The Short Answer
No system won Every scheme we tried — increase-on-loss, press-on-wins, ladders, resets, hybrids — lost money at exactly the same rate: the house edge, on every dollar wagered. A betting system is a variance dial, not an edge dial. It changes the shape of your wins and losses, never the average. The only things that move your real odds are how near your target is and how boldly you bet — and even tuned perfectly, the house still wins most nights.
The Two Iron LawsWhy nothing we built could work
Two principles explain every result on this page. Once you see them, the whole search ends.
1. The edge is untouchable
Each individual bet returns −(house edge) × (bet size) on average, no matter how you chose that size. You get to pick each bet's size from the past — never the future — so the sizes can't correlate with the outcomes. Add up every bet and your expected loss is always edge × total wagered. In our simulations, "net ÷ wagered" landed on the house edge for all ten systems, on both games, every single time.
2. Reaching a target is capped at start ÷ target
Turning $500 into a bigger number is a gambler's-ruin race: reach the target before you hit zero. In a perfectly fair game the best possible chance is simply start ÷ target — 50% to double, 33% to triple, 20% for 5×. No staking system raises that ceiling; the house edge only pushes you below it. The way to get close to the ceiling is to bet bold (few big bets, so less money is exposed to the edge). Timid betting spreads thousands of tiny bets across the felt and the edge grinds every one.
The unavoidable trade-off: turn variance up (bold, press winners) to maximize your shot at a big target — you'll bust more too. Turn variance down (timid, small bets) to win a little more often — paid for by rare deep losses. You genuinely choose which. You never change the edge.
The Deciding ChartIt's the target, not the system
Here is the result that ends the debate. For any given goal, the fancy systems cluster with plain flat betting — and the timid ones fall off a cliff as the target recedes. What moves the needle is how far you're reaching.
Notice what this means: no system beat "just betting normal." To triple your money, a flat $100 bet reaches the goal ~31% of the time — basically tying the theoretical best of 33% and beating every progression we designed. The systems that start at $5 and build actually do worse, because on average they bet small, play more hands, and expose more money to the edge. "Bold" just means big bets relative to your bankroll — and a flat $100 bet on a $500 stack already is bold.
Everything We TriedThe full catalogue
Grouped by shape. Every one was tested on both goals: grind (leave ahead on a 500-hand night) and walk-away (turn $500 into a target and quit).
Negative progressions — bet up after losses
The instinct behind most systems: chase losses back. They book frequent small wins, so they "feel" like they work — but the rare losing streak, betting large, gives it all back exactly to the edge.
| System | Shape | Grind: % nights up | Walk-away: P(triple) |
|---|---|---|---|
| Your +20% / −33% rule | negative | 53.1% | 16.1% |
| D'Alembert (+$5 / −$5) | negative | 21.8% | — |
| Martingale (double on loss) | negative | 25.5% | — |
| Your $5,6,7,9,12,15,20 ladder | negative | ~53% | low |
Verdict: your gentle +20/−33 was the best of the whole field for the grind goal — profitable on ~53% of nights. But that 53% is bought with occasional deep-loss nights, and it's near-useless for chasing a big target.
Positive progressions — bet up after wins
Your later, sharper instinct: press winnings, reset on a loss. This is the right shape for a walk-away goal because it manufactures the variance a distant target needs.
| System | Shape | Grind: % nights up | Walk-away: P(5×) |
|---|---|---|---|
| Your +50% / reset after 3 wins | positive | 47.1% | 3.8% |
| Your hybrid (+75% / halve / cut at 5) | positive | 18.9% | 16.3% |
| Tuned: double / reset after 8 | positive | — | 17.9% |
| Anti-martingale (uncapped press) | positive | 37.5% | 15.4% |
Verdict: your press-on-wins direction was correct, but your original brakes were too tight — resetting after just 3 wins caps the upside. Tuning it (press harder, ride to ~8 wins) took it from a 3.8% shot at 5× up to ~18%, right against the ceiling.
Flat & bold benchmarks
The controls that quietly won. A big flat bet is already "bold," and bold play sits at the ceiling for every target on every game.
The One Thing That WorksAdvantage play — changing the edge itself
This actually wins Every scheme above is a staking system — a rule for sizing bets from past results — and those are mathematically proven never to beat a negative-edge game (Dubins & Savage, "How to Gamble If You Must," 1965). The methods that genuinely win do the one thing staking can't: they change the edge, using information or a physical flaw.
Card counting
Track the ratio of high to low cards remaining; the deck genuinely shifts in the player's favor when it's rich in tens and aces. Bet big only then and blackjack flips to a ~0.5–1.5% player edge. It's legal — but it's exactly why casinos shuffle early, use continuous shufflers, and eject counters. This is the one method that makes blackjack positive, which is why we excluded it from our test on purpose.
Physical roulette prediction
They built the first wearable computer to time the wheel and ball and predict the landing region — reportedly a ~44% edge on a section of the wheel. This beats roulette with physics, not betting patterns. (Such devices are now illegal in most jurisdictions.)
Biased-wheel exploitation
He clocked the wheels, found one with a mechanical bias toward certain numbers, and bet those numbers to win a fortune. No system — just a flawed wheel and the patience to find it.
Edge sorting
He exploited microscopic printing asymmetries on card backs to read high cards in baccarat, winning $10M+. Courts later ruled it cheating and made him hand it all back — a reminder that the line between advantage play and cheating is where casinos and courts draw it.
The Kelly criterion
The one bet-sizing rule that genuinely matters — but only after you already have an edge. It tells you the fraction of your bankroll to bet to maximize long-run growth without going broke. It can't create an edge; it can only grow a real one (which is why card counters pair counting with Kelly).
The through-line: every winning method changes the odds — through information or a flaw. Not one of them is a pattern of bet sizes. Everything you and your cousin ever built is a pattern of bet sizes. That's the whole difference, and it's why the discussion can end here.
If You Ever Actually Do ItThe least-bad way to play
- Pick blackjack, not roulette. A ~0.4% edge versus 5.26% is 5–10× better outcomes on every metric we measured.
- Aim modest. Doubling is nearly a coin flip (~47%); 5× is a 1-in-5 long shot. The nearer the target, the higher the ceiling.
- Bet bold and flat, then leave. Fewer big bets expose less money to the edge. Walk the instant you hit your number — that's the one move that genuinely helps, because it stops the edge from grinding.
- Know the honest odds. Even played perfectly, "double my money" means risking $500 to win $500 on about a 47/53 proposition — a coin flip that pays a little less than fair. Fun for a night; still negative in the long run.