Blackjack Betting Systems: Martingale, Paroli & Flat Betting

Betting systems are rules for changing the starting stake between rounds. Martingale increases it after a loss, Paroli increases it after a win, and flat betting keeps it unchanged. These rules can produce very different balance movements without improving the underlying value of the next wager.

This guide compares flat betting, Martingale, Paroli and 1–3–2–6 using complete stake sequences and clearly stated reset rules. It also explains why real blackjack results, including pushes, naturals, doubles and split hands, cannot simply be squeezed into an ordinary win-or-loss model.

All dollar amounts, budgets and table limits below are invented teaching examples, expressed in Canadian dollars. They are not casino offers, recommended stakes or predictions. The probability example uses a fictional two-outcome game, not measured blackjack results.

What a betting system actually controls

A system needs more than a name. Specify its base unit, when it advances, when it resets, how it treats a zero result and what happens when the next stake is unavailable. Otherwise two people can describe different methods as the same system.

“Positive progression” means increasing after wins; “negative progression” generally means increasing after losses. Neither label describes whether the overall session earns a profit. A positive progression can lose money, and a negative progression can show small wins before a much larger loss.

Betting systems versus basic strategy

Blackjack basic strategy addresses how to play a hand under stated rules. A stake progression addresses how much to start with. Doubling the next starting stake after losing does not identify the correct action on a soft eighteen or a pair of eights.

Keep those two decisions separate in a practice record. First identify the table rules and the appropriate playing action. Then record the amount actually at risk. The blackjack cheat sheet states its reference assumptions; changing the staking pattern does not expand that chart to every variant.

A win also does not prove the preceding decision was correct. A poor action can win a particular hand, while the better action can lose. Judging a progression from a handful of favourable outcomes makes the same mistake on a larger scale.

Read betting systems versus basic strategy for a side-by-side explanation, scoped hand examples and ledgers that distinguish a starting-stake increase from a double down.

The teaching model and its limits

The sequence examples initially assume one starting bet per round, an ordinary win paying profit equal to that bet, and a loss costing the full bet. A $5 ordinary win therefore adds $5 profit; the returned $5 stake is not another $5 of profit.

W means an ordinary even-money win and L means a full loss. Each example begins at a fresh cycle. A cycle ends at its stated reset point, which may occur before the session ends. Where a push occurs, our teaching convention is to repeat the same stage without changing the running net result.

These are accounting assumptions, not a claim that every blackjack round has two outcomes. The later sections explain why actual settlement must be recorded separately. Consult the payouts guide when distinguishing profit, returned stake and total return.

Flat betting: a consistent starting unit

Flat betting uses the same initial stake each round. With a $5 unit, the ordinary sequence L–L–L–W produces −$5, −$5, −$5 and +$5, leaving a $10 net loss. The four initial bets total $20.

The main accounting benefit is consistency: the next starting amount does not jump because the previous result was disappointing. That makes a session ledger easier to read. It does not turn a negative expectation into a positive one or promise a small total loss.

Flat starting bets also do not necessarily mean identical total exposure. A blackjack double or split can add money during the round. If the budget cannot accommodate the permitted action, the original unit was not the full amount potentially needed.

Count repeated rounds too. Twenty ordinary $5 losses cost $100 even though no progression is involved. A small unit is an amount per bet, not a session spending limit.

Keep the unit fixed when making a comparison. Quietly increasing a flat stake from $5 to $20 after a loss is no longer the original flat example. Reducing the unit is a new choice too. Record that change instead of presenting the combined results as evidence for a rule that was not followed.

Read the flat betting guide for full ledgers showing fixed initial stakes, extra blackjack commitments, percentage-based alternatives and the difference between action and spending.

Martingale: doubling after an ordinary loss

In the basic Martingale used here, start with one unit, double after each full ordinary loss and reset after an ordinary win. A completed loss-then-win cycle recovers the preceding losses and produces one base unit of profit, provided every required bet can be placed and the stated payouts apply.

For L–L–L–W with a $5 unit, the bets are $5, $10, $20 and $40. The net result is −$5 − $10 − $20 + $40 = +$5. Total initial stakes are $75. Compare that with the $20 of action and $10 loss in the flat example: the favourable final result required much greater exposure.

