Betting systems debunked

Key facts

No betting progression changes the expected value of a blackjack hand; expected loss stays equal to the house edge multiplied by total money wagered, whatever the bet sizes.
Under six-deck basic-strategy rules analysed by Wizard of Odds, a single hand wins 42.43%, pushes 8.48% and loses 49.09% of the time, so the chance of losing a non-push hand is about 53.6%.
Using that 53.6% figure, six consecutive non-push losses occur about 2.4% of the time per Martingale sequence, and at least once in roughly 70% of runs of 50 sequences.
A Martingale from £10 needs £640 for the seventh bet and £1,270 of total exposure; a £500 table maximum ends the progression after six losses with £630 gone.
Positive progressions such as Paroli and 1-3-2-6 risk mostly house money within a cycle, but the money wagered still carries the full house edge.

A betting system decides how much to wager on the next hand based on what happened on previous hands. None of them changes the expected value of a blackjack hand, because the cards have no memory of your bet history and every pound wagered carries the same house edge. What a system does change is the shape of your results: how often you leave ahead, and how badly the bad nights go.

The arithmetic that applies to every system

Expected value is linear. If each hand has expected return of minus e per pound wagered, the expected result of any sequence of hands is minus e multiplied by the total amount wagered, regardless of the order or size of the bets. Take an illustrative edge of 0.5% and compare two ways of betting across three hands:

Pattern	Bets	Total wagered	Expected result at 0.5% edge
Flat	£10, £10, £10	£30	−£0.15
Martingale after two losses	£10, £20, £40	£70	−£0.35
Paroli after two wins	£10, £20, £40	£70	−£0.35

The progression that wagers more loses more on average, and nothing else changes. Wizard of Odds states the same conclusion from simulation: however bets are varied, the ratio of money lost to money wagered approaches the house edge in the long run, and in Michael Shackleford’s words no betting system “can withstand the test of time”. His ten commandments of gambling include a plain instruction not to believe in them.

The only way to change expected value in blackjack is to bet more when the remaining cards favour you, which is what card counting does. A progression keyed to past results carries no information about the next cards.

A per-hand loss probability for blackjack

Progression maths usually assumes a coin-flip game. Blackjack is messier: hands push, blackjacks pay 3:2, and doubles and splits put extra money out. For streak calculations the useful figure is the chance of losing a hand given that it is not a push, because a push simply repeats the bet. Wizard of Odds’ variance appendix gives, for a six-deck basic-strategy game (dealer stands on soft 17, double after split, late surrender, resplit aces), per-hand outcomes of win 42.43%, push 8.48% and loss 49.09%. Excluding pushes:

0.4909 ÷ (1 − 0.0848) = 0.4909 ÷ 0.9152 ≈ 0.536.

So the labelled figure used below is a non-push loss probability of about 0.536. It is a per-hand frequency, not a house edge: many of the 42% wins pay more than even money, which is why the house edge for those rules is only 0.28% per Wizard of Odds despite losses outnumbering wins.

Martingale and Grand Martingale

Martingale doubles the bet after every loss and returns to the base bet after a win. Each win recovers all previous losses in the sequence plus one base unit. Grand Martingale doubles and adds one base unit, so the win recovers the losses plus more.

Loss number	Martingale bet	Cumulative out	Grand Martingale bet	Cumulative out
1	£10	£10	£10	£10
2	£20	£30	£30	£40
3	£40	£70	£70	£110
4	£80	£150	£150	£260
5	£160	£310	£310	£570
6	£320	£630	£630	£1,200
7 (next bet)	£640	£1,270	£1,270	£2,470

Using the 0.536 loss probability, the chance that a sequence reaches six straight losses is 0.536⁶ ≈ 0.0238, about 2.4%; seven straight is 0.536⁷ ≈ 1.3%. A sequence lasts on average about 2.1 non-push hands, so 50 sequences take roughly 105 decided hands, an evening’s play. The probability of at least one six-loss streak in 50 sequences is 1 − (1 − 0.0238)⁵⁰ ≈ 0.70. In other words, a player who sits down for the evening with a Martingale from £10 has about a seven-in-ten chance of meeting the £640 decision.

