I was playing BJ when a really nice guy sat next to me and started talking about basic strategy. As a fairly new player I didnt know whether to split my 6's against a 4 or not but he helped me out and the rest of the table seemed to agree with his desicions so I figured he really knew what he was talking about. I made a comment after a $100 4 card split I won about how nice it would be to make money this way everyday to which everone chuckled. A couple of hands later I made enough to cover dinner so I got up and headed to the restaurant. After the waitress left with my order the guy from the table saw me, came to the table and sat down and asked me if it was ok to which I agreed. After some meaningless chit chat, he told me he has played for decades and had a winning system that he wasn't trying to sell to me but rather share with me, he called it his "level system."

It works something like this:

With an initial bankroll of $1000, play to win just $50 each day until you make $500. Once you've done that play for $75 a day until you make $1000 etc. He said to continue this until you reach playing for $200 a day.

Some notes:
-He said to play 2 hands to help with losses (lose 1 but win 1= push)
-He only plays shoe games with 6 decks maximum
-He has tried this all over the U.S and never got any heat
-He does not count cards, he observes them? but plays optimal perfect strategy
-If he loses half his daily bankroll he quits for the day and comes back the next but he doesn't chase his loss
-If he goes on a losing streak 3 days+ he reverts back to the previous level. For instance, he's playing for $75 so he goes back to playing for $50 until he back up .

Has anyone heard of somehting like this? To me being pretty naive about BJ this seemed feasible but I'd like some feedback.

Sounds like a martingale variant.

any system where you have -ev and change your betting amount according to magic or anything is bound to fail in the long run. i have had sessions where i lost 10 k without ever being able to be ahead 25 $.
in contrast a system where you have + ev does not also guarantee you will win, you'll win mostly in the long run.

The second statement contradicts the first though the general ideas are functional.

To OP,

Yes, it's martingale, this board's favorite topic starter. Your new friend is (almost certainly) taking the worst of it though his story could still be true. If he has a relatively low standard of living and an outside source of income his story is even feasible though likely incomplete.

OP - the key idea you need to know is that no betting system ever changes the EV of a game, only the rules of the game control the EV. So with basic strategy if the house has a 0.5% edge on you, then over any amount of time your expectation is to lose 0.5% times the total amount of your bets. Period. You can't change that, and your average result is going to converge to that amount.

Secondly, what any progressive betting system like Martingale does is just skew the distribution of wins and losses. You will have many small wins and a few big losses that more than cancel them out. And the big loss can happen on the very first session, or any session.

And let me guess, anyone who PMs is offered a link to where OP's friend is selling a book about his sure fire strategy?

I'm not sure this is martingale at all. This seems to be more of a "bankroll" system.

In a true martingale system, if you lose, you double your bet and keep on until you win.

This one seems to rest upon the "hot/cold" streak theory.

The downside I see to this is that you follow this process until you get to $200/day. 200/8 hours in corporate America is $25/hour (and you get no benefits and no guarantee of winning).

What is his "daily" bankroll and what is his "full" bankroll? Also, if you make $200 today, lose $500 tomorrow, and win $200 the next day, you're still down $100 from where you started.

One of the issues with blackjack as a "job" is that you're HOPING to get a 3% (at best) advantage in the game. While the house may play gladly at a 1% advantage, they have multiple tables and an almost infinite bankroll. Also, the 1% advantage is assuming that the players have a clue how to play. Many don't.

There is something I have never quite understood about criticisms of martindale or other betting systems. The argument is that, for example if you double the bet like martindale, at some point you will be ruined because eventual a bad streak will exhaust your bankroll. And this is true mathematically. It is one of the things you learn in your gambling life. And a lot of people get a bit superior sounding when they learn it.
Okay my question:
The criticism seems to ignore normal streak distribution. That is if you graph wins and loses (each hand that is) you will find 1 in a row, 2 in a row, and 3 in a row clustered in the middle of the Gausian curve. Yes, in a large enough sample 10, 100, or even a million losses in a row can and will occur. But they will still be distributed in the skinny part. That is they will be more rare than 1, 2, or 3.
So couldn't we play a modified martindale type system which simply avoids those anomolus long bad streaks. That is quit (for example) if you lose 4 hands in a row. And also quit if you win 5 hands in a row. The distribution should be close to even. So wouldn't you gain a small advantage.
That is is it really necessarally most correct to think in terms of the largest possible sample? Of course the numbers I gave are just examples maybe it's 6 and 10.
But I think you understand my point. What's wrong with this?