The recovery statement is conditional. A sequence that stops on a loss has not recovered anything. Calling the last loss “unfinished” does not remove it from the balance or guarantee that the next required amount will be available.

How quickly the Martingale requirement grows

The table tracks a $5 base through six successive full losses. Read both the cumulative loss and the next required bet. They are different quantities, and enough money to cover the previous bet does not establish enough money for the next one.

After six losses, an invented $500 starting budget has $185 left. That is a positive balance, but it cannot fund the required $320 bet. Adding more money would change the spending boundary rather than complete the original budgeted plan.

With base unit b, n consecutive losses total b × (2ⁿ − 1), and the next required stake is b × 2ⁿ. Six losses at $5 therefore cost $315. That single failed run cancels the profit from sixty-three earlier $5 winning cycles.

Table limits can stop a sequence too

A hypothetical table maximum of $200 blocks the $320 step even if sufficient money remains. Starting with $1,000 would leave $685 after the same six losses, but the required next bet still exceeds that invented maximum.

MGM’s blackjack guide explains that table minimums and maximums are posted. Check the actual limits and their application before interpreting any stake sequence. Do not assume that changing tables, spreading bets or adding hands reproduces the same model.

A capped progression is a different rule. If you replace the unavailable $320 step with $200 and win an ordinary even-money bet, the accumulated $315 loss becomes a $115 loss. The cap limits that particular next stake; it does not preserve the original one-unit recovery calculation.

Paroli: increasing after wins

Paroli reverses the trigger: increase after winning rather than after losing. For a precise comparison, this guide uses a three-win cycle with stakes of one, two and four units. Reset after any ordinary loss or after the third consecutive ordinary win.

At a $5 unit, the stakes are $5, $10 and $20. A loss at the first stage costs $5. Win then lose gives +$5 − $10 = −$5. Two wins then a loss gives +$5 + $10 − $20 = −$5. Three wins give +$5 + $10 + $20 = +$35.

Under those limited assumptions, every unsuccessful completed cycle loses one unit and a three-win cycle earns seven units. That is not a one-unit session loss limit: repeated unsuccessful cycles keep subtracting money.

The largest starting stake in this version is four units. Other versions may continue longer, which changes their exposure. Money won earlier is part of the player’s current balance; describing the next bet as “house money” does not make its loss inconsequential.

For a direct comparison, read Martingale vs Paroli. It applies both rules to the same outcomes and compares action, reset points, finite budgets and drawdown.

The 1–3–2–6 sequence

Our 1–3–2–6 example advances through starting stakes of one, three, two and six units after ordinary wins. It resets to one unit after a loss or after completing all four winning stages. Pushes repeat the current stage.

The decrease from three to two units is intentional: this is a fixed sequence, not simple doubling after every win. With a $5 unit, the four possible stakes are $5, $15, $10 and $30.

The largest terminal cycle loss shown is two units, while the largest individual starting stake is six units. Those are not contradictory: earlier winning stages offset part of a later loss. Neither number is a guaranteed limit for an entire session containing many cycles.

A successful four-win cycle stakes twelve units in total and earns twelve units of ordinary profit. Reporting only the $60 result without the $30 final stake hides information needed to assess the sequence.

For example, two W–L cycles lose four units altogether. One later W–W–L cycle earns two units, leaving a two-unit session loss. A positive last cycle does not make the whole session profitable. Calculate the running total from the first bet rather than restarting the profit counter at a convenient point.

The full 1–3–2–6 guide follows every cycle ending, repeated cycles, pushes and the extra exposure from blackjack doubles and splits.

Why frequent winning cycles can still lose overall

Consider a deliberately fictional game with independent rounds: every bet wins with probability 49%, loses with probability 51%, and pays even money. Each dollar wagered has an expected loss of two cents. These figures are chosen to explain arithmetic; they are not blackjack probabilities.

Limit a Martingale to three attempts at one, two and four units. Stop at the first win or after three losses. Winning paths are W, L–W and L–L–W, each earning one unit. The only losing path is L–L–L, costing seven units.

The losing path has probability 0.51³ = 0.132651, or about 13.27%. A cycle wins about 86.73% of the time. Yet its expected net result is 0.867349 × 1 − 0.132651 × 7 = −0.061208 units.

