Tuesday, April 7, 2009

There's no escaping the fact that math rules in the casino. And it's also easy to see that mathematics does not rule out progressive betting.

_
The consistent success of target betting or any method that breaks the fixed or random mold is assured by the fact that outcomes in games of chance are collectively predictable.

In blackjack, for example, a reasonably representative sample (say, more than 1,000) will confirm that pushes occur about 9% of the time, player naturals bring in a little extra from about 4% of all hands, and doubles and splits governed by strict basic strategy play will win about 20% more often than they lose.

You can also be sure that paired wins or losses will occur about half as often as isolated wins or losses, triples will be encountered roughly half as often again, and so on.

"Roughly" and "about" are words that academic mathematicians hate to hear, but the smaller the sample of outcomes, the harder it is to predict what will happen with absolute certainty.

Luckily for the target player, absolute certainty is absolutely unnecessary.

Regular readers will know that I say with monotonous regularity that If you can win more when you win than you lose when you lose, then losing more often than you win won't hurt you.

A picture is worth a thousand numbers, so here's a summary for the 100,000-plus baccarat rounds supplied by Lorenzo Rodriguez and Zumma Publishing:-

(Click on the image to enlarge it)

The first thing to notice is that after a single crash-n-burn and 13,640 successful recoveries, the AWB/ALB number is a relatively unimpressive 104%, but still enough to deliver an overall profit.
The overall HA/AV for this set was 1.6%, which is high for baccarat but not desperately so.
Two out of three recoveries were achieved with a single win, although as negative expectation required, more recovery/turnaround attempts were lost than won (the house had a 1.16% edge in that area).
Series requiring 10 or more rounds to recover were less than 9.0% or 1 in 11 and 62% of all series were wrapped up in five bets or fewer.
The average bet value was high according to some at $592 but bets of $1,000 were required only 4.4% of the time (less than 1 in 20).


Many a mythematician might be inclined to dismiss 114,000 rounds as anecdotal or non-representative, so keep in mind that the sample is equal to well over a year of play for a fulltime baccarat fanatic.

Also keep in mind that according to the same self-styled (and often house-trained) "experts" the results you see above are impossible.

Now please consider a much larger sample of baccarat outcomes, more than 200,000 supplied by Lee Jones, a systems promoter who can verify their accuracy and with whom I have no connection (he might even consider me a most unwelcome competitor since I don't charge for my ideas and his are for sale).

(Click on the image to enlarge it)

The AWB/ALB product is higher at 109% vs. a more standard house edge of 1.18% and the average bet value is way down at $352 ($592) but the rest of the numbers in the summary closely resemble those for the smaller and quite separate sample from the Zumma books via Lorenzo Rodriguez:

EOS 1-5 62.5% (62%)
EOS 6-10 29% (29.3%)
EOS 10+ 8.5% (8.7%)
Positive series 35% (35%)
Neutral series (29%) (28%)
Negative series (36%) (36%)
1-win EOS 66% (66%)
EOS wins/tries 49.1% (49%)
Bets of $1,000+ 3.6% (4.4%)


The biggest difference between the two baccarat samples lies in the final target betting win (2.86% for the larger data set vs. 0.16%). The reason, of course, is that the first sample included a bust, one that was triggered by such a precipitous plunge that only the "robot" at the heart of every runaway sim would have sat through it!

Baccarat holds far fewer surprises (good or bad) than blackjack, but the stats for the first 75,000-plus outcomes in the BST blackjack trial are remarkably similar to the baccarat numbers, except for a much higher win percentage for target betting and a dramatic jump in the AWB/ALB product:-

(Click on the image to enlarge it)


And here are data for double-up (a Small Martingale) against the baccarat outcomes. The "win one unit" original version has the double disadvantage of being easily spotted and blocked, and a slow win rate that means that a single "bust" would wipe out years of profits.

(Click on the image to enlarge it)

We are now looking at close to 400,000 "real play" outcomes, which is a representative sample according to any objective, sensible standards.

We can see that ruin (the loss of our $1,000,000 bankroll) is a rare occurrence indeed, and that in spite of a single crash-n-burn, target's notional profit at this point is closing in on $4 million

Spread is the critical factor, we have learned. A spread of less than 1-1,000 is doomed, in the long run.

We do not have a monopoly on that knowledge: the casinos are well aware of the danger to their bottom line posed by progressive bettors with a disciplined plan and a fat bankroll behind them. They and will do whatever it takes to thwart them.

Obviously, a player confined to a single casino or, worse, to a handful of table options, can never hope to spread from 1-5,000 (the recommended optimum) or even 1-1,000, in the same place.

The solution to that is an over-abundance of patience, and strong legs. Spreading from $5-$100 (1-20) from the opening of a new series is plausible, but after that, 1-5 is the limit to aim for ($100-$500, $500-$2,500, $2,500-$10,000 and $10,000-$25,000, for example).

One player flying solo might find the range hard to achieve under ever-increasing casino staff scrutiny, and tough as it sometimes is to run smoothly, team-work is probably the answer.

For now, all we are concerned with is "The Math." And it is on our side.

If you doubt it, here are new numbers from the BST trial (I just began batch #16):


An important reminder: The only person likely to make money out of this blog is you, Dear Reader. There's nothing to buy, ever, and your soul is safe (from me, at least). Test my ideas and use them or don't. It's up to you.
_