Beyond BS - deviating from BS to reduce ROR?

I've just finished reading Schlesinger's "Blackjack Attach" (an original 1997 edition that I had sent over from the States). Although fairly heavy on the maths in part it is (IMHO) an excellent progression from books that cover the basics - rules, BS and the mechanics of counting, play deviations and bet sizing etc.

For me it raises a number of questions that the regulars on the Board might wish to comment on?

1. The prudence of deviating from BS in order to reduce variance and the risk of ruin.
I think Mr S comments excellently on camouflage play and what it costs in $$s. In para 3, page 124 he alludes to the fact that the value of soft doubling is such that it makes little difference to the EV whether you do it or not - "go ahead and do anything you please". As the cost of not doubling, say, an A$ v 4 is microscopic (5c per $100), but at a high count it could mean doubling a 16 unit bet (which could be a hefty proportion of a sessional roll - 32/60ths? - thereby betting over 50% on such a marginal hand) should the byword be to not double on these plays where say, there is more than 2 units in the box? I've made reference to not doubling on such plays at high counts in a previous post, but didn't get any response on this issue. When I was very green, I would have just bet according to BS on these, but understanding more about the "return" on some plays now I'm not so sure. Comments?

Mr Schlesinger's betting ramp.
Mr S's word on increasing and decreasing bets - only ever increase by parlaying a win, never increase after a loss, never decrease after a win. For a recreational player would to do this leave those who play a 6-deck game with not much more than a from of entertainment unless s/he was consistently playing black chips or double greens? Presumably, a pro looking to make $50K+ pa would need to bet far more aggressively? I can do the sums, just interested in others' views.

Playing time to counter the effects of variance and Std Dev.
An interesting figure from the table on page 26 - with a 6-deck game, it is calculated that someone would need to play for 210 hours for the win to equal the Std Dev (and that's assuming only playing at TC+1 and above). Again, for a typical recreational player playing say, 12 hours a month (1-2 evenings?) where the Std Dev is not going to be overcome for a long time, is it better to adopt a strategy of reducing variance, rather than maximising EV, on the grounds that playing BJ will never be more than a form of entertainment? Taking that to it's logical extremes, can the time to learn, practice and apply all of the skills to cover oneself when counting be justified when effectively it's still just gambling? Should there be a huge neon sign somewhere (this forum perhaps?) that tells students in waiting that if they can't play 10 hours a week to just stick to basic strategy and keep their fingers crossed!?

Quite a bit here. Please feel free to comment as you please?

Newb99

1. The prudence of deviating from BS in order to reduce variance and the risk of ruin.

A2 vs 5 is the soft double that jumps out in my mind as one to eliminate to reduce variance. The slight edge you obtain on the double rather than hitting is almost never worth the increased variance, particularly if you have a big bet out.

Also, when counting 10 vs 10 and 10 vs 11 doubles, it is better to use a much higher risk adverse index rather than the mathimatical +ev index (you would double on these with a +4 or +5, but to reduce ROR it is better to wait until +8 before making this move).

2. Mr Schlesinger's betting ramp.

Although this is a extremely sucessful long term strategy for Mr. Schlesinger, there are probably much better 'cover' betting strategies that don't give up so much potential profit. For a recreational small bet player, I wouldn't worry about any cover, and just bet the count and let the 'chips fall where they lay'.

3. Playing time to counter the effects of variance and Std Dev.

Yes, to make a full time living at blackjack requires full time play, just like any other job. However, there is something to be said for a recreational player playing the game to the best of his abilities. I might never be a PGA pro tour golfer, but when playing golf I still try to hit the ball to the best of my ability. I don't say I can only play on weekends so there is no point in practice or trying to hit the ball well.

Thanks for those thoughts.

10vA and 10v10 aren't on my radar as I'm in the UK and the dealer doesn't take a hole card and check for blackjacks. Those plays have the ENHC specific health warning attached to them, especially at higher counts where the possibilities of the house pulling a 10 or A go up.

I think I'm going to use the data in Mr Schlesinger's book to produce a table of marginal BS doubles that I shall drop once the count goes to TC+2 (2 units on my 1-8) and above. I'm surprised this hasn't already been done by someone, although perhaps it'll appear in a future book entitled "Risk adverse blackjack strategies for players who play less than 100 hours a year" ??

