Blackjack Betting Strategy: Every System Tested on the Same Ten Hands
No blackjack betting strategy changes the house edge. We test Martingale, 2-1-2, 1-3-2-6, Paroli, Oscar's Grind and D'Alembert on the same ten hands.
A blackjack betting strategy decides how much you put on each hand. Basic strategy decides how you play the hand once it is dealt. The second one changes the house edge; the first one does not, and this page shows why with numbers rather than assertion. Every system below is run through the same ten-hand sequence with a £5 base unit, so you can see what each one does to your stake, your result and your risk.
All house edge figures assume basic strategy on UK six-deck rules: dealer stands on soft 17, double on any two cards, double after split allowed, no surrender, no hole card, blackjack pays 3:2. That gives about 0.5%. House edge is a long-run average and no betting pattern guarantees a win.
What a betting strategy can and cannot do
The expected loss on any blackjack session is a single multiplication:
Expected loss = house edge x total amount wagered.
The house edge is set by the rules and by how well you play the hands. The total wagered is set by your stakes and how many hands you play. A betting system only touches the second number, and it touches it by moving money between hands, not by removing it. Every pound you bet carries the same 0.5% expected cost whether it is the first pound of a flat bet or the 128th pound of a Martingale.
What a betting system does change:
| Property | Changed by a betting system? | How |
|---|---|---|
| House edge per hand | No | Set by rules and play |
| Expected loss per £ wagered | No | Always edge x stake |
| Total wagered | Yes | Progressions raise the average stake |
| Variance | Yes | Larger swings in both directions |
| Shape of results | Yes | Many small wins and rare large losses, or the reverse |
| Money at risk on one hand | Yes | Can be many times the base unit |
So the honest framing is this: a betting system is a choice about what kind of session you want, not a method for winning more. A negative progression (raise after a loss) buys frequent small wins with occasional large losses. A positive progression (raise after a win) buys frequent small losses with occasional larger wins. Flat betting sits in the middle and keeps the total wagered, and so the expected loss, as low as it can be for the number of hands played.
Here is the test sequence used for every system on this page. Ten hands, five wins and five losses, with pushes, blackjacks, doubles and splits left out so the stake sequence is easy to follow:
L, L, W, W, W, L, W, L, L, W
Five wins and five losses at even money is a break-even result for flat betting. Watch how far each system moves from zero on exactly the same cards.
Flat betting and unit size
Flat betting means the same stake on every hand. On the test sequence at £5 a hand you wager £50 and finish exactly level. The expected loss on that £50 at 0.5% is 25p. That is the figure every other system has to be measured against, because it is the lowest total wagered for ten hands.
The one decision flat betting asks of you is the unit size, and that should come from the session bankroll rather than from habit. A unit of 1% to 2% of the session bankroll (50 to 100 units) survives a normal losing run including a few doubles and splits.
| Base unit | 50-unit session bankroll | 100-unit session bankroll | 200-unit session bankroll |
|---|---|---|---|
| £1 | £50 | £100 | £200 |
| £2 | £100 | £200 | £400 |
| £5 | £250 | £500 | £1,000 |
| £10 | £500 | £1,000 | £2,000 |
| £25 | £1,250 | £2,500 | £5,000 |
Read it the other way round to set a stake: with £100 for the evening, the unit is £1 to £2, not £5. Two hundred units is for anyone running a progression, for reasons the next section makes clear.
Progression systems one by one
Each system gets the rule, the ten-hand table and a note on where it breaks. Running result is after the hand is settled.
Martingale
The rule: double your stake after every loss and go back to the base unit after a win. The idea is that the eventual win recovers every loss in the run plus one unit.
| Hand | Result | Stake | Running result |
|---|---|---|---|
| 1 | L | £5 | -£5 |
| 2 | L | £10 | -£15 |
| 3 | W | £20 | +£5 |
| 4 | W | £5 | +£10 |
| 5 | W | £5 | +£15 |
| 6 | L | £5 | +£10 |
| 7 | W | £10 | +£20 |
| 8 | L | £5 | +£15 |
| 9 | L | £10 | +£5 |
| 10 | W | £20 | +£25 |
Total wagered £95, result +£25. Five wins and five losses turned into a £25 profit, and that is exactly why Martingale is so persuasive: on most short sequences it wins.
