Game certification

Under the hood Certification made simple

Certification looks like a certificate at the end. This is the middle. A build running on the bench, the findings it throws, the test cases behind each one, and a few hundred million simulated spins settling onto the declared return.

Every test case is mapped across the markets it applies to before testing starts. Where requirements overlap, the case runs once and counts for every market it covers, so twenty markets do not mean twenty passes through the same game. In the test execution panel, every line that says counts for a number of markets is one run doing the work of several.

Build under test

cherry-trail v1.4.2
Spin
0
Stake
1.00
Last win
0.00
Session
0.00

Simulation for illustration. No stake, no payout, nothing playable.

Test execution

0 / 0

Findings raised

0 NC, 0 obs

    Maths analysis

    1 of 3

    $ rtp-sim --gameId cherry-trail --bet 1.00 --rounds 100000000

    Declared RTP
    96.20%
    Measured
    —
    95% interval
    —
    Max win
    —
    Confidence Collecting

    Interval narrows with the square root of the spin count.

    Three things run at once: the front end on the bench, the maths behind it, and a line by line read of the source. A non-conformity in any one of them stops the certificate.

    Evidence is captured autonomously as each case runs, and tied to the test case and clause that needed it. The certificate is then generated entirely by machine from that evidence. Nobody retypes results into a report, so there is no transcription step for errors to creep into and no queue waiting for someone to write it up. Our people spend their time on the judgement calls instead.

    Every finding is posted to the client's Slack channel the moment it is raised, so a studio can start on the fix while the rest of the run carries on, rather than finding out from a report at the end.

    Simulation. Sequences shortened. Three hundred million simulated rounds are compressed into a few seconds, and the build, studio, findings and results shown are illustrative rather than taken from a real certification. Test case IDs, titles and requirement references may be a mix of real and combined for illustrative purposes, drawn from the RiskCherry test library.

    That is one game

    Now do it across twenty six markets, each with its own standards, and keep it to a turnaround a studio can plan a release around. That is the job.