Understanding SD

[Dopple]

I am getting interested in SD and wonder if a better working knowledge thereof enhances ones understanding of the swings of the game.

I track my hours and win/loss but not my average bet and or hands per hour.

A chart of what the average bet would be based on one betting a certain ramp would be interesting.

I suppose if you had an good feel for about how far below your EV in total winnings you were you could take some solice in the fact that you should once again rise up to that average again to tetter above and below that value forever.

Maybe there is some simple rule of thumb for figuring your action for an hour of play. You could work off a seperate stack to guage your action for a one hour session but again that damn radar would nail you as an AP I suppose.

 

[FLASH1296]

Average wager / Hands per hour.

Firstly, I am going to assume that you know that most Single Deck games these days pay 6 to 5 on a BJ and must be avoided.

The Single Deck games of Northern Nevada have poor rules but typically have a marginally acceptable House Advantage.

In those games the penetration is paramount.

As to your question re: game speed.

To determine how many hands per hour you are playing - what you need to know is how many players there are at the table (on average), how many decks are in play, and if auto-shufflers are in use. If not is the shuffle relatively simple or is it a complex 6 zone shuffle. Are the dealers absurdly slow, like in the Bahamas, or are they rather fast, as in Las Vegas? Are you "wonging-out" or using a play-all approach. Are there time-consuming side-bets that need to be continuously resolved ?

You are going to play somewhere between 50 and 200 hands per hour.; but the vast majority of BJ games will be played with a half full table dealing at a rate between 70 and 100 hands per hour. Heads up play can double that rate and a full table of amateurs will really bog things down.

Overall, most of us use a ballpark figure of 100 hands per hour.

As to your question re: your average bet ...

that can be computed easily with nothing more than a sharp pencil.
It is a "weighted average" based upon your bet at each True Count and the True Count distribution.

Presuming that your betting "ramp" or "spread" was computer generated and is nearly "optimal" you will find that your AVERAGE bet will generally be about 2 units and will only approach 3 units if your "betting ramp" very aggressively.

I have found that the average bet will most typically be 2.1 - 2.4 units.

So ... You may be betting around 100 hands per hour and averaging a example little over 2 units per hand. Precision here is unwarranted.

200 units per hour is 1,000 units per 5 hours of play.

Incidentally, your objective edge is most likely well under 1% (including errors, not splitting tens, etc. etc.) and in the neighborhood of .75%

So ... if you play 5 hrs. you will have played 1,000 units and your projected average profits at .75% would be approx. 7.5 units.
If your betting "unit" is $15 (a common table minimum at many venues) your expectation for the evening's play is a profit of
$15 x 7.5 = $112.50 on the total "handle" (money placed at risk) of $15,000.

As a beginner, note how much money had to be risked in order to 'expect' a rather modest profit.
This slim edge disappears if your Running Count is incorrect, or your True Count computation is inaccurate, or your indexed plays are done sloppily, or your bet sizing is "off" betting either too much ~ or ~ too little at any True Count. making this even more "dicey" is that you may be using a count ill-suited to the game at hand, like using "Hi-Lo" in a single deck game.

 

[fwb]

You can actually find some great single deck in certain places, like Colorado, with "normal" rules (3:2 bj ect). But be careful about sharing your results here, or else a mod will label you as a voodoo better and move your topic :joker:



The dealers say that the house doesn't make any money off it but it's used to draw people in.

 

[Dopple]

This is excellent information in a very easily digestable format. I sincerely appreciate your generous response to my query. I am a glutton for the analytical end of this game and take as much joy in the study of my longterm outcome as I do in play.

Thanks again Flash.

So if Charlie plays $20 for 20hrs his handle is 20x20x200=80k EV=80x.75= $600

His SD = 1.1/SQRT 4000 = 1.1/63.24 = 1.74% x 80K = 1,392

If Charlie was down $792 or up $1,992 he would be within one SD of his EV which is normal about 68% of the time and should take solice in the fact that should he remain with one SD of his EV over an extended period of time he is in fact proving that this is his true EV. He will in fact stray from this range periodically hower.

 

[zengrifter]

Are there ANY decent 1D games left in LV? zg

 

[StandardDeviant]

...and should take solice in the fact that should he remain with one SD of his EV over an extended period of time ... He will in fact stray from this range periodically hower.

He will be within 1 SD of his EV about 68% of the time. He can also expect to do worse then his EV about half the time (and better half of the time).

 

[Kasi]

I am getting interested in SD and wonder if a better working knowledge thereof enhances ones understanding of the swings of the game.

I track my hours and win/loss but not my average bet and or hands per hour.

