TableWise How casino games actually work

The gambler’s fallacy and other casino myths

The gambler’s fallacy is the belief that a chance event gets more likely because it hasn’t happened lately: black is due after a run of red. It’s false for any game where each result is independent, and most casino myths are versions of it. Here are the common ones, each with the sum that settles it. None of this is a way to win. No system changes the house edge.

What is the gambler’s fallacy?

It’s treating the past as a debt that the next spin has to pay back. On a one-zero wheel, red comes up 18 times in 37. Five reds in a row is a short run, 2.72% of the time, and the fallacy says black must be more likely now. It isn’t. A wheel keeps no record, so after any run the chance of black on the next spin is 48.65%, what it always was.

Myth

“Red has come up five times, so black is due.”

What the maths says

Black on the next spin is still 18/37. A run of five reds is 1 in 37 from a standing start, but the five reds have already happened: only the next spin is in question.

The gambler’s fallacy has a real meaning, then, and a tempting logic. People know that over many spins red and black come out roughly even, and they feel that a short run has to correct itself. It doesn’t have to. Big numbers wash a run out rather than paying it back, and the section “It all evens out” shows how.

Gambler’s fallacy examples

  • At the wheel: “Black’s due.” Each spin is 18/37 either way.
  • At the dice: “There hasn’t been a 7 in ages.” A 7 is 6 ways in 36 on every roll of craps.
  • At a slot: “This one hasn’t paid all day.” Each press takes a fresh draw from the generator on slot machines.
  • At baccarat: “Banker’s won four, so Player’s turn.” Past results shift baccarat odds only as far as the cards they used, and a scorecard doesn’t record those.
  • Lotteries: “That number hasn’t come up in months.” A draw has no memory of the last one.
  • After a loss: “I’m due a win.” Your results don’t enter the sum at all.

Is the gambler’s fallacy real?

The fallacy is real in the sense that people really do think this way; the “due” result isn’t. The difference between the gambler’s fallacy and probability is the difference between two questions. How likely is a run before it starts? How likely is the next result, given the run so far?

Reds on a one-zero wheel: rare as a whole, unremarkable somewhere in a session, and no help on the next spin.
QuestionAnswerWorking
Five reds in five spins2.72%(18/37)⁵
Ten reds in ten spins0.074%(18/37)¹⁰
Five reds in a row somewhere in 100 spins77.04%Exact count over all 100-spin runs
Ten reds in a row somewhere in 1,000 spins31.60%Exact count over all 1,000-spin runs
Black on the next spin, after any of these48.65%18/37, always

So long runs aren’t a sign of anything. They’re what random results look like over time, and 77% of 100-spin sessions have a run of five reds somewhere.

One honest exception: card games dealt from a shoe. Cards that have gone change what’s left, very slightly, so results aren’t quite independent there. That’s about the cards, not about who won. It doesn’t bring anything “due”, and casinos deal with it by shuffling. The blackjack card explains what’s known.

Its cousins: the inverse fallacy and sunk cost

The inverse gambler’s fallacy runs the other way. Someone walks up, sees a rare result such as five reds, and concludes the wheel must have been spinning a long time. A rare result says nothing about how many spins came before it: five reds are as likely on the next five spins as on any other five.

Sunk cost and the gambler’s fallacy are often found together, but they’re different mistakes. The gambler’s fallacy is an error about the next result: “black is due.” Sunk cost is an error about money already gone: “I’ve lost too much to stop now.” What you’ve lost is lost whichever way you decide, so it shouldn’t enter the decision. Together they’re how a short session turns into a long one, as our responsible gambling page says.

The Martingale strategy

The Martingale betting strategy says: bet on red, double after each loss, and go back to your first bet after a win. Sooner or later you must win, the argument goes, and the win pays for all the losses plus one unit. Each of those claims is true for a player with endless money and no table limit. No such player exists.

Take $10 on red, on a one-zero wheel, with enough for 7 bets: $10, $20, $40, up to $640. Most series end in a $10 win. One in about 106 loses all 7 bets, and costs $1,270.00.

Worked example

One Martingale series, 7 bets deep

  1. Chance of losing all 7: (19/37)7 = 0.94%.

  2. Otherwise, up $10: 99.06% of series.

  3. Expected result:

    99.06% × $10 − 0.94% × $1,270.00 = −$2.05
  4. Expected money bet in that series: $75.94. And −$2.05 is exactly 1/37 of it.