What it does to variance: it converts a lot of small losing runs into small wins and concentrates all the losing into the rare long run. The longest losing run in the test was two. Here is what a longer one costs from the same £5 base:
| Consecutive losses | Stake on next hand | Total lost so far |
|---|---|---|
| 1 | £10 | £5 |
| 2 | £20 | £15 |
| 3 | £40 | £35 |
| 4 | £80 | £75 |
| 5 | £160 | £155 |
| 6 | £320 | £315 |
| 7 | £640 | £635 |
| 8 | (£1,280) | £1,275 |
Where it breaks: eight straight losses from £5 means £1,275 gone, and the ninth bet would need £1,280 to win back a £5 profit. Losing eight hands in a row is not rare over an evening of blackjack. The table limit breaks it sooner: a £500 maximum stops the progression at the seventh step, because the eighth bet of £640 cannot be placed, and you are £635 down with no way to recover it. Doubles and splits make it worse, because a doubled £320 hand puts £640 at risk even when the limit allows the base bet.
2-1-2 (Manhattan)
The rule: start at 2 units. After a win, drop to 1 unit; after a second consecutive win, go to 3 units, then 4, then 5 for further wins. After any loss, return to 2 units. There is a full worked guide to the 2-1-2 system with longer sequences and bankroll requirements.
| Hand | Result | Stake | Running result |
|---|---|---|---|
| 1 | L | £10 | -£10 |
| 2 | L | £10 | -£20 |
| 3 | W | £10 | -£10 |
| 4 | W | £5 | -£5 |
| 5 | W | £15 | +£10 |
| 6 | L | £20 | -£10 |
| 7 | W | £10 | £0 |
| 8 | L | £5 | -£5 |
| 9 | L | £10 | -£15 |
| 10 | W | £10 | -£5 |
Total wagered £105, result -£5. The same five wins and five losses that made Martingale £25 cost the 2-1-2 player £5, because the biggest bet of the sequence (£20 on hand 6) landed on a loss.
What it does to variance: it is a mild positive progression with a built-in step back after the first win, so it locks in a little of a winning run before pressing. Swings are larger than flat betting but nowhere near Martingale.
Where it breaks: the starting stake is 2 units, so the base cost of a losing run is double that of flat betting. On a run of losses you bleed 2 units a hand with no mechanism to recover them, and the one-unit step after a win means a W-L-W-L pattern loses money steadily.
1-3-2-6
The rule: bet 1 unit, and on each consecutive win move to 3, then 2, then 6 units. After a loss at any point, or after completing all four wins, go back to 1 unit.
| Hand | Result | Stake | Running result |
|---|---|---|---|
| 1 | L | £5 | -£5 |
| 2 | L | £5 | -£10 |
| 3 | W | £5 | -£5 |
| 4 | W | £15 | +£10 |
| 5 | W | £10 | +£20 |
| 6 | L | £30 | -£10 |
| 7 | W | £5 | -£5 |
| 8 | L | £15 | -£20 |
| 9 | L | £5 | -£25 |
| 10 | W | £5 | -£20 |
Total wagered £100, result -£20. The sequence needs four wins in a row to pay off. Three in a row got the player to +£20 and the 6-unit bet on hand 6 gave it all back and more.
What it does to variance: losses during a run are small (1 unit each) and the pay-off from a completed cycle is large (12 units on 4 bets). Results are a string of small losses punctuated by occasional big wins. Four consecutive wins at a 42% win rate per hand come up roughly one cycle in every 30, and that is before pushes stretch things out.
