I have only seriously played single and double deckers around Vegas. I used the Red 7 with good results (on most trips). A possible job I'm considering taking in L.A. has me thinking about the seis-deckers of SoCal. Anyone use this count/index system for shoes?
I use hi-lo but it's supposed to be close...
Here is red7 versus hilo. S17 DAS 4.5 1-15 play all
Red 7:
hilo:
By its abundance of information your simulator is very transparent and allows some thorough analysis, however some questions remain:
1)In the heading,you say that the spread is 15:1 for both systems but when we look at the actual bets we see that the spread for red7 is 123:8 (15.4)and the one for hilo only 129:9 (14.3).Why didn't you use $10 as the unit with $150 as the maxbet for both systems thereby reflecting real world conditions and allowing a proper comparison?
2)When we look at the respective RORs,we realize that the one for hilo is slightly lower in spite of the bigger maxbet. Shouldn't we adjust the bets in order to arrive at the same ROR?
If we do so, I guess that hilo has a slightly higher win rate than red7.
3)As far as I know,the original red7 starts the running count at -12 and only raises the bet at RC=0,which would correspond to an RC of +12 for an IRC of 0.So why do you raise already at +11?This is most probably an improvement but it's not Snyders doctrine.
4)Were the red sevens really counted as +1 or were all the sevens counted as +0.5?
Thanks for answering.
Francis Salmon
>> 1)In the heading,you say that the spread is 15:1 for both systems but when we look at the actual bets we see that the spread for red7 is 123:8 (15.4)and the one for hilo only 129:9 (14.3).Why didn't you use $10 as the unit with $150 as the maxbet for both systems thereby reflecting real world conditions and allowing a proper comparison?
Because I only wanted to be "close enough" to show the relative power of red 7 - which the original poster was questioning, for shoes. Thus, I checked "User integer bets", and that accounts for the 1-15 (+-1) spreads. My point: if you wouldn't be concerned about using hi-lo against shoes, don't be concerned using Red 7. The post shows why this is so.
>>2)When we look at the respective RORs,we realize that the one for hilo is slightly lower in spite of the bigger maxbet. Shouldn't we adjust the bets in order to arrive at the same ROR?
If we do so, I guess that hilo has a slightly higher win rate than red7.
My software can do that, so course, but the effect in the above post is due, again, to my decision to use the "use integer bets" setting. Again, this was just meant to be a "close enough" example. For a more detailed treatment, see my published works.
But, just to make you happy, here is the comparison with equal ROR:
And here is the comparison with "normal" chip denominations:
And after all that precision, the main point of my original post remains. And it should not be a surprise. Unbalanced counts are every bit as powerful as balanced ones, and knowing something about 7's, is better than not knowing (hi-lo).
>>3)As far as I know,the original red7 starts the running count at -12 and only raises the bet at RC=0,which would correspond to an RC of +12 for an IRC of 0.So why do you raise already at +11?This is most probably an improvement but it's not Snyders doctrine.
Because of the way I designed BJRM, the screen only allows for "13 bins" of TC or RC. Plenty for balanced counts, but when using unbalanced counts, with their greater granularity (due to RC use), this forces me to combine several RCs in a single bin. You can extrapolate from this.
>> 4)Were the red sevens really counted as +1 or were all the sevens
counted as +0.5?
The latter. In the very long run, the same thing. In a player's regular play, more variance.
>> Thanks for answering.
You're welcome, assuming this satisfies you. :-)
-JA
How couldn't I be after such an exhaustive answer?
I never questioned the viability of red7 for shoe games but I still think that yours and other simulators make it look somewhat better than it is in reality (more agressive betting and counting 7s as 0.5).
I don't quite agree with your last sentence.Greater variance forces you to smaller bets and therefore affects the win rate as well.If it didn't matter in the long run we might as well forget about the whole SCORE-business.
Francis Salmon
For those of you who actually count 1/2 the sevens as 1 and ignore the other ones, you will not get the same results, and will experience bigger bankroll fluctuations.
Counting all 7's as .5 (especially if you TC) will be a net improvement upon hi-lo because the count will be better correlated with the betting EOR's, which is what determines the BC.
Unfortunately, all the red7's sims that I am aware of count the 7's as .5, none of them count the red 7's.
But the data shows that counting 7's as .5 without making any TC conversions is about equal to hi-lo WITH TC conversion, thus proving that the system is inherently stronger. NOT so if you count only the red ones.
