Math debunk
Betting Systems Debunked
Every common roulette and blackjack betting system, with the math of why each one fails.
No system changes the house edge - that's a mathematical fact, not an opinion.
- No betting system changes the house edge. The expected value per bet is fixed by the game's math, not the bet sequence.
- Negative progressions (Martingale, Fibonacci) are bankroll bombs. They work right up until they don't, then they cost you everything.
- Positive progressions (Paroli) are safer but still -EV. Same expected outcome as flat-betting, just with different variance.
The fundamental math: nothing changes the house edge
Every casino game has a fixed house edge built into the math model. European roulette: 2.70%. American roulette: 5.26%. Blackjack with basic strategy: 0.4-0.5%. Each individual bet you place is a negative-expected-value transaction - the casino expects to keep, on average, the house-edge percentage of your wager.
A betting system rearranges the size and order of your bets. It does not change the underlying probability of each bet's outcome. Spinning a roulette wheel after losing five reds in a row gives you exactly the same probability of red on the sixth spin (47.37% on European wheels) as it did on the first. The wheel has no memory; the dealer has no memory; the universe has no memory.
This is the gambler's fallacy in formal terms: the belief that past random outcomes affect future random outcomes. Every betting system below is, mathematically, a complicated application of the gambler's fallacy.
The five most common betting systems
Martingale
Negative progression How it works Double your bet after every loss. When you eventually win, you recover all prior losses plus one unit of profit.
Why it feels like it works The Martingale "guarantees" a one-unit win on every winning streak. To a player who hasn't done the math, it looks like a bulletproof system - surely you must win eventually?
Why it actually doesn't Two reasons. First, table maximum bets prevent you from doubling past a certain point - after roughly 8 consecutive losses at a £5 minimum table, you hit a £640 bet that the table will refuse. Second, the bankroll required to survive a losing streak grows exponentially. A 10-loss streak (rare but not absurdly so) requires £5,120 just to make the eleventh bet. The probability of going broke in a long enough session approaches 100%.
Worked example Start with a £5 bet on red. Lose → bet £10. Lose → £20. Lose → £40. Lose → £80. Lose → £160. Lose → £320. Lose → £640. After 8 losses you're down £1,275, the next bet would be £1,280, and most tables cap at £500 or £1,000. You walk away £1,275 down on a 0.4% probability event that the system promised was "impossible".
Fibonacci
Negative progression How it works Increase your bet according to the Fibonacci sequence (1, 1, 2, 3, 5, 8, 13, 21, 34, ...) after each loss. After a win, move back two steps in the sequence.
Why it feels like it works Slower bet growth than Martingale, so feels safer. The Fibonacci sequence has a mathematical mystique that lends superficial credibility - surely a system named after a famous mathematician must work?
Why it actually doesn't The bet growth is slower but the underlying problem is identical: a long losing streak still exhausts your bankroll or hits the table cap. Fibonacci just delays the inevitable by a few steps. The expected value per bet is still negative (the house edge is unchanged). A system that doesn't change the expected value cannot turn a -EV game into a +EV one.
Worked example Start with £5 bets. Lose 10 in a row following Fibonacci progression. Bet sizes: £5, £5, £10, £15, £25, £40, £65, £105, £170, £275. Total lost: £715. To make the eleventh bet you'd need £445 (5+11=16 units × £5 base scale, actually grows faster), and you're still expected to lose 47.4% of the time on red/black even-money. The "system" only works if you never lose 6 or 7 in a row - but in a long enough session, you will.
D'Alembert
Negative progression How it works Increase your bet by one unit after a loss; decrease by one unit after a win. Based on the false intuition that wins and losses "balance out" over time.
Why it feels like it works Bet sizes grow linearly rather than exponentially, so it feels much safer than Martingale. The system implicitly assumes that long losing streaks are "corrected" by wins - a version of the gambler's fallacy dressed up as method.
Why it actually doesn't The "balance out" assumption is the gambler's fallacy. Roulette and other RNG games have no memory - a long losing streak does not increase the probability of subsequent wins. D'Alembert produces small wins frequently and occasional large losses; over enough sessions, the large losses dominate and the player ends down by the house edge times total wagered. The system is harmless in the short run and ruinous over time.
Worked example Start at £5. Win → £4. Win → £3. Lose → £4. Lose → £5. Lose → £6. Win → £5. Over 100 spins of European Roulette (2.7% edge) at average £5 bets, expected loss is about £13.50. D'Alembert produces a more pleasant-feeling distribution (more small wins, fewer big swings) - but the long-run expected loss is identical to flat-betting the same average stake.