I've also written 8v4 out. I've seen some indices tables where this is recommended at TC +4/5, but the fact that in hi-lo the 7s aren't included within the count leaves a window of unknown size open for the house to draw a 7 for an 11. No point in doubling the max bet with that risk in the background. My 6-deck EV table shows that doubling on 6,2, 5,3 and 4,4 against a dealers 4 carries a negative EV (assuming off the top of the shoe). So although this'll alter at a higher count I can't see doubling these will offer anything other than a marginal benefit at most? I don't have the software to validate this.

Also appreciate the comment on golf, although don't entirely agree it's the same thing. Golf's a raw game of skill (I know - I came, I tried, I ended up with a cricket score) whereas BJ always has an element of luck (oops, I mean variance) in it. If, until you'd clocked up 200+ hours on the greens, and regardless of how much practice you put in and how proficient you became, whenever you teed off the ball shot off at a random angle between 45 degs left and right of the axis of aim, I think we'd have a lot less people playing each weekend. But I do take your point.

2) I agree that Schlesinger's betting strategy outline in Blackjack Attack is very conservative, probably the only one that's more conservative is Ian Anderson's in Burning the tables in Vegas. But longevity concerns may be legitimate.

The downside of using such betting cover is that you will find yourself in cases where you are betting too little or too much on a hand. This will decrease your EV, or increase your variance, respectively. The effcts are worse the fewer the decks:
http://www.blackjackincolor.com/cardcountingcover2.htm

3) The figure is also called N0. The ways to shrink it are to increase your EV, or reduce your variance, or both. But yeah, you'd need to put in a lot of hours at the table to have a reasonably high chance of seeing a profit. Otherwise, short term, it still "just gambling", albeit with an advantage instead of a disadvantage.

lol you an i need to talk Newb99! or some one needs to straighten us out.
if your not afraid to gamble read this stuff and comment please:
http://www.blackjackinfo.com/bb/showthread.php?t=11134
but make the comment over in voodoo lmao.
anyway interesting post and questions! :cool2:

Sagefr0g,

I read through your (linked) posting with interest.

If you take the view that you'll never be able to put in enough time at the tables in order to overcome the effects of Std Dev (and therefore can't justify the time and effort to learn and become proficient with a fully fledged counting system), I would think the answer would be to adopt one of the "low overhead" systems that exist. No point in reinventing the wheel?

Mr Renzey's "Blackjack Bluebook II" contains various strategies ranging from common or garden BS through to the Mentor count which is a level 2. Personally I started out with Mr Snyder's Red 7 (just four indexes to recall), but that didn't exercise the brain enough between trips to the House of Chance - no requirement to learn about and understand the underlying dynamics of the game (if I'd stopped at Red 7, I'd still be making max bet doubles on A4 v 4 and increasing the RoR with my minimal bankroll).

Newb99

it's a complicated world

i thought you might find it interesting even if some of my rant may be clueless lol.
could very well be. at least we know that one can look at all this stuff from a lot of differant perspectives from AP pro all the way to a pure gambler.
i guess i'm guilty of trying to in a way reinvent the wheel. just sort of fun for me so what the hey.
there ya go. Mr. Renzey's varous levels of getting to an advantage in his book Blackjack Bluebook II is a great example of the wiggle room i think we are talking about. precious little wiggle room it is too i should suspect. maybe on the order of 2% probably more like 1%. maybe just enough rope to hang ones' self with lol.
so ok thank's for checking out the link. and if you ever think of anything along the way to help me in my delusions over in that voodoo link jump in and let me know.

In the book, is there a table outlining the cost/advantage of doubling, not doubling, splitting etc for each possible scenario?
I have never contemplated comparing the impact on ROR of doubling/splitting in light of the additional advantage generated by the double/split. But this does sound like it is quite important.

Is such a table also available elsewhere?

dbl A4 vs.4: Blast from the past

I am very interested in the above and was wondering whether this table of marginal BS doubles was ever drawn up.

Thanks in advance and regards

No, it wasn't.

