An AI Stock Picker Alternative
Built for Research Agents 面向研究 Agent 的
AI Stock Picker 替代方案

A fixed stock score may not match a fund, fintech app, analyst team, or portfolio workflow. Developers need configurable research steps.

固定股票分数不一定适合基金、金融科技应用、分析师团队或投资组合工作流。开发者需要可配置的研究步骤。

Watchlists, earnings calendars, market movers, filings, and news alerts need scheduled and repeatable agent workflows.

自选股、财报日历、市场异动、文件和新闻预警，都需要定时且可重复的 Agent 工作流。

A useful agent should know which data source, schema, cost, latency, and provider note shaped a result.

有用的 Agent 应该知道哪个数据源、Schema、成本、延迟和供应商说明影响了结果。

Search for stock prices, company profiles, financial statements, earnings transcripts, SEC filings, market movers, or financial news.

搜索股票价格、公司资料、财务报表、财报电话会、SEC 文件、市场异动或金融新闻。

Inspect parameters, return fields, provider notes, latency, and cost so the agent can call tools safely and predictably.

检查参数、返回字段、供应商说明、延迟和成本，让 Agent 更安全、更可预测地调用工具。

Send structured outputs into a stock research agent, market monitor, portfolio alert system, or analyst-ready report.

把结构化结果送入股票研究 Agent、市场监控、组合预警系统或分析师可用报告。

Combine market data, fundamentals, news, earnings, and filings into a repeatable company research process.

把市场数据、基本面、新闻、财报和文件组合成可重复的公司研究流程。

Track watchlists, movers, earnings events, and news spikes, then create automated alerts with context.

追踪自选股、异动、财报事件和新闻峰值，并生成带背景的自动预警。

Build a transparent ranking process that explains which capabilities and data sources shaped each result.

构建透明的排名流程，说明每个结果由哪些能力和数据源影响。

Generate daily or event-driven summaries for analysts, product users, or internal investment workflows.

为分析师、产品用户或内部投资流程生成每日或事件驱动的摘要。

Specify exchanges, countries, security types, sectors, liquidity, market capitalization, trading history, and data-availability rules. Apply point-in-time membership so delisted or newly listed companies do not distort historical tests.

明确交易所、国家、证券类型、行业、流动性、市值、交易历史和数据可用性规则，并使用历史时点成分，避免退市或新上市公司扭曲历史测试。

Document how value, quality, growth, momentum, revisions, risk, and event signals are calculated. Show source dates, missing fields, normalization, sector treatment, and whether a score comes from observed data or model interpretation.

记录价值、质量、成长、动量、预期修正、风险和事件信号的计算方式，展示来源日期、缺失字段、规范化方法、行业处理，以及评分来自观测数据还是模型解释。

Add diversification, liquidity, turnover, exposure, event, and position-size constraints. A high score does not automatically justify a portfolio position when the candidate duplicates existing risk or cannot be traded efficiently.

加入分散化、流动性、换手率、风险暴露、事件和仓位约束。高分不等于自动适合进入组合，尤其当候选标的重复现有风险或无法高效交易时。

Preserve each ranking snapshot, input version, score contribution, exclusions, and subsequent outcomes. Review false positives, regime changes, restatements, and execution assumptions instead of evaluating only the best-performing ideas.

保留每次排名快照、输入版本、分数贡献、排除原因和后续结果，复盘误报、市场环境变化、数据重述与执行假设，而不是只看表现最好的想法。