Peer-reviewed papers on AI, MEV, protocol design, and on-chain agents — all built on real XRPL data.
Our research process is designed for reproducibility, rigor, and direct applicability to production systems on the XRP Ledger.
We index full XRPL history since genesis, covering all transaction types, ledger state deltas, and validator telemetry. All datasets are made publicly available alongside papers.
Models are designed around XRPL-specific constraints: 3-5 second consensus rounds, multi-currency paths, and the unique topology of the XRPL validator UNL graph.
We evaluate on held-out real-world data with walk-forward validation. No data leakage, no cherry-picked benchmarks. Baselines include the strongest publicly available alternatives.
Research that proves out in evaluation gets ported into our open-source SDKs. Every model endpoint in our tools has a corresponding paper with training details and benchmarks.