Two things stop that seventh bet. A £500 table maximum makes it illegal; the progression ends with £630 lost. And a bankroll must hold £1,270 just to place every bet through the seventh, for a target profit of £10 per sequence. Wikipedia’s Martingale article gives the general result: with loss probability q and at most n bets, expected profit per sequence on an even-money game is B(1 − (2q)ⁿ), which is negative whenever q exceeds one half. Shackleford’s Martingale analysis reports the same shape from simulation: the highest chance of a session win of any system, paired with “abject disaster when things go bad”.

Paroli and 1-3-2-6: positive progressions

Positive progressions raise bets after wins and reset after a loss, so the larger bets are mostly funded by money just won.

Paroli doubles after each win and resets after three consecutive wins or any loss. From £10: win → £20, win → £40, win → collect £70 profit and reset. One loss at the third step costs only the original £10 plus nothing further, because the £30 of profit riding was house money. The three-win cycle occurs with probability 0.464³ ≈ 0.10 using the figures above, so roughly nine cycles in ten end short of the target.

1-3-2-6 bets one, three, two and six units on consecutive wins:

Step	Bet	Result if win	Profit if cycle ends here with a loss
1	£10	+£10	−£10
2	£30	+£40	−£20
3	£20	+£60	+£20
4	£60	+£120	£0

The design keeps a two-unit profit if the third bet loses and locks in nothing worse than break-even if the fourth loses. Neither property changes the expected value of any of the four bets.

D’Alembert and Fibonacci: slower loss-chasers

D’Alembert adds one unit after a loss and subtracts one after a win. From £10 the sequence after L, L, L, W, W is £10, £20, £30, £40, £30: three losses cost £60, two wins return £70, net +£10. The system shows a profit when wins and losses are equal in number, which the 0.536 loss probability makes less likely than even.

Fibonacci moves one step up the sequence 1, 2, 3, 5, 8, 13, 21 after a loss and two steps back after a win. A run of six losses from £10 costs 10 + 20 + 30 + 50 + 80 + 130 = £320, and the seventh bet is £210, a gentler climb than the Martingale’s £640 but with the same flaw: the recovery bet is large exactly when the bankroll is smallest.

Oscar’s Grind

Oscar’s Grind, which Wikipedia attributes to Allan Wilson’s 1965 The Casino Gambler’s Guide, aims for exactly one unit of profit per session. Bets stay the same after a loss, rise by one unit after a win, and are trimmed so that a win never overshoots the one-unit target. A session can run for a long time at raised stakes before closing, and a session that never closes is simply a loss. Its appeal is a very high frequency of one-unit wins; its cost is the occasional session that consumes the whole bankroll.

What a system can honestly do

A progression chooses where on the variance curve you sit. Loss-chasers deliver frequent small wins and rare ruin; positive progressions deliver frequent small losses and occasional big cycles. Flat betting at the minimum delivers the smallest expected loss for a given number of hands, because it wagers the least. If you enjoy the rhythm of a progression, treat it as entertainment paid for at the house edge, keep the base bet small enough that the top of the sequence fits inside your session budget, and never confuse a winning streak with a working method.

Frequently asked

Does the Martingale work in blackjack?

No. It raises the chance of a small session win and trades it for an occasional very large loss. Table maximums and finite bankrolls cap the doubling, and the expected loss per pound wagered stays equal to the house edge.

Which betting system is the safest?

Flat betting at the table minimum gives the lowest variance and the lowest expected loss for a given number of hands. Among progressions, positive ones such as Paroli limit the damage of a losing streak, but none reduces the house edge.

Why do betting systems seem to win so often?

Loss-chasing systems produce many small wins and rare huge losses, so most sessions end ahead. The rare loss is large enough to cancel the wins; the frequency of winning sessions says nothing about expectation.

Is a betting progression the same as card counting?

No. Card counting raises bets when the remaining cards favour the player, which changes the expected value of those hands. A progression sets bets by previous results, which carry no information about the next deal.

Sources & editorial notes

By Blackjack World Tour · Updated 2026-10-12 · Owner review pending. Sources were checked on the update date.

Where this guide states a strategy decision without naming rules, it assumes the site profile: six decks, dealer stands on soft 17, double after split, no surrender. Basic strategy cannot guarantee a winning hand.

From theory to the table

Practise the decision.

Ten hands under the stated profile, with the reasoning after every answer.

Open the trainer