Engine cross-validation
Pass probabilities compared between the PropSim engine and the open-source LuxAlgo engine (MIT license), on rules from our verified YAMLs and a frozen reference profile. No business contact: only the public npm package.
No published results yet. The draft exists (frozen protocol, specs generated from rules.json) and awaits human validation — gaps will be published with their causes, never smoothed.
Drafts: seo-agent/reports/cross-validation-<date>.md — regenerated at each engine version.
Why gaps exist
- Two different RNGs — same declared seed, unpaired draws.
- Intra-trade evaluation order of the daily barriers.
- Neighbouring trading-day models but distinct implementations (Poisson).
- FundedNext consistency rule simulated by LuxAlgo, not by PropSim (expected gap on FundedNext).
- The5ers daily anchor approximated differently (max balance/equity vs balance).
Two Monte Carlo engines under neighbouring assumptions give estimates of the same order: the gap is not a “true” reference probability.
Open dataset — prop firm rules · Glossary
Educational tool. Probabilities come from simulations under simplified assumptions; they do not guarantee any outcome. Past performance is not indicative of future results. Trading involves a substantial risk of loss. This site is not affiliated with or endorsed by the prop firms mentioned — partner links, marked as such, are the only exception.