现有股票研究输出要么是没人看的密集 spreadsheet 导出,要么是营销味满满的终端截图。这个项目是第三种选择 — 单命令、单文件、阅读体验对标一线卖方研究 note 的 HTML 报告。Most equity research output is either dense spreadsheet exports that no one reads, or marketing-fluff terminal screenshots. This project is the third option — single-command, single-file HTML reports that read like a top-tier sell-side research note.
本站每份报告都由三阶段流水线生成:Every report on this site was produced by a three-stage pipeline:
1. 数据层。1. Data layer. 并行拉取 20 个 Financial Modeling Prep API 端点 — 公司画像、5 年年报、8 季度趋势、估值倍数、分析师预期、同业、18 个月价格历史。20 Financial Modeling Prep API endpoints pulled in parallel — profile, 5-year annuals, 8-quarter trends, valuation multiples, analyst estimates, peers, 18 months of price history.
2. 分析层。2. Analysis layer. 自动计算 4 种 DCF 并对比(季度年度化、历史 FCF、前瞻 EPS、前瞻 FCF 调整)。周期股从盈利趋势判定。情绪分数从价格动量得出。4 DCF methods auto-computed and compared (Q-annualized, Historical FCF, Forward EPS, Forward FCF-adjusted). Cyclicality detected from income trend. Sentiment score from price action.
3. 渲染层。3. Render layer. 自含 HTML,内嵌 SVG 图表,中英双语切换,45+ 个内联术语提示,加完整术语词典章节。Self-contained HTML with inline SVG charts, bilingual EN/中 toggle, 45+ inline term tooltips, and a full glossary section.
视觉语言刻意编辑化:暖米色背景、Charter 衬线标题、单一 Claude 珊瑚强调色、买卖信号用柔和的状态色。目的是让研究阅读起来像深度思考,而非激进推销。The visual language is deliberately editorial: warm cream background, Charter serif headlines, a single Claude coral accent color, and muted state colors for buy/sell signals. The goal is to make research feel considered, not aggressive.
每份报告都是单一自含 HTML 文件。没有外部 CSS,没有 JS 图表库,没有服务端依赖。任何浏览器都能打开,打印效果也漂亮。Each report is a single self-contained HTML file. No external CSS, no JS chart libraries, no server-side dependencies. Opens in any browser, prints beautifully.
本项目不构成投资建议。This is not investment advice. 报告由程序自动生成,只是众多决策输入之一。任何评级都请先读完空头案例再行动。Reports are programmatically generated and should be one input among many. Read the bear case before acting on any verdict.
DCF 永远是估算。DCF is always an estimate. 季度年度化在周期顶部会高估。情绪缓冲可能掩盖基本面风险。要用多个视角交叉验证。Q-annualized can overstate at cycle peaks. Sentiment buffer can mask fundamental risk. Use multiple frames.
模型局限。Model limitations. 系统针对传统上市公司调优。稳定币发行方(Circle)、银行、REIT、资产管理公司的商业模式不适用标准 DCF — 这些报告应当作研究起点,而非最终答案。The system was tuned for traditional public companies. Stablecoin issuers (Circle), banks, REITs, and asset managers have business models that don't fit standard DCF — treat those reports as a starting point for further work, not a final answer.
数据来源。Data source. 所有财务数据来自 Financial Modeling Prep(Premium 套餐)。季度数据是 Q-annualized DCF 必需的。每跑新报告本站随之更新。All financials come from Financial Modeling Prep (Premium tier). Quarterly data is necessary for Q-annualized DCF. The site updates as new reports are run.