Novus Odds
 
Getting startedBeginner· 6 min read· 4 steps

How Novus Odds simulations work

The end-to-end mental model: a seeded RNG, a synthetic price, an underlying true probability, and thousands of resolved events.

ShareXFacebookLinkedInRedditPinterestEmail
Open the experiments directory
  1. 1

    Everything is synthetic

    There are no live odds, players, or casino tables. Each lab builds a fictional market from parameters you set, so every number is reproducible and safe to dissect.

  2. 2

    A seed makes it reproducible

    The random seed drives a deterministic generator. Reuse the same seed and settings and you get the exact same run — that is what makes the experiments repeatable research rather than one-off gambling.

    Tip: Change only the seed to see a different 'history' of the same model; change a setting to see a different model.

  3. 3

    Two probabilities, not one

    Every event has a priced implied probability (what the American price breaks even at) and a simulated true probability (the hidden chance the lab uses to settle outcomes). The gap between them is the whole lesson.

  4. 4

    Resolve, then aggregate

    The engine draws outcomes for thousands of events, then reports win rates, ROI, confidence intervals, and the discrepancy detector. Large samples converge; small samples stay noisy.

Key takeaways

  • Seeds make every result reproducible.
  • Priced vs true probability is the core comparison.
  • Read aggregates, not single events.
Related breakdowns guide →All tutorials →
Responsible use: these tutorials teach probability mathematics with synthetic data. They do not identify real-world profitable selections. Read the responsible-use page.