Duel Dice: the Simplest Zero Edge Game on the Site

The payout on a zero edge dice game is a hundred divided by your win chance, and you can check every line of the paytable against that in your head. Roll under 49.50 and it owes you 2.0202x, because 100 divided by 49.5 is 2.0202. Take a 2 per cent chance and it owes you 50x. Take 98 per cent and it owes you 1.0204x. Anywhere a house keeps an edge, those numbers come out lower than the division says, and that gap is the entire business model of every other dice game on the internet. Here the division holds at every setting, which is the most boring possible way to prove something interesting.

Verifying a roll takes about a minute and is worth doing once, so you know what you actually have. Before you bet, the server seed already exists and you are shown its hash. You supply the client seed or accept the one handed to you, and every bet moves a nonce along by one. Those three things combine and hash into the number the game rolls. While you play, the server seed stays hidden, so all you hold is a hash and a promise. Rotate your client seed and the old server seed is released. Hash it yourself, compare it against the string you were shown before the first bet, and if it matches, the whole sequence of rolls was locked before you placed any of them.

Be precise about what that does and does not prove, because plenty of sites are sloppy here. It proves results were not altered mid session and not chosen in response to your bets. It says nothing about whether the paytable is generous and nothing about whether you will win. A miserly paytable published honestly and verified cryptographically is still a miserly paytable. On this game the arithmetic and the verification both hold, which is rarer than it should be.

Two habits survive from dice games that did keep an edge, and both are dead weight here. Switching between roll over and roll under changes nothing whatsoever; they are the same bet described from opposite ends, and the payout follows the win chance either way. Nudging the target after a losing run changes nothing either, because the next roll has no memory of the last one and the seed for it was fixed before you touched the slider.

What does deserve thought is the bankroll a low chance setting quietly demands. A 2 per cent target pays 50x and lands, on average, once in fifty rolls. Average is doing heavy lifting in that sentence. Going a hundred and fifty rolls without a hit at that setting is an ordinary occurrence, not bad luck, and you need enough left in the balance to still be there when it arrives. Most people who complain that a zero edge game took their money were funding a 50x dream with a 2x bankroll.

Move the slider, read the payout, that is the entire game

You set a win chance anywhere from 1 to 99 per cent. The game hands you a payout multiplier of exactly 100 divided by that number. Pick 50 per cent and you get 2.00x. Pick 25 and you get 4.00x. Pick 2 and you get 50.00x.

Multiply the chance by the payout and you land on 100 every single time. That is what zero house edge means, written where anyone can see it. Every other casino quietly makes that product come to 97 or 96, and the difference is their revenue. Even Beef, the one original here that keeps a cut, only takes 0.8 per cent.

Duel Dice provably fair: what the panel actually lets you check

Before the roll you get a hashed server seed. You supply your own client seed. After the roll the server seed is revealed, and you recompute the result from the three values. If the number matches, nothing was touched.

Publishing seeds is not a Duel invention, plenty of operators do it. What is unusual here is that the maths underneath has nothing to hide in the first place, so verifying it is a formality rather than an accusation.

Where Dice sits next to the rest

If Dice is the cleanest expression of the model, Crash is the loudest and Mines is the one that makes you decide when to stop. All three pay the same over a long run. What differs is how violently they get there, and how many chances they give you to talk yourself into one more click.