Every few weeks a reader emails me the same message with different numbers. They found a betting system, maybe Martingale or the 1-3-2-6, and it worked. Twelve sessions, twelve wins. They want me to confirm what they already believe: that they have beaten blackjack by managing their bets cleverly.
I always answer the same way. The system did not work. You haven't yet hit the session that pays for all the others. This guide explains why, using numbers you can check yourself, so you never have to learn it the expensive way.
The One Equation Every Betting System Runs Into
Every betting progression collides with one fact of casino mathematics: expected value is linear. Your expected loss on any wager is the house edge multiplied by the size of that wager. Because expectation adds, your expected loss for an entire session is:
E[Total Loss] = House Edge × Sum of All Bets
Look at what this equation leaves out. It has no term for the order of your bets. It has no term for whether you raised after a loss or after a win, and none for streaks, "due" outcomes, or how a pattern felt. The only inputs are the house edge and the total amount of money you push across the felt, which professionals call your action or turnover.
A betting system can only rearrange the sizes and sequence of your wagers. Rearranging the terms of a sum does not change the sum's expected value. If each hand carries a 0.5% house edge, every dollar you bet is expected to lose half a cent. No sequence of bets turns that into a positive number.
Key Concept
A betting system changes the shape of your results, meaning how often you win and how large the wins and losses are. It does not change the center. Your expected loss is fixed at house edge × total action. Most progressions also inflate your total action, so they usually make your expected loss bigger, not smaller.
Shape vs. Center: Why Systems Feel Like They Work
This is the part that fools smart people. A betting system can dramatically change the probability distribution of your session outcomes. Martingale, for example, turns a roughly symmetrical spread of results into a lopsided one. You get a large cluster of small wins and a thin tail of catastrophic losses far out to the left.
Most of the time you sample from that big cluster of small wins, so your personal experience says the system works. The tail is rare, which is exactly why it gets overlooked. When it arrives, it erases dozens or hundreds of small wins in one sequence. Averaged across all outcomes, the center of that distribution sits right where the house edge says it should.
If you want the formal treatment of how theoretical expectation turns into real casino revenue, read about house hold and expectation on BlackjackMath.org. I won't reproduce those derivations here. For practical purposes, the casino does not need to win every session. It only needs you to keep betting.
Negative Progressions Under the Microscope
Negative progressions raise your bet after a loss. The idea is that a single win, or a short run of wins, will recover everything you lost plus a profit. These are the most popular systems and also the most dangerous.
The Martingale: Doubling Into the Abyss
The Martingale is the simplest system. Double your bet after every loss. When you finally win, you recover all prior losses plus one base unit, and then you reset. On paper it looks bulletproof, because eventually you have to win a hand.
Blackjack is not a pure coin flip. Blackjacks pay 3:2, and doubles and splits change the payoffs. For a clean model of a Martingale player, though, take the resolved even-money hands and set pushes aside. On that basis the player wins roughly 49.2% of decisions and loses roughly 50.8%. This simplified model is actually a little harsher than real basic-strategy blackjack, where the 3:2 payout and doubling opportunities pull the true edge down to around 0.5%. It still captures the losing-streak risk correctly, and losing streaks are what destroy Martingale players.
The probability of losing seven consecutive hands is (0.508)7, which comes to just under 0.88%. That works out to approximately one Martingale sequence in every 113 to 115. If you play a few hours a week at a full table, you will see that streak many times a year. It is not a freak event.
| Hand in Sequence | Required Bet ($10 base) | Cumulative Loss if Lost | Probability of Losing This Many in a Row | Net Profit if Won |
|---|---|---|---|---|
| 1 | $10 | $10 | 50.8% | +$10 |
| 2 | $20 | $30 | 25.8% | +$10 |
| 3 | $40 | $70 | 13.1% | +$10 |
| 4 | $80 | $150 | 6.7% | +$10 |
| 5 | $160 | $310 | 3.4% | +$10 |
| 6 | $320 | $630 | 1.7% | +$10 |
| 7 | $640 | $1,270 | 0.88% | +$10 |
| 8 | $1,280 (exceeds $1,000 max) | Sequence is dead | n/a | Cannot be placed |
Read the right-hand column again. On hand seven, you are putting $640 at risk to net $10. If you lose it, you would need $1,280 on hand eight to chase that same $10. That is a terrible risk-reward trade, and it is the entire point of the system.
