Provably-Fair Engine — Evaluation Kit
Sylox Tech · a result engine for licensed gaming operators that players can
verify and exploiters can't predict.
▶ Live prediction tool: https://deepankar1993.github.io/Test-Prediction-engine-show/
— learns a game's result history, posts its 8 most-likely numbers for the next round to
check against the live game (or 3 reel digits for deeper structure), and shows that the
same engine gets no edge against the commit-reveal engine. Your test is saved in the
browser across days; use export/import to carry it between machines.
This kit lets you evaluate it yourself in a few minutes. It has three pieces and
two ways to open them — pick whichever fits you.
Quickest start (one click)
- Windows — double-click
Start-Windows.bat
- Mac / Linux — double-click
Start-Mac-Linux.command
That opens the interactive demo in your browser and shows a small menu for the
rest. (Requires Python 3.8+, which most machines already have. If a technical
review isn't needed, you can also just open the two files below directly.)
One command (for engineers)
python sylox_kit.py # opens the demo + menu
python sylox_kit.py --demo # runs the predictable-vs-fair audit, no browser
python sylox_kit.py --audit path/to/results.csv
python sylox_kit.py --check # confirms the kit is intact
What's inside
1 · fairness_demo.html — the interactive demo
Double-click to open in any browser; works offline. Press Run comparison to
watch the same betting attack run against an incumbent-style engine and the Sylox
engine. Drag Incumbent predictability and watch the exploit rise and fall;
the Sylox side never moves. The live preview computes a real commit/reveal hash
on your device — the crypto is genuine, not a mockup.
2 · sylox_brief.pdf — the one-page summary
For the decision-maker: the two numbers that matter, what "provably fair" means
in plain language, and why it matters for compliance, trust, and margin.
3 · predictability_audit.py — the audit tool
Point it at any results file and it reports whether the next result is tied to
the previous ones — the test a regulator's auditor runs. No installs needed.
python predictability_audit.py results.csv
python predictability_audit.py results.db --table rounds --column result
python predictability_audit.py --make-samples # writes test fixtures
A clean engine passes all five tests; a predictable one is flagged.
samples/ — ready-made data to prove the tool before using real data:
fair_sample (passes) and predictable_sample (flagged), each as CSV and
SQLite, plus observed_history.csv.
Suggested 3-minute walkthrough
- Open the demo, press Run comparison, then drag the predictability slider.
- In a terminal:
python predictability_audit.py samples/predictable_sample.csv
→ five flags. Then samples/fair_sample.csv → five passes. Same tool, only
the engine differs.
- Read
sylox_brief.pdf.
Note: the audit measures predictability from logged results — it is an
evaluation tool, not a live-betting client. Point it at any published history,
including ours.
Deepankar Singh · Founder, Sylox Tech (OPC) Pvt. Ltd.
[ your email ] · [ sylox.tech ]
