The premise needs a correction
Blackjack has fixed rules and calculable probabilities, so it is tempting to think that the right system of bets could turn those probabilities into a reliable profit. That is not the case. A player who uses basic strategy still faces a small house edge, and no way of varying bet sizes on a game with a negative expectation changes the sign of that expectation. Betting systems can change how results are distributed, but not the average.
This article looks at the best-known betting systems, uses small numbers you can check by hand, and explains what is and is not possible.
What a betting system is
A betting system is a rule for choosing how much to wager based on what has happened earlier. It does not tell you how to play the cards; basic strategy does that. Common examples are:
- Martingale: double the bet after each loss, return to the base bet after a win.
- Reverse Martingale (Paroli): increase the bet after wins, return to the base after a loss.
- D’Alembert: raise the bet by one unit after a loss and lower it by one after a win.
- Fibonacci-style: step along a number sequence after losses.
Each aims at the same feeling: that a win will “recover” previous losses, or that wins can be pressed into larger gains.
Martingale with real arithmetic
Suppose the base bet is 10 units and a player doubles after every loss. A run of losses looks like this:
| Loss number | Bet placed | Total staked so far |
|---|---|---|
| 1 | 10 | 10 |
| 2 | 20 | 30 |
| 3 | 40 | 70 |
| 4 | 80 | 150 |
| 5 | 160 | 310 |
| 6 | 320 | 630 |
The next bet would be 640. If it wins, the player receives 640, which covers the 630 already lost and leaves a profit of 10, the original base bet. So the system “wins” 10 units each time a winning hand eventually arrives.
The problem is what it takes to get there. The bet needed grows exponentially, so a short run of losses demands a very large stake. If the chance of losing a hand is about 0.49 (using a rounded figure that ignores pushes), the chance of six losses in a row is 0.49 to the sixth power, which is about 1.4%. That sounds small, but it happens regularly over many sessions, and when it does the player is already 630 units down with a 640 bet required. The chain also breaks if the bankroll or the table maximum runs out first.
Table limits matter here. Every table has a maximum bet. If the maximum is 500, the sequence above stops at the sixth bet and the recovery step is not available. The small, frequent gains of 10 units are exposed to rare but very large losses.
Why the average does not change
The key fact is that each hand is a separate bet with the same negative expected value. The expected value of a series of bets is the sum of the expected values of each bet, no matter how the amounts were chosen. If every bet has a negative expectation, the total does too.
An illustration with deliberately simple numbers: suppose each unit wagered loses 0.5% on average. A flat bettor who wagers 1,000 units in total can expect to lose about 5 units. A Martingale player who ends up wagering 3,000 units in total, because of doubling, can expect to lose about 15 units. The system has not removed the edge; it has multiplied the amount exposed to it. The expected loss goes up with the amount wagered, which is why betting systems often raise the expected cost of an evening rather than lowering it.
The system does change the shape of results. Most sessions end with a small gain, and a few end with a large loss. The two balance out to the same negative average, or a worse one. A long record of small wins can make the method look sound until the rare bad run arrives.
Other systems, same result
Positive progressions such as Paroli make the opposite trade: many small losses and occasional larger gains. D’Alembert and Fibonacci-style schemes slow the growth of bets compared with Martingale, so large losses arrive more slowly, but they are still sums of negative-expectation bets. In each case the expected loss per unit wagered is unchanged, and the total expected loss depends on how much is wagered.
There is also a detail that matters in blackjack specifically: not every win pays even money. Naturals pay more, doubling and splitting change the amount at stake, and pushes return the bet. This means the “one win recovers everything” arithmetic of Martingale is only approximately true even before the house edge is considered.
The gamblers fallacy behind many systems
Many systems rest on the idea that a loss makes a win more likely next time. Cards have no memory in that sense: each hand is dealt from a shuffled shoe, and a run of losses does not shift the next hand’s odds in the player’s favour. Believing otherwise is the gamblers fallacy. For the same reason, the urge to bet more to get back to even is a form of chasing losses, which tends to deepen losses.
What can be said reasonably
- Bet sizing can affect how quickly a bankroll rises and falls, and the chance of running out. This is the idea behind risk of ruin.
- Bet sizing does not create an edge. Methods such as the Kelly criterion, which size bets in proportion to an edge, are meaningful only if a positive edge exists. Without one, the correct size is zero.
- The best-known method of gaining an edge in blackjack is card counting, which is hard to do accurately, is limited by casino countermeasures, and is not a betting system in the sense above.
If you choose to play, treat it as entertainment with a cost. Set a loss limit before you start, do not raise stakes to recover money, and never borrow to gamble. If gambling is causing you harm, a local support service can help.