Nganya. Our own provably fair crash game

What the game actually is

You are a passenger on a nganya. When the round starts it pulls away and the multiplier climbs with its speed. At some point it breaks down, and every stake still on board is lost. Cash out before the breakdown and you keep whatever the multiplier read when you did.

The controls use the words the matatu does: panda to get on with a stake, shuka to get off with your winnings. The makanga hanging off the door is the announcer, and he never falls: the breakdowns are a rotating set of comic mechanical failures rather than crashes, because nobody needs a gambling product that shows people getting hurt.

Mechanically it is the same genre as Aviator, and everything on our Aviator page about how crash games work applies here. The differences are in what you are able to check.

How the fairness check works

The problem with every crash game is simple to state: the operator decides where the round stops, and you cannot see that decision. You are asked to trust that the number was not adjusted once it knew what you staked. Provable fairness replaces that trust with arithmetic.

1. Before betting opens, the server publishes a hash of the round's secret seed. A hash is a one-way fingerprint: it proves the seed already exists without revealing what it is. The crash point is fixed at this moment, before anyone has staked anything.
2. You bet. Nothing the server does now can change the outcome, because the commitment is already public and any change would break it.
3. After the round, the server publishes the seed. Your browser hashes it and checks the result against the commitment you were shown, then recomputes the crash point from it and checks that against what you just watched.

The seeds are also chained, each one derived from the next, so the whole history is verifiable as a single sequence rather than a set of unrelated claims. Removing or rewriting one round breaks every link after it.

The crash point comes from the seed like this: take HMAC-SHA256 of the round seed with your client seed, read the first 52 bits as a number X between 0 and 1, and compute 0.97 / (1 - X), truncated to two decimals with a floor of 1.00. That formula is what produces the property in the next section, and it is published rather than described so you can check our arithmetic as well as our honesty.

Why the RTP is the same at every target

That derivation gives the probability of a round reaching any multiplier m as 0.97 / m. Multiply the chance of reaching a target by what it pays and the target cancels out: cash out at 1.5x and the expected return is 97% of your stake; at 50x it is also 97%.

This is worth sitting with, because it kills an entire industry. If every target returns the same, then no cash-out strategy is better than another, not a fixed 1.2x, not martingale, not waiting for a run of low multipliers. There is no configuration of the game where the edge is smaller, so there is no “system” to discover and nothing anyone can honestly sell you. Our page on why no Aviator strategy works makes the same argument about everyone else's game.

What the edge does mean is that the game is not a way to make money. 3% of everything staked is the cost of playing, it compounds over a long run, and no amount of verifiability changes it.

Past rounds do not affect future rounds. Every crash-game round is generated independently. A run of low multipliers does not make a high one “due”, and no pattern in this history has any predictive value whatsoever. Anything claiming otherwise is selling you something.

The instant-crash number, stated properly

Roughly 3% of rounds crash at the floor with no multiplier accrued. You will observe closer to 4% of rounds showing 1.00x, because any result between 1.00 and 1.01 truncates down to 1.00 on the display. Both figures are correct and they answer different questions. One is how often the game truly ends immediately, the other is how often it looks like it did.

We publish both because quoting only the lower one would be flattering and quoting only the higher one would be alarmist, and because a difference of one percentage point in a number a regulator reads is the kind of thing worth getting right in public.

What we do not claim

Nganya is not a better bet than any other crash game. The maths is the maths, and a verifiable 3% edge takes your money at the same rate as an unverifiable one. Anyone telling you a particular crash game pays better is either reading noise as a pattern or wants a referral.

What is genuinely different is the checking. On most crash games the fairness claim is a sentence in a help page. Here it is a computation you can run, and if we ever broke it the evidence would be in the hands of every player who saved a commitment. That is a weaker promise than “trust us” sounds, and a much stronger one than it is.

Our conflict of interest

We review crash games and we run one. Naming that is the minimum, so: Nganya never appears in our rankings or comparison tables, it is labelled as ours everywhere it appears, and the unflattering arithmetic on our crash-game pages is applied to it unchanged. You can read the full version on our about page, including how the site is funded.

Treat this page as interested writing, because it is. The claim on it that does not depend on trusting us is the verification, and that is deliberate.

Before you play

Set a loss limit before you start rather than during, because the point at which you most want to raise it is the point at which it is doing its job. The limits and self-exclusion controls are inside the game itself, and we honour them without asking you to justify anything.

If betting has stopped being fun, or if you recognise yourself in the pattern of chasing a loss, please start with responsible gambling instead. That page is not a formality. It lists help that is free and reachable from Kenya.