Our method

The simulator on this site is not decoration and not a random number generator put there for show. Below: where the data comes from, what the 151.8 million real spins behind our long-run figures are, how the model is calibrated and where the limits of its accuracy lie.

Where to play

Where the numbers come from

We work with three kinds of source and never mix them.

First, the game itself. Mechanics, RTP and symbol payouts are read off the rules and paytable screens inside the game; release dates come from Pragmatic Play's own pages. This is also where our casino comparison comes from: in September and October 2026 we opened the rules page of nine Dog House games inside fifteen real player accounts — 132 readings in all — and recorded which build each operator had loaded. Those figures are the ones with a screenshot beside them.

Second, aggregator catalogues. This is how we know a game exists in several certified builds at all. A catalogue entry is a version's passport, not the result of anyone's play, and the two must not be confused.

Third, aggregated statistics of real rounds collected by clashofslots across many casinos. From these come bonus frequency, median and mean round payouts, and the shape of the distribution. We use them to describe how a game behaves — never to judge a casino, because the sample pools operators running different builds, and an average across them says nothing about any one of them. If a figure came from a press release rather than these three sources, we say so in the text.

What the 151.8 million spins are

For The Big Dog House, Megaways 1000 and Dice Show there is no such data yet: the first two were released recently and the third runs on separate markets. For those games we quote the spec sheet only, and we say so plainly rather than papering over the gap with somebody else's estimates.

How the simulator is built

At its core is a model of the per-spin payout distribution, split into tiers: dead spins, small returns, mid-sized wins, bonus hits and the tail of big rounds. The parameters of each tier are fitted so that three quantities match: the return, the hit frequency and the spread of payouts. The model runs in the browser, with no server, and is not connected to any casino; you can try it on The Dog House page and on the game pages.

How we check accuracy

We verify the model's return analytically — by computing the expectation from the distribution parameters, not by trusting a single run. A million-spin run can drift a percentage point and a half purely because of rare large payouts, and an estimate like that is worth little.

Across The Dog House series the analytical return sits within 0.3 of a percentage point of the official figure for seven games out of eight. Where a game's mechanics could not be confirmed from its paytable, it does not go on the site at all.

What the simulator does not do

It shows one thing: what the shape of the results distribution looks like over the long run if you play a given stake for a given number of spins. The virtual money in it cannot be withdrawn and has nothing to do with a real balance.

What we do when a number is wrong

If a figure turns out to be wrong, we correct the page and update the modified date. Games in the series whose mechanics we could not confirm from a paytable do not make it onto the site: two disputed titles in the series are unpublished for exactly that reason, while Dice Show went the other way, as we checked it in the demo and added it once we were satisfied the game is real.

Common questions

Where does the figure of 151.8 million spins come from?

We did not run them ourselves: this is external material that bonus frequency, median round payouts and actual return are checked against. For the 2026 releases there is no such data yet.

Is the simulator connected to real play?

No. It is an independent mathematical model that runs in your browser: it is not connected to any casino, does not know your real balance and has no effect on the outcome of real spins.

Why can't RTP be checked with a single run?

Because rare big payouts dominate the result, a single random run gives a noisy estimate: even over a million spins it can drift by as much as 1.5 percentage points. So we work out the model's return analytically, as the expected value from the parameters of the distribution, and use random runs to picture the spread rather than to verify accuracy.

Where do the symbol payouts come from?

From the paytable inside the game itself, read in demo mode. We treat values on third-party sites as secondary and re-check them: some of them publish payouts per line, which puts their numbers out by a factor of twenty against ours. We convert everything into multiples of the player's total bet.

Who writes the material?

Owen Marsh, the editor of the site. The material is informational: the site takes no bets, is not an operator and does not offer real-money play. If an error turns up in a published figure, we correct the page and update the modified date, and data we could not confirm against a source does not make it onto the site.