TAO
>>I never questioned the viability of red7 for shoe games but I still think that yours and other simulators make it look somewhat better than it is in reality (more agressive betting and counting 7s as 0.5).
You may "think" that, but it is not true. In the example I posted above, where ROR is equal and non-integer bets are used, there is no "more aggressive betting" in the Red 7, and, over a billion round sim, also no difference in win rate between counting only red sevens, and counting all 7's as .5.
Red 7 is slightly better than hi-lo, for the game in question. Furthermore, if you want to get technical, it is QUITE a generous assumption that a sim gives to a balanced count, to assume that the counter, in real life, is doing accurate true count estimation. So "actual" hi-lo performance could easily be worse than simmed.
>>I don't quite agree with your last sentence.Greater variance forces you to smaller bets and therefore affects the win rate as well.
Huh? That makes no sense. In the short run, if you count only "red" 7's, then sometimes those will come more, less, or about equal to the 50-50 proportion they actually are in the shoe. That will create variance, and sometimes cause you to be betting more, sometimes less than "you should". That is the only effect, which goes away in the long run.
>>If it didn't matter in the long run we might as well forget about the whole SCORE-business.
Huh?
Reegards,
JA
As I posted above, I find it mildly amusing how quick folks are to jump on the fact that "counting all 7's as .5 is not the same as counting only the red ones", using that to caution about the predicted results, while giving a complete pass to errors in calc-ing True Counts, that SURELY plague almost all true count system users, but are avoided by RC-based, unbalanced systems. Even if you are off on the RC by a point or two ( which you also could be on a true count system! ), this is only the equivalent of being off by a fraction of a TC.
In other words, regarding "red" sevens versus .5 sevens - don't fret about it, unless you constantly also fret about things like sims assuming accurate true count calculations.
-JA
You are right that in the long run it will even out, but the further away you get from the actual EOR, the more of a chance that you will never reach that long run, because you will catch "streaks" of divergence.
Here is an extreme example: The 3's and the 6's are worth just about the same, so say that instead of hi-lo, you count the 3's as +2, and the 6's as 0, in the long run both of these mistakes will cancel each other out.
Now I admit that I am exagerating a little bit, and I didn't actually do the math (like I did for red7) to verify the degree of veracity of this affirmation, but there will be a cancelling out and the bc will not be terribly affected. When you are in a deck where the distribution of 3's and 6's is near even, you will be near unaffected.
Generally as card counters, when we face "normal" conditions (i.e. no great divergence from even distribution or wild fluctuations), we have no problems and we make money. As long as you have a decent bankroll, the only thing that can ruin you is an "abnormal" serie of divergences.
No matter what, we have "streaks" (that we can only recognize after the fact), and if we are always slightly overbetting or underbetting, tho it will even out in the long run it will increase the magnitude of all streaks, and even if it doesn't lead to ruin, it will lead to emotionnal turbulence, and that is a definite negative.
Please don't interpret this as an attack, it's just that I think there are certain loose ends in the generally accepted theories, and one of them is the way we calculate BC; more specifically the cancelling out that occures when you over-value one number and undervalue another. This is the reason why the BC for every system is so high and I find this misleading even if it is technically accurate in the "long run".
I hope I am being clear enough to be understood,
TAO
Not to put to fine a point on it:
Here is an extreme example: The 3's and the 6's are worth just about the same, so say that instead of hi-lo, you count the 3's as +2, and the 6's as 0, in the long run both of these mistakes will cancel each other out.
That is really not the same thing as "missing" 7's if too many black ones come out before the same number of red ones do.
Adding "the keeping track of 7's" to your count adds to PE, for certain indices more than others. Call them the "seven sensitive" ones.
"3" and "6" are both low cards, to be balanced against 10s and aces, but knowing something about the distribution of 3's versus the distribution of 6's, will affect other of the indices. Some are "3 sensitive", some are "6 sensitive", some combinations of both, etc.
In other words, it also makes a difference as to which card tag you are tracking. This is the reason why counting the "9" as a big card is not that good of a thing. It adds to BE, but throws off the most important index of them all: Insurance.
Bottome line: I would not worry much at all about the distribution of color on the 7's. But if you do, switch to KO or UBZII, then.