Where it breaks: the system is designed so that the big bet is the one you are most likely to lose, because every bet in blackjack is more likely to lose than win. The 6-unit stake on the fourth hand carries the same 0.5% house edge as the 1-unit stake on the first.
Paroli
The rule: start at 1 unit and double after each win, for up to three wins in a row. After the third win, or after any loss, return to 1 unit.
| Hand | Result | Stake | Running result |
|---|---|---|---|
| 1 | L | £5 | -£5 |
| 2 | L | £5 | -£10 |
| 3 | W | £5 | -£5 |
| 4 | W | £10 | +£5 |
| 5 | W | £20 | +£25 |
| 6 | L | £5 | +£20 |
| 7 | W | £5 | +£25 |
| 8 | L | £10 | +£15 |
| 9 | L | £5 | +£10 |
| 10 | W | £5 | +£15 |
Total wagered £75, result +£15. The three-win run on hands 3 to 5 completed a cycle for +£35, and the rest of the sequence chipped at it.
What it does to variance: you are betting the house's money during a run, so the downside on any hand is limited to 1 unit of your own stake plus whatever the run had earned. Wins are lumpy and losses are a slow drip.
Where it breaks: three wins in a row happen about 7% of the time from any starting hand (0.42 x 0.42 x 0.42), so most cycles end with a loss of 1 unit or with a 2-unit bet losing back the first win. Over a long session the drip of 1-unit losses adds up to the same expected loss as flat betting at the same average stake.
Oscar's Grind
The rule: the aim is to win exactly 1 unit per cycle. After a loss, keep the same stake. After a win, raise the stake by 1 unit, unless that would take the cycle past a 1-unit profit, in which case bet only what is needed to finish the cycle at +1 unit. When the cycle shows +1 unit, start again.
| Hand | Result | Stake | Cycle position | Running result |
|---|---|---|---|---|
| 1 | L | £5 | -£5 | -£5 |
| 2 | L | £5 | -£10 | -£10 |
| 3 | W | £5 | -£5 | -£5 |
| 4 | W | £10 | +£5 (cycle complete) | +£5 |
| 5 | W | £5 | +£5 (cycle complete) | +£10 |
| 6 | L | £5 | -£5 | +£5 |
| 7 | W | £5 | £0 | +£10 |
| 8 | L | £5 | -£5 | +£5 |
| 9 | L | £5 | -£10 | £0 |
| 10 | W | £5 | -£5 | +£5 |
Total wagered £55, result +£5. On hand 7 the stake stayed at £5 rather than rising to £10, because a £10 win would have taken the cycle to +£10, past the 1-unit target.
What it does to variance: it is the gentlest progression on the page. The stake creeps up only during a recovery and is capped by the cycle target, so the total wagered (£55) is barely above flat betting.
Where it breaks: a long losing run leaves you in a cycle that is deep in the red with a stake that has not risen, and then each subsequent win raises the bet into a hand you are more likely to lose than win. Cycles can take dozens of hands to close, and bankroll requirements during a bad cycle are higher than the modest stakes suggest.
D'Alembert
The rule: raise the stake by 1 unit after a loss and lower it by 1 unit after a win, never going below the base unit.
| Hand | Result | Stake | Running result |
|---|---|---|---|
| 1 | L | £5 | -£5 |
| 2 | L | £10 | -£15 |
| 3 | W | £15 | £0 |
| 4 | W | £10 | +£10 |
| 5 | W | £5 | +£15 |
| 6 | L | £5 | +£10 |
| 7 | W | £10 | +£20 |
| 8 | L | £5 | +£15 |
| 9 | L | £10 | +£5 |
| 10 | W | £15 | +£20 |
Total wagered £90, result +£20. Wins tend to land on larger stakes than losses when wins and losses alternate, which is where the profit on this sequence comes from.
What it does to variance: slower than Martingale, since stakes grow by addition rather than doubling. Ten straight losses from £5 put you at a £55 next bet and £275 down, which is survivable; the same run on Martingale would have asked for £5,120.
