金融给AI代理的严苛信号：PnL，Questflow四层金融栈解析
Finance gives AI agents an unusually unforgiving outcome signal: PnL.
Finance gives AI agents an unusually unforgiving outcome signal: PnL.
金融为AI代理提供了一个异常严苛的结果信号：盈亏（PnL）。
And the harder part of an AI trading agent starts after the model has an opinion.
而AI交易代理更困难的部分，始于模型形成观点之后。
@questflow calls its 4-layer stack the Financial Harness:
@questflow 将其四层架构称为“金融驾驭层”：
models for reasoning, Skills for investment methods, plugins for live financial data and tools, and execution for read/write connections to exchanges, brokers, wallets, and accounts.
模型负责推理，技能用于投资方法，插件接入实时金融数据和工具，执行层则建立与交易所、经纪商、钱包和账户的读写连接。
Questflow also let me choose among OpenAI GPT, Claude, Gemini, Grok, Kimi, DeepSeek, Qwen, MiniMax, Mimo, GLM, and Hunyuan for different research, analysis, and trading tasks.
Questflow 还让我在 OpenAI GPT、Claude、Gemini、Grok、Kimi、DeepSeek、Qwen、MiniMax、Mimo、GLM 和 Hunyuan 之间选择，以应对不同的研究、分析和交易任务。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力