August 5, 2026 · 8 min read

The Perya Color Game: Why Betting on Every Colour Cannot Win

The color game has 216 possible rolls, so its maths can be counted exactly rather than guessed. Here is the real house edge, why the most popular strategy — covering every colour — has no winning outcome at all, and the one number on the screen that actually changes the game.

Everyone knows this one. Three dice under a cup, six colours on the board, and a crowd leaning in. It has been at every fiesta and every perya for as long as anyone can remember, and now it is on the phone too — Color Game Live, ColorPlus, perya tables inside half the casino apps in the country.

It is also the rare gambling game where nobody has to take anyone's word for anything. Three dice have exactly 216 possible outcomes. That is small enough to count all of them, so every number in this article is counted rather than estimated. You could check it yourself on paper.

And the most popular way people play it — spreading bets across several colours, or all six — turns out to have a property almost nobody knows about.

First, the game as it actually pays

The standard perya ladder is simple: you pick a colour, and if it lands on one die you get paid 1:1, on two dice 2:1, on all three 3:1. Count all 216 rolls and here is the whole game:

• Your colour shows on no dice — 125 rolls out of 216, or 57.87%. You lose the bet.
• On one die — 75 rolls, 34.72%. Pays 1:1.
• On two dice — 15 rolls, 6.94%. Pays 2:1.
• On all three — 1 roll in 216, 0.46%. Pays 3:1.

Work the arithmetic through and the house edge comes out at exactly 17/216, which is 7.87%. That is the same figure the guides quote — and it is worth pausing on, because it is unusual. Most published gambling numbers do not survive being checked. A slot machine's advertised RTP is often the best setting the machine can be shipped at, not the one you are sitting at. The color game's stated edge is simply correct.

Now the part nobody tells you: covering every colour has no winning outcome

Watch a perya table for ten minutes and you will see it — chips going down on three, four, five colours at once. It feels like the sensible play. Cover more of the board, and surely something comes back every roll.

Something does come back every roll. That is exactly the problem.

Put one unit on each of the six colours, so you have staked six. Whatever the dice do, each colour that appears pays you its stake back plus one unit for each die it landed on — and three dice means three units are paid out in total. So your return is always the number of different colours showing, plus three. Against a stake of six, that gives only three possible results:

• All three dice different — three colours showing, so you get back 6. Net zero. This happens on 120 of 216 rolls, or 55.56%.
• Two dice the same — two colours showing, you get back 5. You lose 1 unit. 90 rolls, 41.67%.
• All three the same — one colour showing, you get back 4. You lose 2 units. 6 rolls, 2.78%.

Read that list again. There is no outcome where you finish ahead. The best roll in the game hands you back exactly what you put in. Every other roll takes money. Covering the board does not reduce your risk — it removes the possibility of winning, and leaves the 7.87% edge completely unchanged.

And if that trade sounds familiar, it should. It is the same one a slot machine makes when it lights up on twenty paylines and pays you back less than you staked — the mechanic our Myth Lab covers under losses disguised as wins. Coverage buys the sensation and spends the substance. The colour board is just the honest, countable version of it.

The one number on the screen that actually changes the maths

If you only check one thing before playing an online color game, check what a triple pays. Everything else about the game can stay identical while that single number moves the house edge a long way:

• Triple pays 3:1 — the classic perya ladder — house edge 7.87%.
• Triple pays 5:1 — house edge 6.94%.
• Triple pays 10:1 — house edge 4.63%.
• Triple pays 12:1 — house edge 3.70%.

Some app guides claim their color game pays 10× on a triple while also quoting the standard 7.87% edge. Those two statements cannot both describe the same game, and 7.87% is the figure that belongs to a 3:1 triple. When a page contradicts itself like that, the paytable inside the game is the only thing worth trusting.

Speed is the part nobody prices

The edge is only half of what a session costs. The other half is how many times you pay it. Cost per hour is your stake × the edge × the number of rounds — and the rounds are where the online version quietly differs from the perya.

A live-dealer dice table runs roughly 40–50 rounds an hour. A software version runs 200 or more. At ₱10 a round on a 7.87% colour bet, that is about ₱35 an hour against a live dealer and ₱157 an hour against software — same bet, same odds, four times the cost. At ₱100 a round it is ₱1,570 an hour.

This is why bet-picking matters less than it feels like it should. On a sic bo table, the best bet available played at software speed costs about as much per hour as one of the worst bets played slowly with a live dealer. Choosing well and playing fast cancels the choosing.

So what actually helps

Nothing changes the odds. There is no colour that is due, no pattern in the last ten rolls, and no arrangement of chips that beats 7.87% — the dice have no memory and the board has no strategy. What is left is the boring thing, and it is the only thing with a real effect:

• Decide the number before you sit down. A loss limit and a time limit, set while you are calm — not in the fifteen seconds between rolls.
• Check the paytable, especially the triple. It is the one number that genuinely varies between games wearing the same name.
• Notice the speed. An app dealing 200 rounds an hour is charging you the edge four times as often as the perya table did.
• Play money that has no other job. If it is rent, bills, tuition or borrowed, it is not a bankroll.

If you want the number worked out for your own income and bills rather than a rule of thumb, the safe-to-play calculator does it in about a minute, and 1% of monthly income is the figure the international research actually lands on.