Backtesting is the foundation of systematic trading algorithms — yet there’s still no open, verifiable standard for how backtests should be recorded, structured, reproduced, or audited. Everyone seems to be using their own JSON/CSV formats. You can usually read another person’s backtest output, but you can’t reliably verify it.
I’m thinking about a no-trust protocol: a specification defining how backtests should be logged, hashed, documented, and reproduced. It’s not a product or a platform, just an open protocol anyone can implement.
Key ideas could include:
fixed, open schemas for inputs and outputs
cryptographic consistency checks
required metadata for full reproducibility
deterministic execution guidelines
fully open-source reference tools
complete auditability, zero-trust assumptions
A decentralized, peer-to-peer implementation could ensure backtest data remains publicly verifiable while avoiding central control. The protocol would need to remain neutral and non-commercial to preserve its integrity.
I’m just a beginner exploring this idea, so this is more a thought than a proposal. Does anyone know if something like this already exists?