A chart of what the average bet would be based on one betting a certain ramp would be interesting.....

I am a glutton for the analytical end of this game and take as much joy in the study of my longterm outcome as I do in play.
Thanks again Flash.
So if Charlie plays $20 for 20hrs his handle is 20x20x200=80k EV=80x.75= $600
His SD = 1.1/SQRT 4000 = 1.1/63.24 = 1.74% x 80K = 1,392

Seriously, you need to buy a sim if you are serious about analyzing results. There is no other way. A sim will tell you your avg bet etc.

Flash's pencil is not sharp enough to give you info like freq by TC, etc. Neither is mine.

Charlie, assuming in your example you are assuming, for some reason, Flash's 2-units avg bet/rd, is way off on both EV and SD.

Flash assumed 1000 "units" (ie avg units/rd) bet in 5 hrs. So, with a $15 unit, $15K bet in 500 rounds. Or, $30 avg bet/rd for 500 rounds = $15K. The win rate of,say, 0.75% is, typically, based on avg bet/rd, not on min unit used.

Hopefully you can relate this to Charlie.

Also, Flash neglected to mention how SD can be 4 or more units per round. Your "1.1" seems to indicate flat-betting which a counter obviously doesn't do. Your "4000" seems to be rounds played? - obviously Charlie has only played 2000 rounds in 20 hours?

Only a sim can tell you stuff like "avg bet/rd" and "SD/rd" after you tell it what game and count and pen counting system w or w/o indexes, back-counting, wonging-out, playing-all etc you are playing against.

The exact numbers can vary greatly.

There is no "generic" set of assumptions to use for all games. Hope you come to understand that.

 

[UK-21]

Seriously, you need to buy a sim if you are serious about analyzing results. There is no other way. A sim will tell you your avg bet etc.

Flash's pencil is not sharp enough to give you info like freq by TC, etc. Neither is mine.

A sim will, indeed, provide an accurate picture of the longer term results of particular games played with particular strategies - and as such a comparison can be made between them. So which is better - a uniform approach to playing 6D, H17, DOA, DAS, NS or UK ENHC, SD, S17, DOA, DAS, NS where you can only insure a natural (this is the most commonly found ruleset here in the UK) ? Only simming these options over say, a billion hands, will provide a reasonably precise answer.

As to stuff like freq by TC etc, a lot of this can be gleaned from sources online and in the noted publications on the game such as Schlesinger's BJA. With this you can build spreadsheets that, although not accurate to the nth degree, will provide a pretty fair indication of things. If you use these figures and apply your betting ramp and unit value to them you will be able to calculate your average bet over, say, 1,000 hands, 10,000 hands or whatever.

http://www.blackjackforumonline.com/w-agora/view.php?bn=bjf_forum&key=1242052399&pattern=billion

The figures on the end of the link don't tally to 100% (due to rounding to two decimal places I suspect), so I adjusted mine slightly so my calcs would reconcile out. Here are my adjusted figures:

TC Freq

-7 0.61%
-6 1.12%
-5 2.04%
-4 3.69%
-3 6.61%
-2 12.19%
-1 18.17%
0 29.43%
+1 11.82%
+2 6.43%
+3 3.55%
+4 2.03%
+5 1.09%
+6 0.60%
+7 0.31%
+8 0.16%
+9 0.08%
+10 0.04%
+11 0.02%
+12 0.01%

These are based on a sample of two billion hands, and the "adjusted" figures above have proved to be just fine for my purposes.

If you want or need the accuracy to the nth degree, then you will need to invest in a sim or even program one yourself if those currently available don't meet the need - the instance that springs to mind here is analysing a new counting system that deviates radically from those currently in use?

Trust this helps.

 

[kewljason]

For what game are these true count frequencies for??

 

[UK-21]

Fair question. Don't know - the poster doesn't say.

But with these being the result of a 2 billion hand sim, whether the figures are based on a single deck or 8 deck shoe I reckon the difference will be insignificant unless you want a very precise answer (in which case run your own sim).

The more important question I think is whether these figures are based on dealing to a point where all cards are dealt out, or whether a cut card was used - and if so at what penetration was it set? Again this is not stated.

It's been published within the forum that instances of TC+3 or more occur 8%of the time. The tally of these figures is 7.89%, which bear that out. There will be a margin of error depending on the number of decks and penetration, but for my purposes I can live with it.

 

[Kasi]

There will be a margin of error depending on the number of decks and penetration, but for my purposes I can live with it.

May I ask what your purposes are - are you saying you use that generic set of freqencies for all the games you play?

Whatever, i think the frequencies above would change significantly depending on game and spread as you hint at. maybe what I call significant is what you might call insignificant lol.

