Monday, October 3, 2011

If you're still on the fence about wide spreads and "twins" here are some numbers that might tempt you to climb down on my side.

_
(For updated information about Target's ongoing sports betting experiment, please go to the Sethbets website, or click here for an introduction to progressive betting, and here [20] and here [5] for current picks)

Most gamblers feel very strongly that any analysis of casino games of chance is a waste of time that could be better spent playing at the tables - and probably losing.

"You know there's a house advantage and it is whatever it is, so why worry about it?" sums up the prevailing attitude, along with, "No one's forcing you to make a bet."

All true.

But risking good money on any proposition without a full understanding of the numbers at work makes no sense to me, and I am comfortable being a member of a tiny minority.

I know this much: Even without the huge bankroll that I know is the most important weapon against negative expectation, I lose a lot less often - and a lot less money - than most of the people I know.

And the regular gamblers I come across out there in the real world are all experts as far as they are concerned, convinced that their way, and only their way, works.

You have to admire a blackjack player so dedicated to his own infallibility that he will risk a smack in the eye by telling a stranger, "You never hit 13 against a 2" or "I never hit a three-card 16 unless the dealer is showing a 10 or an Ace, and I don't care what the book says!"

As it happens, it really doesn't matter what the book says, and the self-styled experts I'm quoting are probably right about as often as they are wrong, just like anyone else who prognosticates about casino games.

And that's the reason why when I get an idea, I round up as much data as I possibly can and test it with the help of complex spreadsheet models that have been years in the making.

Every test says the same thing: How you play (in the few games where you actually get to make choices after placing your initial bet) is a whole lot less important than how you bet.

That conclusion is, of course, anathema to the thinly-veiled con artists who charge hundreds of dollars for software and systems that claim to "teach" optimum blackjack play, using card-counting or "advantage play."

It's true, there are times when the odds of winning at blackjack improve when the composition of cards yet to be dealt matches certain criteria.

But the improvement is so tiny that if you don't have the right bet on the table, knowing the deck count won't help you.

Ace prediction, shuffle-tracking and all that malarkey are money-makers for only one group of people, none of whom actually play blackjack because they know that with or without their magic tricks, it's a losing proposition.

Unless you bet right, of course.

A while back, I mentioned baccarat data that originally featured in "System Tester" books put out by Zumma Publishing, and were supplied to me in spreadsheet format by Lee Jones.

It's about five years since I have looked at them, but baccarat has been coming up a lot lately as a suitable platform for Target Betting (I guess it never really went away) and my models have improved significantly since 2006.

So out came the ancient Zumma files, and I spent half of Sunday picking up all those numbers and plugging them in to new models specifically set up to evaluate different spreads all the way from 1-10 to 1-5,000.

I'm often criticized for saying that a spread as wide as 1 to 5,000 may be the only tenable option at any casino game, the primary complaint being that not more than one player in 10,000 can afford to risk that kind of money.

I understand that, I really do, but I can't apologize for accurately reporting what my research uncovers.

I am also convinced that the basic principles of Target Betting can serve any player well, as they do me, even without a bankroll that runs into six figures.

First, a summary of "bare bones" results for the entire Zumma data set, which I think may have been augmented by Mr. Jones, and all of which he claims is drawn from actual shoes and not cooked up by a Wiz-style sim.

As I see it, the most important thing about the Zumma/Jones sample is that it isn't mine, and I did nothing to alter it.

There are people out there who will say that 186,000 rounds is not a "representative sample" - but there are also people out there who say global warming is a Liberal hoax, and I accord both groups the same degree of respect.

The sample depicted below adds up to more than two and a half years of professional play (five hours a day, five days a week, nine months a year) so I contend that it has something useful to tell us.


I am happy to see that even stripped down to its barest skivvies, Target Betting delivered a 4.0% win overall against a 1.30% house edge, which is probably about right for a large baccarat data set.

(Note that Target did not lose even once against 26 separate data sets).

I'm not so happy about the size of the spread that was needed, but since it just confirms what I have been saying for years, there's nothing I can do about that.

"Bare bones" Target means playing the ultimate waiting game - no win progression, no loss progression, just flat betting until a mid-recovery win comes along, prompting an LTD+ bet that has just a 1-unit win goal.

So: -10,-10,-10,-10,+10,-40,-40,+40,+80 = +10/250 = +4.0% vs. AV -3/9 = -33%, Expectation -83.

Nothing fancy, and it would drive me nuts to bet that way, but often I just want to test the primary rule of Target Betting, and this was one of those times.

These were all Player bets, by the way, because in spite of all the stick I take from baccarat aficionados, I flat-out refuse to hop back and forth between two choices in any casino table game.

And giving up a substantial percentage of my winnings by betting Banker is just plain daft - tell me "it's just 5 percent" and I'll know you're a global warming denier!

In a game with two or more options, pick one and stick with it through thick and thin to avoid zigging when the cards zag and vice versa. It's just statistical common sense.

The spreads breakdown from the Zumma/Jones sample looks like this:


Action values are right next to the overall result, win or loss (mostly loss!) for each spread range.

The big stand-out in a sample this enormous is always confirmation that spreading 1-5,000 does not, in the end, call for much bigger bets on average than spreading 1-10.

The explanation for that "surprise" is that the tighter your spread, the sooner you'll hit your green ceiling, and the longer you'll have to keep placing top-dollar bets in pursuit of turnaround.

Which, of course, hardly happens, like hurricanes in Hertford, Hereford and Hampshire.

Spread as wide as you possibly can, and you'll enjoy faster recoveries and smaller bets overall because each turnaround prompts you to drop back to a minimum bet and start over.

"It's the arithmetic!"

I'll just briefly refer to the inertia effect for those who haven't been here, done that already.

The problem with any large data set and the process of applying a fixed set of betting rules to it is that real people are not computers, and they respond to influences such as "hot" and "cold" tables and gut feelings - not to mention dealers with bad breath and drunks with big cigars! - in ways that a computer cannot imitate.

So this very large sample includes preposterous losing streaks that no human player would put up with. I don't worry about matching winning streaks, because I think most people would stick around for those!

The most important lesson from all those boring, never-ending mathematical exercises is that it is essential to bet consistently in such a way that you can win back prior losses in fewer bets than it took to lose the money.

You have to Win more when you win than you lose when you lose, in other words - something Target achieved here with wins averaging $46 each and losses averaging $41 a pop. There were, as expected, more losses than wins - but, as always, Target didn't care.

It's important not to throw wins away - always know how much you're behind, and when the potential for even a short two-bet winning streak arises, as it does each time you win a mid-recovery wager, make the most of it.

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. One more piece of friendly advice: If you are inclined to use target betting with real money against online "casinos" such as Bodog, spend a few minutes and save a lot of money by reading this._