Myths and misconceptions, taken apart
Seven ideas that follow casino games around — due wins, hot streaks, doubling systems, near misses — checked against the actual arithmetic, with the numbers from our own games as worked examples.
Seven claims, and what the maths says
Claim
"It hasn't paid in ages, so it's due."
A game with independent rounds has no memory. Two dice that have missed a total of seven twenty times in a row are exactly as likely to roll seven on the next throw as they were on the first: six chances in thirty-six, or 16.67%. The belief that a run of misses makes a hit more likely is the gambler's fallacy, and it is the single most expensive idea in gambling.
What is true is that long-run frequencies settle down. That happens because more results dilute an early imbalance, not because the game corrects it.
Claim
"Hot and cold streaks are real, so ride them."
Streaks are real; they are just not predictive. Flip a fair coin two hundred times and a run of six or seven heads is likelier than not — randomness clusters, and human eyes are excellent at finding shape in the clusters. The next flip is still an even chance.
Reef Diver makes the point neatly. Clear ten safe tiles with three urchins and the multiplier reaches 5.00×, which reflects the risk you have already survived. It says nothing about the eleventh tile, where the chance of coral is simply the number of safe tiles left divided by the tiles left.
Claim
"Doubling after a loss guarantees a win in the end."
The Martingale system asks you to double your stake after each loss so that one win recovers everything plus one unit. The arithmetic works. The requirement does not: stakes grow exponentially, and both your balance and any table limit are finite.
| Losses in a row | Next stake | Already staked |
|---|---|---|
| 3 | 800 | 700 |
| 5 | 3,200 | 3,100 |
| 8 | 25,600 | 25,500 |
| 10 | 102,400 | 102,300 |
On a bet you win half the time, eight losses in a row turn up once every 256 runs — rare, and certain to arrive if you keep going. When it does, it takes back every unit the system had collected. Averaging over every round, the expected result of a Martingale is identical to the expected result of flat stakes on the same bet: the system changes the shape of the swings, not the odds.
Claim
"That was so close — I nearly had it."
Two matching symbols on a scratch card with the third one panel away feels like progress. It is not. Lantern Scratch draws the card's result first, from the printed table, and only then lays out the nine symbols: the winning symbol three times, everything else at most twice. A card that shows two Gold Cranes was never one panel from paying, because the result was already settled.
Nothing here is weighted to produce near misses, and no animation on this site is tuned to make a loss look like an almost-win.
Claim
"Free casino games are good practice for the real thing."
They are good practice at reading a rule sheet, and that is where it ends. Two differences do the damage. First, the payouts: several bets here return the full mathematical value of the risk, while commercial games keep a permanent margin on every bet. Second, the money: no decision made with coins that cannot be bought carries the weight of one made with rent.
Success at social casino games does not imply future success at real-money gambling. A winning streak on this site is evidence of nothing beyond a pleasant half hour.
Claim
"Bigger stakes give you better odds."
Stake size scales what a round pays and what it costs. It does not touch the chances. Backing under seven on Scarab Dice wins 41.67% of the time at 100 coins and 41.67% of the time at 2,500. The only thing a larger stake reliably increases is the size of the swings.
The same goes for playing faster: more rounds per hour means more of everything, including the effect of any margin. That matters far more with money than with free coins, which is exactly why it is worth knowing.
Claim
"Somebody can predict what comes next."
Not from past results, when each round is drawn independently. This site takes its numbers from crypto.getRandomValues, the browser's cryptographic random source, with a fresh draw for every throw, card, snap point and urchin layout. Past rounds are not part of the input.
Doubloon Climb shows the shape of it. The snap point is drawn before the climb starts and never adjusted, so a rope that has snapped low five times running is neither owed nor denied a long climb on the sixth.
How randomness works on this site
Every payout table on this site prints the chance next to the payout, so the maths is checkable.
Each round starts by asking the browser for random bytes and turning them into the numbers the game needs: two dice faces, one outcome against a prize table, a snap point, or a shuffle of urchin positions. Nothing about your balance, your recent results or how long you have been playing takes part in that draw.
The one deliberate margin is small and stated. Doubloon Climb and Reef Diver both keep one per cent — the snap-point formula gives a 0.99 ÷ m chance of reaching any multiplier m, and Reef Diver's multiplier is 0.99 divided by the chance of the run you have survived. Lantern Scratch returns about 88.5 coins per 100 staked over a very long run, which the prize table adds up to on its own. Scarab Dice pays true odds on several bets, and its under and over bets break even because a seven returns your stake.
Read that list once and the general point lands: in every game of chance, the numbers behind it are knowable, and they never include a reason to expect a win because you are owed one.