The next bet after seven losses would be $1,280.00, past most players’ budgets and possibly over the table’s limit, which is posted at each table.

So the Martingale doesn’t change the price. It moves the losses into rare, large lumps, with lots of small wins between them, and that feels like a working system until the lump arrives:

Chance of at least one full 7-bet loss, by the number of series played. Each series won is $10.
SeriesWon if none loseAt least one full loss
10$100.009.0%
50$500.0037.7%
100$1,000.0061.2%
200$2,000.0084.9%
500$5,000.0099.1%

Any betting system

The same goes for every staking plan, however it’s named: Fibonacci, d’Alembert, Labouchère or something new. Each bet carries its own price, made before the spin that settles it, and expected values add up. So over any series of bets, your expected loss is the house edge times what you staked, whatever pattern set the stakes.

Six plans on red, one-zero wheel. Each row is computed by walking every win-and-loss path. In every row the cost is exactly 1/37 of the stake.
PlanExpected money betExpected loss
Doubling after a loss, up to 7 bets$75.94$2.05
Doubling after a loss, up to 4 bets$41.65$1.13
Doubling after a loss, up to 10 bets$113.08$3.06
Flat $10, 12 spins$120.00$3.24
Up $10 after a loss, down $10 after a win, 12 spins$290.15$7.84
Double after a win, back to $10 after a loss, 12 spins$413.84$11.18

A plan can change the shape of a session: many small wins and a rare big loss, or the reverse. It can’t change what the session costs per dollar. In a game where your decisions do count, such as blackjack or video poker, the decisions change the cost. The size of your bets doesn’t. Five systems are played out over sessions of 1,000 spins, with a budget and a table limit, in roulette systems, tested.

Hot and cold numbers

Over a short stretch, random results look lumpy, and we read patterns into them. Spin a one-zero wheel 37 times, once per pocket on average. You might expect each number to come up about once. In fact, about 13.4 of the 37 numbers won’t come up at all, and about 2.9 will come up three times or more. Those are your “cold” and “hot” numbers, from a fair wheel. On the next spin, each still has the same chance. To watch it happen, spin the wheel 1,000 times in our simulator.

A machine that’s due

A slot’s result is a draw at the moment you press. On the teaching machine on our slots page, the top prize comes on average once in 65,536 spins. That’s not a schedule: 36.8% of the time, a machine goes that many spins without one, and the chance on the next spin doesn’t move. Ontario’s technical standards require the generator to be independent of play history and bar a game from using words such as “due” or “ready”, and a machine can’t change how it works based on what it has held. “Near misses” aren’t close calls either: a loss with a 7 next to the line is the same loss as any other.

Skill where there isn’t any

A game where nothing is decided after the bet has no skill to learn. Ways of throwing dice, pressing a button or timing a spin don’t change results, and Ontario’s standards bar a game from suggesting that they do. Where decisions exist, as in blackjack and video poker, correct play lowers the cost and stops there. A strategy chart isn’t a route to profit, and anyone selling a sure method is selling something else.

“It all evens out”

Results do even out, but only as a share. Over more spins the percentage of reds settles near 48.65%, while the difference in the count of reds from its average actually grows. And the money doesn’t even out at all, because the edge never stops working:

$10 on red per spin, one-zero wheel. ± is one standard deviation, about how far a typical result strays.
SpinsExpected redsCount strays byShare strays byExpected loss
10049±5±5.00%$27.03
10,0004,865±50±0.50%$2,702.70
1,000,000486,486±500±0.05%$270,270.27

So the share of reds gets closer and closer to its average, without anything “paying back” the early results. A player betting $10 on red still finishes 100 spins ahead 35.53% of the time, and 1,000 spins ahead 18.76%. Keep playing and that keeps shrinking.

When a myth becomes a reason to keep playing

“I’m due”, “I’ll win it back” and “one more will turn it round” are also signs that gambling is costing more than you meant. If you hear yourself saying them, set a limit, or stop for the day. Our responsible gambling page has what’s available in Ontario, and ConnexOntario answers at any hour on 1-866-531-2600.

Where to go next

19+ Play responsibly ConnexOntario 1-866-531-2600