Home / Tools / Dice roller

Virtual dice roller

Fair server-side dice: pick the count and the die, every submit is a fresh independent roll

Choose your dice and roll. Each throw is generated server-side with a cryptographic random source — no roll depends on the one before.

Your roll

Choose how many dice and how many sides. Every submit rolls fresh dice on the server.

Number of dice 1 to 10 dice per roll.

Die type From the d4 up to the d20.

Reset

No JavaScript required. No JavaScript required. The dice are rolled server-side with a cryptographic random source.

How the math works

Fair dice, honestly rolled

Each die is an independent uniform draw: every face has exactly the same chance on every throw, and no streak changes the next roll. The expected total is the number of dice times the average face, (sides + 1) / 2 — for two six-sided dice, 7.

That independence is the same property certified RNGs bring to casino games — and the same reason no betting pattern can outguess the next outcome, here or at a table.

# independent uniform rolls

roll = random_int(1, sides) × dice

expected_total = dice × (sides + 1) ÷ 2

Methodology

Each die is an independent draw from a uniform distribution via PHP's random_int (CSPRNG). The expected total is dice × (sides + 1) / 2 — the same independence property certified casino RNGs are audited for.

Reading a run of dice

Six faces give a 1 in 6 chance each, and that is the whole model — but the shape of a short run surprises people. Over 60 rolls the expected count per face is 10, while anything from 4 to 17 is unremarkable. Over 6,000 rolls the expected count is 1,000 and the spread narrows to a few per cent. Watching that gap close is the clearest demonstration of why a casino needs volume and a player does not have it.

Two sums are worth doing with the roller open. The chance of no six in ten rolls is (5/6)^10, about 16% — common enough to happen in any session. The chance of three sixes in a row is 1 in 216, which is roughly once per 200 rolls and therefore not evidence of anything. Streaks in a fair sequence are expected, which is exactly why they are useless as a signal.

FAQ

Are these dice results actually random?

They come from the platform's cryptographically secure generator, the same class of source a licensed game uses. Nothing here is weighted, and no result depends on previous rolls.

Why do runs of the same number appear so often?

Because independent trials cluster. Over 600 rolls a streak of four or five identical faces is expected, not remarkable — which is exactly why streak-based systems fail.

Is a die 'due' to land on a number it has missed?

No. Each roll has no memory, so the probability stays 1 in 6 regardless of history. The gambler's fallacy is the most expensive idea in casino play.

What is this tool good for?

Seeing how slowly observed frequencies converge on the theoretical ones. It is a demonstration of variance, not a predictor of a real game's next result.