Where it breaks: it assumes wins and losses will roughly balance, and in blackjack they do not. You lose about 49% of hands and win about 42%, so the stake drifts upward over time and the average bet, and with it the expected loss, climbs with the length of the session.
Comparison table
The same ten hands, side by side. "Chance of a small session win" and "risk of a large loss" are qualitative, because the exact numbers depend on session length and limits, but the ordering is not in dispute.
| System | Result on test sequence | Total wagered | Expected loss at 0.5% | Risk of a large loss | Chance of a small session win | Bankroll needed (units) | Verdict |
|---|---|---|---|---|---|---|---|
| Flat | £0 | £50 | 25p | Low | Moderate | 50 to 100 | Lowest expected loss for the hands played; the default |
| Martingale | +£25 | £95 | 48p | Very high | High | 127 for seven steps, 255 for eight | Wins often, loses catastrophically; blocked by table limits |
| 2-1-2 | -£5 | £105 | 53p | Moderate | Moderate | 100 to 200 | Harmless structure, but the 2-unit base doubles the cost of a losing run |
| 1-3-2-6 | -£20 | £100 | 50p | Moderate | Low | 100 to 200 | Needs four straight wins; the big bet is the one most likely to lose |
| Paroli | +£15 | £75 | 38p | Low | Low | 50 to 100 | Limits risk to your own base unit; wins are rare and lumpy |
| Oscar's Grind | +£5 | £55 | 28p | Moderate | High | 100 to 200 | Closest to flat; long cycles can trap a bankroll |
| D'Alembert | +£20 | £90 | 45p | Moderate to high | Moderate | 100 to 200 | Stakes drift up because losses outnumber wins |
Notice the "expected loss" column. On identical cards, the systems vary between 25p and 53p of expected loss for ten hands, and the only reason is how much money each put on the table. The results column (from -£20 to +£25) is noise around those small numbers, and a different ten-hand sequence would shuffle it completely.
Table limits and why they matter
Every table has a minimum and a maximum stake, and the maximum is what kills negative progressions. Take Martingale from £5 at a table with a £500 limit:
| Step | Stake | Within £500 limit? | Cumulative loss if it loses |
|---|---|---|---|
| 1 | £5 | Yes | £5 |
| 2 | £10 | Yes | £15 |
| 3 | £20 | Yes | £35 |
| 4 | £40 | Yes | £75 |
| 5 | £80 | Yes | £155 |
| 6 | £160 | Yes | £315 |
| 7 | £320 | Yes | £635 |
| 8 | £640 | No | cannot bet |
The seventh step is the last one the table allows. If it loses, you are £635 down and the system has no next move. The casino is not setting a £500 limit to stop you winning; a £5 to £500 spread is a perfectly normal range. The limit simply makes it impossible to double indefinitely, and it is the rare long run, not the typical short one, that carries the whole expected loss of the system.
Online tables are often tighter at the low end (£1 to £100, or £5 to £250), which brings the ceiling down to five or six steps. Live dealer VIP tables run to thousands, but the base unit scales with them and so does the loss at the wall.
The ratio matters more than the raw numbers. A 100 to 1 spread between minimum and maximum allows six doubles (5, 10, 20, 40, 80, 160, 320 needs 64 to 1 for seven bets). A 20 to 1 spread allows four.
The only betting strategy with an edge
There is one betting strategy in blackjack that changes expected value rather than variance, and it only works because something else has changed first: the composition of the remaining shoe.
A card counter using Hi-Lo tracks the ratio of high cards to low cards left in the shoe. When the true count rises, the player's edge rises by roughly 0.5% per point, turning positive at around +1 to +2 depending on the rules. The counter then bets more, typically on a 1 to 8 unit spread in a six-deck game: 1 unit at a true count of +1 or below, rising to 8 units at +5 or above. The bet spread is not a progression; it does not respond to the last result at all. It responds to the deck. The card counting page sets out the count, the index plays and the bankroll, and the advanced strategy page covers how it fits with everything else.