Expected total action per cycle is 1 + 2 × 0.51 + 4 × 0.51² = 3.0604 units. Two per cent of that action is 0.061208 units, matching the expected loss. A high proportion of small successful cycles has not removed the cost of the losing ones.

For the three-win Paroli in the same fictional game, three wins occur with probability 0.49³ = 0.117649. The other completed paths lose one unit. Its expected result is 7 × 0.117649 − 0.882351 = −0.058808 units, with expected action of 2.9404 units.

Both retain the same 2% expected loss relative to expected action. Comparing their per-cycle losses alone would ignore different average amounts wagered and different cycle lengths. The odds and house edge guide explains why expectation and one session’s outcome are separate measures.

Blackjack cards and streaks are different information

Actual shoe blackjack is not the independent two-outcome model above. OLG PlaySmart explains that removing cards changes the composition of the remaining cards. Rules, information and playing decisions matter when describing an advantage.

A list of recent wins and losses does not measure that composition. “I lost three hands” is not equivalent to knowing which cards remain. A prescribed loss progression therefore cannot claim a better next-hand expectation merely because it has moved to a higher step.

The narrower mathematical point is this: when the next wager retains a negative expected return per dollar, choosing a larger stake from previous results changes the amount exposed, not the sign of that return. This does not pretend that every conceivable card-information method is the same as Martingale.

Pushes, naturals and actual round settlement

A push returns the bet without ordinary profit. Under our stated convention it leaves the progression at the same stage. It adds a round and time to the cycle, so a sequence described as “three wins” can take more than three actual rounds to complete.

A natural blackjack can pay differently from an ordinary win. On a $20 stake, 3:2 produces $30 profit while 6:5 produces $24. Neither is the $20 profit assumed in the ordinary even-money model. Use the posted payout rather than assigning every win the same unit value.

A surrender result also needs separate treatment. A half-stake loss is not a full loss, and blindly advancing a full-loss ladder changes its arithmetic. Our surrender guide covers that settlement; a progression must explicitly define how it records it.

Insurance is another wager with its own result. Keep it separate from the main hand instead of labelling the whole round from only one outcome. See insurance and even money for the distinction.

Doubles and splits change the money at risk

Suppose a $10 starting hand is doubled to $20 and loses. The actual loss is $20. If the next starting bet is $20 and wins ordinarily, the two-round net is zero, not the $10 profit predicted by doubling only the original $10 starting amount.

Recalculating a recovery target after every settlement produces a different system from the simple ladder. It may require an amount above the next table limit or outside the chosen budget. The double down guide explains the additional stake within a round.

After splitting a $10 pair into two $10 hands, one ordinary win and one full loss produce a zero net round result but $20 of initial hand stakes. Further doubles could increase exposure again. A record saying only “push” would conceal the extra money committed.

The splitting guide explains hand and stake separation. Record each result, then calculate the net round result before applying any expressly defined progression rule. Do not switch between per-hand and per-round accounting halfway through a sequence.

Side bets, slots and restricted stakes

A sequence calculated for ordinary even-money wins does not transfer unchanged to a wager with many losing rounds and occasional larger payouts. A winning payout may be too small to recover prior losses or much larger than the assumed unit return.

Blackjack side bets have their own pay tables and probabilities. Slots involve different prizes and features again. A label such as Martingale does not calculate their expected return, and increasing the stake does not make a jackpot due.

Account restrictions may also set an eligible maximum stake or exclude particular activity. Read the applicable bonus terms before comparing an unrestricted teaching model with a restricted balance. A staking sequence does not override those conditions.

How to keep a useful practice ledger

Use a paper or spreadsheet exercise before interpreting a system from a short demonstration. Record the cycle number, starting balance, stage, initial stake, extra stakes, net settlement, ending balance and proposed next stake. Keep the actual outcome distinct from the rule that would follow it.

Include failed and interrupted cycles. If a demonstration resets whenever the sequence becomes unaffordable, carry that loss into the session total. Starting a fresh row does not restore the missing money.

Compare methods using the same outcome sequence and disclose total action, maximum stake and ending balance. The earlier L–L–L–W examples deliberately share an outcome sequence, but their four bets expose different amounts. Neither alone establishes which ending balance is typical.