Here is the arithmetic in plain terms, without the full matrix. In the coin-flip model, a seven-step Martingale wins $10 about 99.1% of the time and loses $1,270 about 0.9% of the time. The small wins add up to roughly $9.91 per sequence. The rare wipeout costs about $11.09 per sequence on average. The result is a net expectation of about −$1.17 per sequence. Your average action per sequence is about $73, and $1.17 divided by $73 lands right on the edge built into the model. The linearity rule holds, as it always does.
Table Limits: The Casino's Firewall
Players often tell me Martingale only fails because people don't bring enough money. That is wrong, and the table limit is the reason. On a common $10 minimum / $1,000 maximum table, the eighth bet in the sequence ($1,280) is not allowed. You could walk in with a million dollars and your Martingale would still be capped at seven steps.
The ratio between maximum and minimum is what matters. A 100:1 spread caps a pure doubling sequence at seven hands, whatever your bankroll. Casinos did not set limits by accident. The limit truncates any doubling scheme into a bet with a small, frequent win and a large, guaranteed-eventual loss.
Warning:
Martingale players routinely refuse correct doubles and splits because "the bet is already too big." Every time you skip a basic-strategy double on 11 vs. 6 to protect your progression, you raise the house edge on that hand. The system damages your strategy as well as your bankroll. If you notice this pull in yourself, read our Chasing Losses mistake guide before your next session.
D'Alembert, Fibonacci, and Labouchere: Slower Fuses, Same Bomb
The other negative progressions are mostly attempts to soften Martingale's exponential growth. They change how fast you reach disaster. They don't change where you end up.
- D'Alembert: Add one unit after a loss and subtract one unit after a win. Bet growth is linear instead of exponential, so ruin comes slowly. The system quietly assumes wins and losses will "even out" in count, which is the gambler's fallacy. Even if the counts did balance, you would profit only by the number of units in the sequence, while your total action keeps climbing. At the 0.5%–1.6% edges in our models, the house collects on every one of those dollars.
- Fibonacci: After a loss, move one step forward in the sequence 1, 1, 2, 3, 5, 8, 13, 21, 34, 55. After a win, move back two steps. It grows more slowly than doubling at first, but by step ten you are betting 55 units, and a bad run against you hits the table max as surely as Martingale does.
- Labouchere (Cancellation): Write a list of numbers such as 1-2-3-4. Bet the sum of the first and last numbers. Cross both off when you win, and add the lost amount to the end of the list when you lose. It feels like accounting, which is why analytical players like it. In practice, a losing streak makes the list grow faster than you can cancel it, and the bet sizes climb toward the limit.
All four systems share one feature: they bet the most money when you are behind and emotionally compromised. That is the worst time to make large decisions.
Positive Progressions: Emotionally Safer, Mathematically Identical
Positive progressions reverse the logic. You raise your bet after a win, so you are pressing with "the house's money," and you drop back to base after a loss. I'll give them credit: they are far less likely to empty your account in one sequence. They still don't change the expectation.
The Paroli (Reverse Martingale)
In the Paroli, you double after each win, usually for three steps (1, 2, 4 units), and then reset. You also reset after any loss. Using our 49.2% resolved-hand model, the chance of winning three in a row is about 11.9%. When that happens you bank +7 units. Otherwise your worst case on any single sequence is −1 unit, because each raise is funded by your winnings.
The downside is capped and the occasional upside is satisfying. The expectation, though, is still house edge × total action. Each pressed bet carries the same negative EV per dollar as your base bet. You have rearranged the variance, not removed the edge.
The 1-3-2-6 System
This system bets 1 unit, then 3, then 2, then 6, advancing after each win and resetting after any loss. If all four bets win, you net 12 units, which happens roughly 5.9% of the time in our model. The structure limits your worst sequence:
- Lose bet 1: −1 unit.
- Win bet 1, lose bet 2: +1 − 3 = −2 units, which is the system's maximum sequence loss.
- Win bets 1 and 2, lose bet 3: +4 − 2 = +2 units.
- Win bets 1–3, lose bet 4: +6 − 6 = break-even.
The design is clever: losing on the third or fourth bet still leaves you at or above zero. That clever design still gives you no edge. You are spreading money across four negative-EV bets with a distribution that feels pleasant.