Labouchère (Cancellation)
Cancellation How it works Write down a sequence of numbers (e.g. 1-2-3-4). Bet the sum of the first and last numbers. If you win, cross both off. If you lose, add your bet to the end of the sequence. Continue until all numbers are crossed off.
Why it feels like it works The structured "cross off the numbers" approach gives the system a sense of progress and control. Players track their session against the sequence, which creates a satisfying gamified feel and the illusion that you're working toward a defined goal.
Why it actually doesn't When you lose, the sequence grows; when you win, it shrinks. A losing streak makes the sequence grow faster than wins can shrink it, leading to ever-larger bets to cancel the inflated sequence. The same death-spiral as Martingale, just dressed up in a more complex bookkeeping system. Mathematically identical expected value per bet.
Worked example Start with 1-2-3-4 (goal: win 10 units total). First bet: 1+4 = 5. Lose → sequence becomes 1-2-3-4-5. Next bet: 1+5 = 6. Lose → 1-2-3-4-5-6. Now bet 1+6 = 7. After five consecutive losses, your sequence is 1-2-3-4-5-6-7-8-9-10-11 and your next bet would be 12 - you're down 33 units and the system says you need to win 11 more bets in the right order to "complete" the sequence.
Paroli (Reverse Martingale)
Positive progression How it works Double your bet after each win, up to a chosen target (e.g. 3 wins in a row), then reset to the base bet. After a loss, reset.
Why it feels like it works Lower-risk than negative progressions - you're only ever risking your initial small bet plus accumulated winnings. Wins compound during a hot streak. Feels like "letting it ride" with a system.
Why it actually doesn't Less catastrophic than Martingale but mathematically still -EV. The expected value per bet is unchanged (the house edge applies to every bet regardless of progression). Paroli produces a session distribution with more small losses (the common case: a quick first-bet loss) and rare big wins (a 3-in-a-row hot streak). Over enough sessions, the expected outcome is again negative by the house edge.
Worked example Start at £5. Win → bet £10. Win → bet £20. Win → reset, take £35 profit. The probability of three roulette reds in a row (even money) is roughly 11%. The probability of an immediate loss (89% over the same three-bet horizon) loses £5. Run this 100 times: expected outcome is roughly £350 of "hot streak" wins (11 × £35) and £445 of single-bet losses (89 × £5). Net: -£95 over 100 sessions, very close to the -£100 you'd expect from flat-betting the same volume at the same house edge.
What actually changes your expected value
If betting systems don't work, what does? Three things that genuinely move expected value, in order of impact:
- Game selection. Playing blackjack with basic strategy (0.5% house edge) instead of slots (3-6% house edge) cuts your expected loss by roughly 8x at the same wagered volume. RTP by game →
- Bonus selection. A no-wagering free spin offer has positive expected value; a 100% match with 50× wagering has heavily negative expected value. Bonus EV methodology →
- Bankroll management. Bet sizing matters not because it changes EV, but because it controls variance. Smaller bets relative to bankroll mean longer sessions and lower probability of ruin. Limits don't change the long-run outcome but they smooth the path.
What about advantage play?
Genuine advantage play exists in three narrow contexts:
- Card counting at blackjack can flip a 0.5% house edge into a 0.5-1.5% player edge, but requires perfect basic strategy, accurate count tracking and aggressive bet sizing - and casinos will back you off the table once they detect it. Online with infinite-shoe RNG: impossible. At live dealer tables: theoretically possible with shoe-tracking software, practically futile because shoes are reshuffled before they\'d become countable.
- Video poker game-selection at full-pay tables (9/6 Jacks or Better, 10/7 Double Bonus Poker) with optimal strategy can produce a player edge of 0.5-1%. Requires precise play and disappearing game variants - most casinos run shorter paytables.
- Sports betting line-shopping and value-finding can produce edge against retail bookmakers, but this is a craft requiring years to develop, not a "system" you can read about and apply.
None of these is a betting system in the Martingale/Fibonacci sense. All require specific games, specific conditions and substantial skill. If a betting system is being sold to you as a path to consistent profit, it isn\'t one.
The honest bottom line
Casino games are designed to be entertainment with a negative expected value. The house edge is the cost of entry - the price you pay for an evening of play. Betting systems don't change that cost; they just redistribute when you pay it.
If you want to gamble, the rational approach is the same as the entertainment approach: set a budget you can afford to lose, pick the lowest-house-edge games available, play for a fixed time, and walk away when the budget or time runs out. Anyone selling you a system that promises more is selling you the gambler\'s fallacy with extra steps.
Related guides
Written by Daniel Chen, Strategy & US Market Lead.
