Tag
文章探讨了人工智能领域巨额投资形成的自我强化"资金循环"，正在像2008年一样大量吸引外国资本流入,引发对AI泡沫与金融风险的担忧。
作者用 AI 代理回测了一个策略：每月买入上月表现最好的股票，自 1998 年以来 1000 美元可增长到 880 万美元，远超标普 500 的 220 万美元（但收益主要来自苹果、英伟达等少数离群股）。整个实验是用 Agents API 构建的。
OpenStocks 用无利益冲突的 AI 智能体预测公司财报数据，与华尔街分析师共识对比，自 8 月 4 日开始评分以来 AI 共识在 90 份财报中赢下 60 份，典型误差 2.5% 低于华尔街的 2.9%。
Coinbase CEO Brian Armstrong praises the SEC's proposed merit-based private market access in the US, arguing it would let retail investors participate in private markets rather than leaving opportunities only to the wealthy.
swyx 提醒 AI Engineer New York 2026 的门票本周末即将售罄，这是纽约规模最大的技术大会，且首次设立金融主题主会场。Leadership 门票已售罄，Oct 12-14 的 Engineering 门票仍可购买。
A data center company's ambitious $30 billion IPO plans reportedly collapsed within 48 hours amid shifting market conditions and investor skepticism.
Shane Parrish 用 bot 创建了一个「Knowledge Project 播客嘉宾指数」，统计上过节目的创始人和 CEO 所在公司自首次登台以来的股价表现，结果显示该指数大幅跑赢标普 500。
WIRED published an interactive web feature, 'Spend Elon Musk's Money,' letting users virtually allocate the tech billionaire's fortune across purchases and donations — a lighthearted viral simulation.
本文提出 FinVector-Market-4B，使用 rank-16 LoRA 在 22,000 条语料上对 Qwen3.5-4B 进行结构化金融任务的适配研究，表明在匹配提示契约下，紧凑型领域适配能带来超出输出格式学习的显著任务收益。
作者讨论了 AI agent 的 ROI 衡量难题：节省的时间常常被简单等同于裁员，导致财务团队不信任。提出了一套更可信的计量方法，包括只统计外部系统确认的工作、计入人工审核时间、扣除后续出错的部分，并对每个数字标注是实测还是估算。
A builder shares a concept for a deterministic, node-based graph system that LLMs can both understand and construct for complex time-series financial calculations, arguing that black-box LLM math output is untrustworthy for finance agents. The project is still a proof of concept, with longer-form deep dives planned for community feedback.
OpenAI customer story on how Jump Trading uses GPT-6 Astra agents to scale quantitative research, from coding to multi-day hypothesis-validation workflows, while keeping human review at the center in a regulated environment.
An article reporting that the only billionaires increasing their wealth this year are those in the technology sector, highlighting tech's outsized role in wealth generation amid broader market conditions.
HKU's lab open-sourced Vibe-Trading, an MIT-licensed financial research platform with 34K+ GitHub stars that lets users run full quant research chains—from data sourcing and factor mining to multi-agent backtests—entirely via natural language.
Wall Street firms are increasingly using AI to handle the research, modeling, and presentation work traditionally done by junior bankers, potentially hollowing out the entry-level rungs of the finance career ladder that train future senior bankers.
A historical overview tracing how financial information delivery evolved from telegraph tickers and pigeon dispatches to the Bloomberg Terminal, founded by Michael Bloomberg in 1981 after leaving Salomon Brothers.
Bill Ackman warns that the current abundance of venture capital echoes the dot-com bubble, urging founders to raise now but spend judiciously and keep cash in the bank to survive an eventual high-profile blow-up and market reset.
Robinhood has rolled out AI-powered trading agents to all its 29 million customers, allowing them to use OpenAI and Anthropic models to execute trades and build investment strategies. This move aims to democratize finance by providing tools previously reserved for institutional investors.
An experiment where the Claude AI agent researches news daily and trades a $1,000 paper portfolio against the S&P 500 over a year, with Day 1 showing a slight gain compared to SPY's loss.
A Brookings paper projects $10.3 trillion in AI infrastructure investment from 2025 to 2032, highlighting financing risks and potential inefficiencies in demand.