Parlay: the system that lets the winnings ride.

What is the Parlay system in roulette?

Parlay (also known as Paroli or Reverse Martingale) is a positive progressive strategy. The player raises the bet after a win rather than after a loss. In its basic form it is played on an even-money chance paying 1:1, such as red or black. Win, and the whole payout rides on. Lose, and the run ends and starts again from the base unit. The system does not alter the probability of any individual spin — on European roulette red still has odds of 18 in 37, roughly 48.65%.

Why does Parlay feel so safe?

Parlay behaves in exactly the opposite way to Martingale. Martingale tries to chase a loss back with aggressive raises after every defeat, so the risk piles up at the worst possible moments. Parlay instead accepts small losses and waits for several wins in a row. Psychologically that is far more pleasant: after a run of losses you never risk ever-larger sums, and each new run puts only the base unit on the table.

That pleasantness is precisely the trap. Because the individual losses are small, they are easy to overlook. But they arrive again and again — most runs break before they reach the target. If successful runs do not show up often enough, the bankroll declines quietly and steadily. So Parlay does not create the occasional extreme pressure Martingale does; it creates a constant small drain.

The rules of Parlay and the run target

The classic variant uses four rounds. You start with one unit. After the first win you bet 2 units, after the second 4 units, after the third 8 units. Win the fourth spin too and the whole run ends, and you take a net profit of 15 units. Lose at any step and you do not double after the loss — the run simply ends and you start again from 1 unit.

A target fixed in advance is essential. Without one you end up never taking the riding winnings off, and sooner or later a single loss wipes them all out. A target of four wins is a compromise: long enough for a worthwhile profit, short enough to come off now and then. Thanks to the reset after a loss, Parlay does not demand an enormous bankroll for chasing losses — but it does need winning streaks often enough.

What exactly does the house edge mean for Parlay?

At first glance Parlay sounds favourable: you risk 1 unit and four wins in a row bring you 15. If we ignored the zero and counted only red and black, the probability of four wins in a row would be exactly 1/16. One successful run would bring +15 units, roughly 15 unsuccessful runs would take −15, and over the long run the result would be about zero — a fair trade.

But roulette does have a zero, and for a colour bet it is not a win. The chance of one win is therefore not 1/2 but 18/37 on European and 18/38 on American roulette. That small shift alone is enough to tip the theoretical balance into a loss. So over the long run Parlay tracks the house edge — 2.70% on European and 5.26% on American roulette — exactly like flat betting.

What did the long-run Parlay tests show?

The long test ran Parlay over 25,000 spins. The outcome matched the statistical expectation: the system drifted below zero and finished roughly 1,500 units down. No single-evening catastrophe as with Martingale, but a persistent downward trend that tracks the house edge precisely.

Variants with different run lengths were tested too — lowering the target to two wins, and raising it to eight or nine in a row. The long-term trend did not change materially. A shorter target means more frequent but smaller wins; a longer one rarer but larger. The expected value stays negative either way; only the shape of the curve and the volatility change. You can switch between the same variants in the simulator above.

How long a run target should you choose?

The choice of target is purely a trade-off between frequency and size of win. A two-win target closes in roughly 23.7% of runs on European roulette and earns 3 units. A four-win target comes off in about 5.6% of runs for 15 units. An eight-win target paying 255 units succeeds only about once in 320 runs — that is closer to a lottery than a system.

A practical rule: the longer the target, the longer the stretches in which the bankroll just drifts down a unit at a time waiting for a miracle. If you are testing Parlay, pick the target by how long a dry spell you can bear financially and psychologically — not by the size of the payout at the end.

Why not to add a progression after a failed run

A tempting adjustment: raise the base bet by one unit after every failed run until the system is back in profit. It looks like insurance that “catches up” on the lost runs. The test showed the exact opposite — the adjustment did not help and instead pushed the total loss to very high figures.

The reason is simple: a more aggressive base only magnifies the effect of bad stretches. Once you start raising after failed runs, Parlay stops being a positive progression and starts behaving like a more dangerous loss-chasing system — the very mechanism it originally set out to avoid. You can switch this variant on in the simulator and watch how the curve changes.

How to read the bankroll chart

A Parlay chart is the mirror image of a Martingale one. Instead of long calm stretches with an occasional plunge, you see a gradual decline in small steps — those are the broken runs — and the occasional jump up when a run closes. Watch above all whether the jumps up keep pace with the small losses between them. Over a long sample they usually do not.

That is why the simulator has a toggle between the bankroll chart and the bet-size chart. The second view shows that the large bets appear only inside winning streaks — they grow out of the winnings riding on the table, not out of your bankroll. That is the main structural difference from systems that raise after a loss.

European, French and American roulette

Parlay is played on even-money bets: red/black, odd/even or low/high. European roulette has 37 numbers including one zero, American 38 because of the double zero. For Parlay the American wheel is considerably worse — the chance of one win drops from 18/37 to 18/38, and over a run that difference compounds: the chance of 4 wins in a row falls from 5.60% to 5.03%.

French roulette with La Partage or En Prison softens the impact of the zero on even-money bets, making it the friendliest of the variants. But even the best rules do not change the conclusion: the expected value of a run stays negative, just less so.

How to use Parlay sensibly in the tester

The best way to understand Parlay is to run several simulations with the same settings and compare them. A short close-up run shows the rhythm of the system — small losses, the occasional closed run. A long run over 25,000 spins shows the trend a short sample can hide. One time you finish ahead thanks to a lucky cluster of runs, the next the curve simply falls.

Watch four figures above all: the final bankroll, the deepest drawdown, the number of runs closed against the theoretical odds, and the longest winning streak. If the system finished ahead only thanks to one exceptionally long run, that tells you something important about how fragile the result is. To compare it with Martingale, Fibonacci or your own strategy on identical spins, use the Roulette system tester.

The most common beginner mistake

With Parlay a beginner usually gives in to the feeling of “playing with the casino’s money”. Once a run is under way they are betting money they won, so a loss does not feel like a loss. But the winnings riding on the table are already your money — the moment you leave them there, you are risking them exactly as you risked your stake. In accounting terms it makes no difference whether the $800 leaves your wallet or the table.

The second common mistake is judging the system on a short sample. A few hundred spins can happen to catch a cluster of successful runs and Parlay looks profitable. That does not mean the system has gained an edge over roulette — only that the long-term trend has not shown up in the sample. Compare several runs, lengthen the simulation, and watch the results converge on the negative expected value.