What's wrong is that you can't "gain a small advantage" like this. The average result is going to be a net loss of the house edge times the total wagers no matter how you split up the bets and no matter what stop loss and stop wins you use. You don't change the EV of the game at all by betting strategies. Ever.

If your rule is to quit after losing 4 bets and quit after winning 5 bets, then you will have more of the losing sessions than the winning ones and the sum will still be negative.

What's wrong is you need to take a math class.

$10/hand blackjack. If you win 5 hands in a row, you've won $50 assuming simple wins and no blackjack). If you lose 4 hands in a martingale system, you've lost $10, $20, $40, $80, for a total of $150. Since you have a -EV without anything else being considered, you should lose more often than you win. Therefore, losing "streaks" should be longer and more often than winning "streaks".

Martingale works if you a) have an infinite bankrolls, b) have a casino that will let you double your bet an infinite amount of times, and c) won't ever cap your betting or stop taking your bets. In other words, unless you have Bill Gates' bankroll, it's useless. Also, in the end, you may win a few small bets overall, so starting at $10, you might end up $100 in a day (again, assuming the infinite bankroll, etc. In which case, why in the bleep are you playing blackjack anyway?

Well I don't use such systems but I do wonder about it. The usual answer is that it is a mistake to think in terms of sessions. It is all one big session is the correct thinking I am told.
Anyway lets pretend were not talking about about BJ but coin flipping. Something with no inherent advantage to either side. I am pretty certain that a betting system as I described would be profitable (in the long term). A long streak would come up ( say 1 every 100 tosses) We would profit from ours. But not lose the bad ones because we would not be playing those.
Of course we lose a little because some of the bad streaks would have come back our way.
But I don't think you can argue that we would not profit.

So does it really make sense to say that this has no affect once the game is not level? Perhaps it would not be sufficient. But can we really say that the normal distribution of results can not, and should not, be looked at?
--
BtW I forgot about the doubling in Martindale. Just assume straight bets. Leave out any double up strategy. Can't distribution be a factor in overcoming or lessening house advantage?

That's absolutely wrong.

The total EV of a session of any length and with any composition of bets, is the sum of the individual EVs of the bets. On a coin flip every bet has $0 EV.

The exact same thing is true for a -0.5% edge in BJ. Each bet has an EV of the wager * edge. The session has an EV of the sum of all those.

It is impossible to add up a set of negative numbers (or zeroes) and get a positive number. That is what you are speculating you can do.

First sentence is fine. How do you go from there to the second?

Start thinking of it as one long session. Does a .250 hitter in baseball become a .300 hitter simply by spacing out his at bats? Of course not, and a .250 hitter is "close" to a .300 hitter in much the same way you're saying these distributions are close. At the end of his career he's still going to be loser vs the .300 hitter the same way the guy paying 5% commission on his banker bet, taking 7-6 odds on a 6-5 proposition, or hit and running a blackjack dealer or superior poker player is not going to gain any kind of advantage by breaking his "career" up into small sessions. The .250 hitter is a .250 hitter one at bat at a time.

Also, I agree Martingale might not be the best label for this system to who said that. It basically seems to have the same fundamentals, go in with a big bank and try to book a small win then use the magic of driving away from the casino and returning to gain an edge, so it's easy to just use the martingale word.

Okay. I can accept what you are saying. And I do understand the idea of independent events. I certainly am not saying that heads are due just because 4 tails have occurred.
I was simply thinking that we can avoid some of the longer bad streaks, which will occur, by simply not playing.
Apparently what I am calling "normal distribution of results" can not be used in a constructive manner. But it certainly feels like I should be able to squeeze something out it. Its hard to believe its completely useless information. But apparently it is
So, probability but no statistics in gambling