For a random simulation, also disclose the game probabilities, payout assumptions, number of sessions, stopping rules and whether every failed sequence was retained. A screenshot of one successful run cannot answer those questions.

Keep observation and hindsight separate. If a sequence was selected because it happens to contain the four wins needed for a successful cycle, that selection cannot establish the chance of encountering it beforehand. Likewise, showing only players who finished a target omits those who reached a spending limit first.

A practical ledger should reconcile the ending balance to actual money added and removed. If an exercise starts with $100, adds $50 and ends with $120 without withdrawals, the net gambling result is a $30 loss. The $20 increase over the original displayed balance is not profit because another $50 was supplied.

Budget and stopping rules belong outside the ladder

Choose a spending boundary independently of the next proposed stake. If the sequence asks for more than the remaining allowance, it has reached that boundary. A smaller replacement bet or a new deposit changes the plan and should be described honestly.

The session budget calculator can organise an amount and time limit. Its output is not a prediction that the balance will last for a particular number of wins or losses. Money reserved for other expenses is not a reserve for completing a progression.

A profit target can end a session, but it does not remove earlier risk. A loss limit can stop further bets while leaving a real loss. Neither stopping rule makes the wager itself more favourable. If keeping to boundaries becomes difficult, our gambling help page lists support and practical controls.

Return to the guides directory for the relevant hand rules and settlement explanations.

Common betting system questions

Does a larger bankroll make Martingale certain to work?

No. It allows more steps before money becomes the constraint, but any finite budget has a point where the required next stake cannot be funded. A table maximum can stop the sequence earlier. More permitted steps also mean a larger possible accumulated loss.

Is Paroli the same as basic strategy?

No. Paroli changes starting stakes after wins. Basic strategy chooses playing actions from the hand, dealer card and rules. Following one does not automatically supply the other, and neither guarantees a winning session.

Can a progression guarantee that a loss is recovered?

Only a conditional calculation can describe recovery after a particular future win. It cannot guarantee that win, the ability to place the required stake or the absence of limits. An incomplete sequence remains part of the actual session result.

What is the clearest way to compare systems?

State the rules, show every terminal path, include total stakes and retain failed cycles. Separate starting stakes from additional blackjack exposure. Return to the guides directory for the relevant hand rules and settlement explanations.

What happens to a progression after a push?

In this guide’s examples, a push repeats the same stage and changes the net result by zero. It still uses a round and time. Other versions may define a different convention, so state the rule before comparing results rather than changing it after seeing the outcome.

Is the 49% win probability a measured blackjack figure?

No. The example is a fictional independent game that wins 49% of bets, loses 51% and pays ordinary even-money profit. Its 2% expected loss explains the arithmetic of the capped progressions. Actual blackjack includes pushes, different payouts, extra wagers and changing card composition; those example probabilities must not be presented as its statistics.

Why does a doubled blackjack loss change the recovery calculation?

A $10 starting hand doubled to $20 can lose $20. A following $20 ordinary win then leaves the two rounds at zero, not a $10 profit. Record the full accepted stake and actual settlement. Recalculating a new target from that loss is a different rule from doubling only the preceding starting stake.

Can I count a natural blackjack as an ordinary one-unit win?

Not when its posted payout differs. At 3:2, a $20 natural earns $30 profit; at 6:5 it earns $24. Record that profit separately from the returned stake, then apply the progression’s explicitly chosen reset rule. Labelling every win as plus one unit loses the actual settlement information.

References

Mathematical examples and available reference pages reviewed November 1, 2026. The BetMGM terminology article was last checked October 31 and could not be reopened during this review. Worked sequences and the fictional probability model are calculations for this guide; they are not results reported by these organisations. Ontario and US sources explain the game, not Alberta availability or a particular table’s rules.

1. Wizard of Odds: The Truth about Betting Systems — mathematical analysis of progressions and their limitations.
2. Wizard of Odds: Martingale Betting System — doubling after losses and finite-budget considerations.
3. BetMGM Canada: Understanding Blackjack Betting Systems — terminology for Paroli and 1–3–2–6 progressions.
4. OLG PlaySmart: Blackjack Odds — card removal, rules and playing decisions.
5. MGM GameSense: Guide to Blackjack — table limits, payouts and additional stakes.