Since my original post in late 2008, I have continued to be pelted with electronic rotten fruit by some for daring to suggest that not playing the mathematically advantageous option, with the aim of reducing variance and avoiding big losses at high counts, might be a prudent move for those with limited bankrolls.

If you PM me an e-mail address, I can e-mail you a spreadsheet with all of the hand values for a six deck shoe at a neutral (off the top) count. What I don't have is the same sheet with values at each TC point - both neg and pos - to base further decisions on. From the number's crunched for me by a respondent in another thread (index for 12 v 3) they could be generated by the sim software and entered into a spreadsheet - although it would be a boring time consuming job.

One of the considerations is to put a value on one's own, personal, risk aversion, ie if a player's risk aversion value is 0.20, then where the difference between hitting and doubling at a high count was 0.07 (7p per £1 bet), you would hit rather than double, despite winning less if you win the hand. I haven't really thought about a mathematical basis for arriving at a value for this, although it would have to be relative to RoR on whatever bankroll one has.

I haven't got around to writing the book yet either.

The problem with dropping marginal BS doubles on high counts is that they are no longer marginal !!

For example A2 vs 5 may not matter all that much in EV whether you hit or double if just playing basic strategy but that all changes with a higher true count. Your EV increases on this play the higher the count because the dealer will bust more often.

However, I agree that steps should be taken in high counts to reduce variance on close plays, but not according to your BS strategy but according to your BS deviations.

For Example...

8 vs. 5 You estimate the true count to be exactly 6.

Let's say you have a max bet out there...

Of coarse basic strategy is to hit...

But maybe your index says to double on a true count of 6. Maybe this double is not worth the risk in variance.

Though, before making this decision knowing more information will be helpful. If the actual true count needed to increase your EV by doubling was only 5.25 then maybe it would be better to double... but if the actual TC number is 5.90 required to increase EV then doubling may not be worth the risk.

Actually, I'd like to know myself.. how to find the actual index numbers instead of the rounded ones presented for all counting systems.

Risk Adverse Strategy

A lot of counters implement risk adverse strategy. Here's the basic idea and principle...

Suppose you bet $100 and the index for the hand you draw is to double at +1 or greater, and currently the count is at +1. So if we double, perhaps our EV for the hand is a penny or two higher than if we just hit. Is it worth putting twice as much money at risk just to make an extra penny or two? Most people would say no. Thus we knowingly accept a slightly lower EV% in order to decrease variance a good bit by just hitting instead of doubling. And since our variance is now lower, we can bet a little bit more without increasing our risk of ruin, which increases hourly EV.

Now if the count was say +4, the EV for doubling would likely be substantially higher than just hitting, and thus we become readily willing to double as we are getting paid nicely instead of a mere penny or two to put our additional $100 at risk.

The same principle goes for splitting. Risk adverse bettors are slightly less aggressive on splitting which puts twice as much money at risk.

A quick, down and dirty RA strategy that many use is to add 2 to the dbl and split indexes (assuming single level count). In other words, say the index for your system on a given play is to split at +2 or higher. You can instead implement RA strategy by not splitting until the count is +4 or higher. By doing this, you will only give up a tiny bit of EV and in exchange you will lower variance a good bit. Technically, adding 2 to all dbl/split indexes isn't correct as different hands vary in the amount of EV change per change in count, but the +2 is quick, easy, and pretty close over all, and I think more detailed analysis is beyond the scope of this forum. Also note that hit/stand indexes remain the same since those decisions don't require additional money to be put at risk.

Won't the usual risk averse indices as determined by CVData or other sim software do exactly what is proposed in this thread? Don't they automatically tell us when to deviate from BS in such a way as to maximize our SCORE even if doing so means decreasing variance at a slight cost to the EV? I guess I'm sort of missing the point of this thread

Yep, that's the point of RA indexes. Individual decisions are looked at and indexes created that accurately balance risk and reward for the specified counting system.

This is true. Marginal doubles off the top of the shoe like A/2 vs. 5 and A/4 vs. 4 yield around an extra half percent EV over just hitting. But at +3 TC, they yield around an extra 3%.

And A/8 vs. 5 or 6 yield around an extra 7% EV!