Our Fair Draw Assurance

Every Swishes competition is drawn by the Swishes Fair Draw Engine, the software that picks the winning ticket number. It has been examined and tested by an independent, accredited laboratory, and this page records what that involved.

Who tested it

The Fair Draw Engine was evaluated by GLI Europe B.V., an independent testing laboratory in Haarlem, the Netherlands. GLI is accredited by the Dutch Accreditation Council (RvA) under registration L372 for testing and C577 for product certification.

The evaluation ran from 19 June 2026 to 21 August 2026, and the result was a pass. It is recorded in report RN-246-SW5-26-01, dated 28 August 2026, covering Swishes Fair Draw Engine RNG version 1.1.

What was examined

• The source code. GLI was given the full source code behind the draw and traced it from the moment a draw starts to the numbers it produces, looking for any undisclosed switch or parameter that could influence the outcome or affect fair play.

• The build itself. The exact file tested was fingerprinted, so the certified version can always be identified.

• The numbers it produces. GLI analysed both the final draw outcomes and the raw output of the algorithm before any scaling was applied to it.

The testing covered draws selecting as many as 1,000 winners from a pool as large as 10,000,000 tickets, drawn without replacement, meaning no ticket can be selected twice within the same draw.

The tests applied

The draw outcomes were put through fourteen separate statistical tests: Runs, Serial Correlation, Interplay Correlation, Adjacency Max-Min, Adjacency High-Low, Adjacency Blocks, Coupon Collector's, Duplicates, Overlaps, Permutation, Total Distribution, Total Distribution by Position, and Count of Counts. The raw binary output was tested separately against the Diehard battery, a standard suite designed to find patterns in the individual bits an algorithm produces.

Each test compares what the draw engine actually did against what genuine randomness would be expected to do. A result that is too irregular fails, and so does one that is too neat, because real randomness is not tidy.

The Fair Draw Engine passed every test at the 95%, 98% and 99% confidence levels.

The version that was tested

These are the fingerprints of the exact file GLI examined, taken with Verify+ by Kobetron.

Questions

If you have a question about how a competition was drawn, email [email protected].