It appears to me that QFIT's CV Data/CVCX does have the ability to count red or black sevens. At least, there is the ability to define a counting system based on this attribute. I am not certain whether the Sim actually follows through and tracks color or simply assigns a 1/2 in the background. For what it's worth, I ran two quick sims changing only from the color approach to assigning a value of 1/2. Both sims were 1,000 million rounds for S17 6D DAS, split to four hands no RSA heads up. For a pen of 5/6 and betting spread of 1:12, the 7 = 1/2 approach generated a c-SCORE of 33.50 while the Red 7 = 1 approach generated a c-SCORE of 32.76.
I think the sim was large enough to demonstrate a difference (Norm uses 2,000 million in most of the sims supplied with CVD and CVCX), but I am not certain. Clearly the difference shown here is modest.
Of course, if CVCX is really counting the red 7's as 1/2 the sims are obviously meaningless.
Perhaps Norm or someone could clarify whether the sims actually do incorporate the color.
It's the best there is.
User error is your only caution. :-)
Thank you. Your message speaks volumes about both of you. Koo
First I would like to say that I am only refering to the betting efficiency in my example, and not the overall effectiveness.
I also understand that there may be extra advantages to simplicity, but that is not part of my feild of interest, since I am more interested in the academic aspects of the game than anything else, and anyways I don't use any of the published counts for myself.
If I had to use a published count, my first choice would be a true counted red7 where 7=.5. But the actual red7 would be pretty far down my list: below hi-lo.
The most important point I was trying to make was that the numbers can be misleading in the long run, because some mistakes cancel each other out, but in the short/medium term (within a lifetime of playing) these mistakes might not cancel each other. As the current collective knowlege about the game goes, we simply dismiss it as "variance" that is intrinsically random and accept it as part of the game. But "randomness" is not inherently random; once it is understood, it is no longer random.
I must admit that I am surprised that koolipto's sim shows no more than a 2.4% difference in score, but never-the-less I am pretty sure that this difference is about entirely attributable to the extra variance.
Maybe you can help me clarify this point, I beleive that since the score is a mix of EV and variance, it would take a 4.8% increase in variance alone (with identical EV), in order to reduce the score by 2.4%. Is this correct?
4.8% increase in variance would not be so insignificant.
IMO controlling the variance is more important, because that is where we are more vulnerable, and even if we do not tap out, big losing streaks are extremely hard to take emotionally, and can even be damaging to the inner self, which makes it a matter that goes beyond the financial aspect of the game.
Question for koolipto: If this is not too much to ask, could you re-run the same sim, but with TC conversion, because my math (based on EOR) assumes that everything is true counted, and I am wondering if this would widen the gap.
Thanks in advance,
TAO
Again, you worry about the non-random distribution of red sevens, but not about the ability of the counter to estimate the height of the cards in the discard tray, and then do a quick math division, and then make a quick bet and/or departure decision based on that?
I don't get the priority. You would never use red 7 because of the occasional non-near-random distribution of red sevens, and the extra variance it causes, but you then don't worry about the increased variance of inaccurate discard height estimation, and faulty on-the-fly division to get true count?
I'm really not trying to be difficult. I just don't understand the emphasis on the one, but not the other.
Regards,
JA
Your huh-comments are out of place because the thing is obvious:
Greater variance with equal win rate means smaller SCORE by definition.Smaller SCORE means smaller win rate with equal bankroll and ROR. q.e.d.
The long run doesn't change anything about this fact.
For the other point, I just consulted Snyders "Blackbelt in BJ",and he is ramping up his bets much slower than you do it in your sim.
Francis Salmon
>>For the other point, I just consulted Snyders "Blackbelt in BJ",and he is ramping up his bets much slower than you do it in your sim.
So who do you suppose is correct?
Apologies to any and all befuddled by my earlier message regarding the simming of Red 7. I now suspect that CVD/CVCX also sims using only 7= 1/2.
It turns out that you can manipulate strategies in CVBJ (the practice program), and indeed in that program you can set counting parameters based on the color. Separately, you can use these same strategies in the sim modules. However, when I went to revise these to TC per TAO's request from within CVD, I found that the color attribute was not available and that the 7's were set = 1/2.
Based on the above, I am now guessing that both sims described earlier were showing results for the same system, with differences due to standard error alone.
Question for TAO: I tried to e-mail you privately (through this board) some material on side counting 7's. Did you get this?