This requires a shoe game with penetration of 75% or better, a table that tolerates the spread, and accurate counting. RNG games reshuffle every hand and most live dealer games cut at around 50%, so for online play it is rarely available, which leaves flat betting as the best blackjack betting strategy for almost everyone reading this.
Setting a session budget
Since expected loss is edge x total wagered, the budget question is really a question about how much you will wager, which depends on stake, speed and session length. The table shows an hour's expected loss playing basic strategy at about 0.5% and playing casually at around 2%. Live and online games often deal 100 hands an hour to a single player; a full land-based table runs closer to 60.
| Stake | Hands per hour | Wagered per hour | Expected loss at 0.5% | Expected loss at 2% |
|---|---|---|---|---|
| £5 | 60 | £300 | £1.50 | £6 |
| £5 | 100 | £500 | £2.50 | £10 |
| £10 | 60 | £600 | £3 | £12 |
| £10 | 100 | £1,000 | £5 | £20 |
| £25 | 60 | £1,500 | £7.50 | £30 |
| £25 | 100 | £2,500 | £12.50 | £50 |
A progression raises the "wagered per hour" figure above the flat-bet number, so a Martingale player at a £5 base is not a £5 player for budget purposes. On the test sequence the Martingale wagered almost twice as much as flat betting; over an evening with a few long losing runs it can be several times more.
Three numbers to fix before the first hand: a loss limit (a session bankroll of 50 to 100 units is the usual guide), a win target at which you stop, and a time limit. Then stop at whichever arrives first. The actual result in any hour will usually be far from the expected figure in the table, in both directions, and that is why the limit has to be set in advance rather than decided in the moment. The safer gambling page lists the deposit and session tools that enforce this online.
Keep stake sizing separate from hand selection. Use a rules-matched blackjack strategy chart or check a conventional multi-deck hand with our blackjack strategy calculator. A staking progression does not change which card decision is best.
Frequently asked questions
What is the best blackjack betting strategy?
Flat betting at a stake of 1% to 2% of your session bankroll. It gives the lowest total wagered for the hands you play, and so the lowest expected loss, and it keeps the swings within what a 50 to 100 unit bankroll can absorb. Every other system has the same expected loss per pound wagered and most of them wager more.
Does Martingale work in blackjack?
It wins small amounts often and loses large amounts rarely, with the same long-run expectation as flat betting at the same average stake. From a £5 base, eight consecutive losses cost £1,275, and a £500 table limit stops the progression at the seventh step with £635 lost. Doubles and splits make the exposure larger still.
What is the 2-1-2 blackjack betting system?
Start at 2 units, drop to 1 after a win, then rise to 3, 4 and 5 units on further consecutive wins, and return to 2 units after any loss. It is a mild positive progression that banks a little of each winning run. The 2-1-2 worked guide runs longer sequences and shows the bankroll it needs.
Should I raise my bet after a win or after a loss?
Neither changes your expected result. Raising after a loss (Martingale, D'Alembert) produces frequent small wins and rare large losses. Raising after a win (Paroli, 1-3-2-6) produces frequent small losses and occasional larger wins. If you want the smallest expected loss and the most predictable session, do not raise at all.
What is the optimal bet size in blackjack?
For a basic strategy player, the smallest stake that makes the game worth your time, within a session bankroll of 50 to 100 units. For a card counter, bet size follows the true count on a spread such as 1 to 8 units, with the top bet set by bankroll and risk of ruin rather than by feel.
Is there a blackjack betting strategy chart?
Not in the sense of the basic strategy chart, because there is no betting decision that depends on your hand or the dealer's up-card. The nearest thing is a counter's bet ramp, which maps true count to units wagered, and that is covered on the card counting page. For everyone else the "chart" is one line: same stake, every hand.
Back to Blackjack strategy: basic strategy explained, with the full chart