Not to mention the adv% that might be associated with each TC.

And that sims can figure out just when to bet how much to keep things optimal for you. And every other question you may want to know the answer too.

No big deal - we agree sims can be useful at times lol.

I just don't see how one can analyze one's results after the fact when one never knows what to have expected in the first place lol.

 

[UK-21]

I use those TC frequencies within a spreadsheet I have produced that crunches approx EVs, returns and three Std Devs each side, based on a betting ramp, unit value and number of hands played. It's not accurate enough to land a man on Mars but accurate enough to determine which strategy still leaves a game with a neg EV, which breaks even, which has a pos EV and by roughly how much. I'm not sure how you could use the frequency figures to analyse results of actual play (and what would be the point anyway)?

Building this spreadsheet has been part of the learning process and understanding the game.

I'm a recreational player I don't play enough BJ to justify getting into analyses that require a sim, and anyway, I don't have the time to learn to use yet another piece of software which probably won't get that much use if the truth be told.

 

[Kasi]

I use those TC frequencies within a spreadsheet I have produced that crunches approx EVs, returns and three Std Devs each side, based on a betting ramp, unit value and number of hands played. It's not accurate enough to land a man on Mars but accurate enough to determine which strategy still leaves a game with a neg EV, which breaks even, which has a pos EV and by roughly how much. I'm not sure how you could use the frequency figures to analyse results of actual play (and what would be the point anyway)?

Building this spreadsheet has been part of the learning process and understanding the game.

I'm a recreational player I don't play enough BJ to justify getting into analyses that require a sim, and anyway, I don't have the time to learn to use yet another piece of software which probably won't get that much use if the truth be told.

I understand newb99.

All I'm saying is, once one has the Freq, ADV% and SD per TC from a sim, one's sheet can be very accurate indeed. That's all one needs, as near as I can tell.

To me, it's the Holy Trinity from which All Things Flow lol.

You can't even approximate an EV without an adv % at each TC. (well, you can but you'd have to estimate some adv % at each TC - how do u go about that?). Not only do the frequencies at each TC vary alot by game and pen etc but also the adv% one enjoys at each TC also varies alot depending on game, pen, style of play etc.

One uses the freq's from a sim to analyze actual results becasue it is only that that will produce avg-units-bet/rd. Usually based on a min unit.

One uses the adv% at each TC to develop the win% per hand, usually based as a % on the avg bet/hd.

One uses the SD per each TC to develop SD in units per round kind of thing.

I mean one's sheet should be able to duplicate the results (very very closely from ROR to N0) of any sim based only on the freq, adv and SD of the sim.

But one still needs the Holy Trinity that only a sim can produce.

A sheet can tell you easily what changes if one changes spread etc but it just can't know the freq, adv % at each TC and SD at each TC.

You see what I mean?

 

[UK-21]

You're right. There are a lot of variables and it's possible to dig down deep into the dynamics of the game. I don't think I need to go any deeper than I have for the amount of time I spend at the felt however, as I'm never going to play enough so I reach the point where variance doesn't swamp (someone else's description although a good one I think) the EVs etc. As I've said, I've done the sums purely as a part of the learning curve and so I can compare the different playing strategies - I have a fair idea of the cost of playing all against WO at -2 say, although it's somewhat academic in most of the places I play at where typically all slots are taken, and there are players waiting to jump in. We don't have any SD/DD in the UK, so one game is much like another with the main difference being the number of cards cut out of the game.

As I've said to Flash in a PM, what's the point of everyone standing around naked at the North Pole discussing what the exact temperature may be?

 

[Kasi]

As I've said to Flash in a PM, what's the point of everyone standing around naked at the North Pole discussing what the exact temperature may be?

Let me guess - he replied, agreeing, "why worry about freezing to death now because, after all, we are all still alive and summer is coming and things are bound to turn-around soon because we came with adequate clothes to begin with and we all knew exactly what we were doing every point along the way that brought us to this naked point".

No big deal, I'm with you. Just a recreational player who never saw a reason to buy a sim either.

Just trying to say, as you are I think, you can plug the frequencies from Don's table 10.11 (7/8 S17 DAS) into Table 10.8 (5.5/8 S17 DAS) and get the sameish avg units bet/round but the changes in adv % per each TC changes the world how many units you could have in roll with same risk, avg units won per round etc.

Notwithstanding, your avg-bet units/rd, also would change fairly substantially if you kept frequencies the same.