Oscar's Grind
Oscar's Grind tries to earn +1 unit per series. You raise your bet by one unit after a win, but only if the raise won't overshoot the +1 target. After a loss you keep the same bet. It produces a lot of small winning series and occasionally a long, grinding drawdown that is hard to sit through. The arithmetic is the same as everything above: total action times house edge.
| System | Type | Typical Session Feel | Worst-Case Sequence | Effect on Long-Run EV |
|---|---|---|---|---|
| Martingale | Negative | Many tiny wins | −127 units (7 steps, 100:1 table) | None; higher turnover means larger total loss |
| D'Alembert | Negative | Slow drift, long recoveries | Open-ended linear growth | None; inflated action |
| Fibonacci | Negative | Steady until a cold run | Hits table max around step 10–12 | None |
| Labouchere | Negative | Feels controlled | List growth outpaces cancellations | None |
| Paroli (3-step) | Positive | Mostly small losses, occasional +7 | −1 unit | None; same edge per dollar |
| 1-3-2-6 | Positive | Occasional +12 run | −2 units | None |
| Oscar's Grind | Positive | Frequent +1 series | Long drawdowns | None |
| Flat Bet + Basic Strategy | None | Lowest variance | Bounded by buy-in | Minimum possible loss for your action |
Sharp Tip (Marcus Vance):
If you enjoy pressing a hot hand, use Paroli or 1-3-2-6 as entertainment, and size the base unit so that your entire session fits inside a predetermined buy-in. Set that buy-in with our Session Buy-In Units framework. A positive progression with a hard cap is a reasonable way to spend recreational money. A negative progression without one is how people lose a month's rent in twenty minutes.
What Actually Moves the Needle
If bet sequencing can't help, what can? Two things, and both concern the game rather than how you stack your chips.
Lower the Edge You're Multiplying
Because total loss equals edge times action, the cheapest improvement available to you is a smaller edge. Perfect basic strategy at a good table can bring the house advantage to roughly 0.3%–0.5%. A 6:5 blackjack payout or a dealer who hits soft 17 adds meaningful edge before you place a single chip. Run any table's rules through our Table Grader tool before you sit down. Choosing a good table does more for your bankroll than any progression ever written.
Control the Action You Feed the Edge
The other lever is total action. Flat betting a modest unit with flawless basic strategy gives you the lowest expected loss per hour for the entertainment you're buying. Pair that with firm session boundaries. Our guide on Stop-Loss and Win Goals shows how to protect a bankroll without pretending that stopping changes the math. Stop-losses don't create an edge either. What they do is stop tilt from inflating your action.
Why Advantage Players Vary Their Bets and System Players Shouldn't
"Card counters spread their bets too," someone always says. "Isn't that a betting system?" No. The difference is the core of this article.
A card counter raises the bet because the remaining shoe composition has shifted. When the true count is high, the undealt cards are rich in tens and aces, and the player's expectation on the next hand is measurably positive. The counter bets more at those moments because the next hand is a better bet. That decision rests on information about cards that have not been dealt yet.
A Martingale player raises the bet because the previous hand lost. Past win/loss results carry no information about the next hand's expectation. The cards behind the discard tray matter. Whether you won or lost with them does not. Advantage players change bet size only in response to the statistical state of the shoe, and they size those bets with methods like the ones covered in Kelly criterion and betting sizing on BlackjackMath.org, wagering a fraction of bankroll proportional to their measured edge.
Key Concept
A legitimate bet spread puts more money into positive-EV hands and less into negative-EV hands, which changes the weighted average of your edge. A betting progression puts more money into hands that are just as negative as all the others, so the edge stays the same and you only add risk.
A Practical Checklist for Your Next Session
- Grade the table before you sit down. Rules set your edge.
- Play perfect basic strategy, including every double and split, regardless of bet size.
- Flat bet unless you are counting. If you want the fun of a progression, use a capped positive one with recreational money only.
- Never raise a bet because you lost the last hand.
- Set your buy-in, stop-loss, and win goal in units before the first card comes out.
- Treat any long streak of system "wins" as a sample from the frequent part of the distribution, not as proof that the system works.
Bottom Line
Expected value is linear, so your expected loss is the house edge multiplied by everything you wager, in whatever order you wager it. Martingale and its relatives trade frequent small wins for rare, devastating losses that table maximums make certain over time, and they usually increase your total action as well. Paroli, 1-3-2-6, and Oscar's Grind cap your downside and feel better, but they carry the same negative expectation per dollar. The only bet variation that changes your edge comes from information about the cards left in the shoe. Otherwise, choose good rules, play perfect basic strategy, flat bet, and keep firm session limits.
