example surely identifies an issue. No in the sense that your example would literally never happen in a real life situation for someone card counting properly. The average bet based off spread. A better mouse trap would be to weight your bets with the frequencies of actually betting that amount and then get an average bet. Basically, my disclaimer that a simulation will give much better results is the case. HOWEVER, if one isn't betting $1, $1, $1, $10 million... the above formulas can certainly be used for a fair estimation and be done much quicker/simpler. In the time it would take you to just get TO a site to download simulation software (not even download it) I could crank out several EV's and SD's.

For any counter with a good spread it should do just fine (again my assumptions above were $10, $20, $40, $60, $80) for TC = 0 through TC +5, with wonging, etc. I've run my spreadsheet calculations and these formulas against simulations and the're quite on par, except for more extreme scenarios (such as 90% PEN, or a WILD betting spread of like 1-50). I would say 90%+ of the time I've double checked the mathematical formulas above with a SIM they've been rather close (and usually "underestimating" a bit - which I prefer to do with all my numbers). Basically if my EV comes out to be like $46.75, the SIM EV is usually about $50. The PEN makes the difference (both ways). So I'd sum all this word garbage up in 2 sentences:

1) For the most accurate results possible you should absolutely SIM your situation. For anyone who wants to take the game quite seriously and make any kind of serious money at it, I'd recommend getting the simulation software.

2) For the do it yourself-ers who enjoy working through the math and are okay with a "good" estimation that's quick and dirty, these will do just fine. Aka I saw this thread and was able to put that post together in 5 minutes... and unlike a SIM where you plug and chug I feel the formulas (combined with the spreadsheets, frequencies, etc) give one a deeper understanding of how all of the math pieces to the puzzle go together in the game.

You are right it is actually an underestimation.

I just did some calculations based on your spreadsheet for the ratio of bets at each count, and used the sum of variance formula, please let me know if it's correct:
for example, the max bet of 150 is bet on 5.44 hands out of 84.55 hands played, so 0.064*n hands.

( https://docs.google.com/spreadsheets/d/1-jCr5jRUiAg_s59IFTz6B2wqKyDJmyQkiSlddT7RdNc/edit#gid=206313892 )
Approximation: SD = 1.15*sqrt(n)*30.27 where 30.27 = avgbet
Edit; had a mistake ^_^"
Exact: SD = 1.15*sqrt(n)*sqrt(0.064*(150)^2 + 0.047*(110)^2 + 0.077*(60)^2 + 0.136*(30)^2 + 0.675*(10)^2)
SD = 1.15*sqrt(n)*49.76
I am getting a SD that is much higher than the one using the avg.bet, is it supposed to be that different?
Also about your point on making it quick to calculate. A multiplication factor can be calculated for a specific spread to correct the SD calculations from flat betting avg bet to SD with a bet spread for quick calculations.
For instance in my example above it would be: 49.76/30.27 = 1.64
So when using this spread (or a close one) on 6D I should multiply my flat betting SD by 1.64 to get the actual SD, what do you think?

could be variance, could be you are making fundamental mistakes

Care to elaborate? I am not perfect but counting is easy in pitch game as slow as they play. AM I miscalculating the true count? Is it not running count divided by decks remaining DD?

There are any number of mistakes that aspiring counters make. You might be incorrectly estimating deck depth, using incorrect index numbers, applying index numbers incorrectly, not knowing basic strategy perfectly for the game you are playing, not betting correctly etc.

A downswing doesn't really indicate anything one way or the other. Having big (20 max bet) losing sessions is just part of the game, even if you are playing perfectly.

Trying to spread 1-8 with a bankroll that is 20X your minimum bet is an invitation to disaster. Your ROR must be sky high.

Don t know where you got that but my bankroll is much more than that, something like 100 times max bet

Quote: dwight56

Care to elaborate? I am not perfect but counting is easy in pitch game as slow as they play. AM I miscalculating the true count? Is it not running count divided by decks remaining DD?

Everyone learns eventually (usually in their first big losing streak) that you think your game is much better than it really is. When my teammate and I had 300 hours in the red, we re-did everything we could with our game. More testing out, counting drills, more indexes, more revisiting our bet spread to see if we could optimize it better both for our game and our bankroll... etc. It was during this time we saw we had a pretty good game... but after this losing streak and re-evaluating everything about our game, afterwards we could say we had ONE HELL of a polished game. Keeping really good playing records helps SO MUCH with seeing how the numbers are "actually" stacking up as opposed to "expected." A really big eye opener for some... and others can't deal with that and blame "luck" or what have you.

Never take for granted that you've "got it down" and always be trying to learn something new about the game... because trust me, there's always something new to learn.