 

[sagefr0g]

cold day in hell

Originally Posted by newb99 View Post
As I've said to Flash in a PM, what's the point of everyone standing around naked at the North Pole discussing what the exact temperature may be?

ok, i haven't read all the posts in this thread.
i'm just wondering if maybe the point or a point of them is the idea that like the fact that there can be a generic basic strategy that is better than no strategy then is it also true that one could come up with a generic betting strategy for all common games or maybe a generic betting strategy for various class's of games sort of thing.
i mean, heck i think it's an interesting sort of an idea.:whip:

 

[Kasi]

ok, i haven't read all the posts in this thread.
i'm just wondering if maybe the point or a point of them is the idea that like the fact that there can be a generic basic strategy that is better than no strategy then is it also true that one could come up with a generic betting strategy for all common games or maybe a generic betting strategy for various class's of games sort of thing.
i mean, heck i think it's an interesting sort of an idea.:whip:

O heck = I lost my last reply lol.

The long and short is why not lol.

All I do anyway is have fun with Don's tables anyway when I make up my stuff lol.

So, to newb99, attached are Don's tables 108 thru 10.11 for a S17 DAS game dealt to diff pen levels.

Change anything you want that's in red (plug 7/7 freq into 5.5/8 freq etc)and see what would change in this example.

 

[Dopple]

Wow thanks

That is very interesting, thanks so much for the data. Two questions right off the bat please:

1. I see you showing an advantage as increased bet at TC1. I thought we had to wait til TC 2 for an advantage. It makes alot of difference in 6d with 11 half decks out when you begin to raise the bet. Either RC 11 or RC 22.

2. Would there be anyway to field test this data just out of curiosity or does the computer programs algorythm somehow prove itself provided you have the correct inputs?

Thanks again, I will be looking at this more.

 

[Kasi]

That is very interesting, thanks so much for the data. Two questions right off the bat please:

1. I see you showing an advantage as increased bet at TC1. I thought we had to wait til TC 2 for an advantage. It makes alot of difference in 6d with 11 half decks out when you begin to raise the bet. Either RC 11 or RC 22.

2. Would there be anyway to field test this data just out of curiosity or does the computer programs algorythm somehow prove itself provided you have the correct inputs?

Thanks again, I will be looking at this more.

Glad you liked it.

Well, the sheet was based on the published sims in Schlesinger's BJAIII. Do u have that book to compare to by any chance?

They assumed, basically, I think, a guy using Hi-Lo with the I 18 indexes and truncating his TC's (I think lol). Maybe it assumes he was flooring his TC's - I'd have to look it up. Also assumed he was playing "all hands" "practically" as opposed to "optimally". The math is the same, more or less, for all that and whether he is "back-counting" "optimally" or "practically".

I have the same type sheet for every line in every table in his book becasue I got bored one day a while back.

Now, since you speak of RC 11 or RC 22, it sounds to me you may not even be using Hi-Lo as your counting sytem. Maybe you are using KO?

The very reason, in a nutshell, why one may choose to invest in a sim in the first place.

One tells the sim one's counting system, the rules of the game, the number of players, the extent to which one uses indexes, the pen level, whether one is playing all hands, or entering at a + count and at what count is leaving, or whether one starts every shoe but leaves at some count etc, whether one is spreading and, if so, how much and when, etc and then the sim spits out the frequencies of TC's (or I assume RC's) occuring, the advantyage associated with that count when it happens and the SD associated with that TC when it happens.

Then it produces the same crap my sheet does lol.

Make no mistake - the sheet is not creating anything and does absolutely nothing a sim wouldn't do in the first place. And probably not as well - it's just my own experiment of seeing how Freq, Adv and SD interact AFTER a sim provides them to me.

There's no way to "field test" the data I know of - the results of a sim ARE what to expect from any field test. That's why one uses the results of a sim to measure one's "field tests" results by. SIM=TRUTH. The ultimate act of faith in just accepting that is what will happen. One's actual result's=variance from SIM. One's field results vary by 10SD from SIM, no, don't blame the SIM. Look inward lol.

Oh yeah, the sim also figures out the best amount of units to bet at each count to maximize the avg ratio of EV to SD given the assumptions you tell it to begin with. Go ahead and change a few units spread a little - if N0 goes up, it's a worse, less efficient, way of betting.

Hope that helps a little in answering your questions. If not, keep asking and I'll do my best.

 

[Dopple]

Yes that answers alot of questions Kasi. I use the Uston APC count and I bet the TC correlates to most of the common systems. I might get Schlessingers book, I have bought alot in my day. If my luck (lol) holds out I may invest in a sim too. I am ready to start getting more money on the table but want avoid getting too aggresive and over zealous after a few good wins. I can afford to suffer downturns twice as bad as any I have seen in the last 1000 hrs of play.

Thanks again for all the help.