无信任协议用于回测系统交易算法
简而言之：回测交易算法缺乏透明且可重复的标准，结果无法可靠验证。一个无需信任的开放协议能否有所帮助？
回测是系统化交易算法的基础，但目前仍没有开放且可验证的标准来记录、结构化、重现或审计回测结果。每个人似乎都在使用自己的JSON/CSV格式。通常你可以阅读其他人的回测输出，但无法可靠地验证其准确性。
我在考虑一个无需信任的协议：一个定义回测应该如何记录、哈希、文档化和重现的规范。这不是一个产品或平台，而是任何人都可以实现的开放协议。
关键理念可能包括：
- 固定的、开放的输入和输出模式
- 加密一致性检查
- 完全可重现所需的元数据
- 确定性执行指南
- 完全开源的参考工具
- 完整的审计能力，零信任假设
去中心化的点对点实现可以确保回测数据保持公开可验证，同时避免中央控制。该协议需要保持中立和非商业化，以维护其完整性。
我只是一个初学者在探索这个想法，因此这更像是一个思考而非提案。有没有人知道类似的东西是否已经存在？
查看原文
TL;DR: Backtesting trading algorithms lacks a transparent, reproducible standard. Results can’t be reliably verified. Could a no-trust, open protocol help?<p>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.<p>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.<p>Key ideas could include:<p>fixed, open schemas for inputs and outputs<p>cryptographic consistency checks<p>required metadata for full reproducibility<p>deterministic execution guidelines<p>fully open-source reference tools<p>complete auditability, zero-trust assumptions<p>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.<p>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?