Skip to main content
WeiMindWeiMind
ChineseView Course
Quant Radar

Research Directions

Large Language Model Applications Latest Activity

Using large language models and agents for research-report reading, data processing, coding, and research automation.

50 related activitiesLast 30 days 36 itemsFeatured among them 1 items

Updated

Activity Archive · Page 2 页

15 OctThuIssue 21–40 items
  1. Hong Kong Monetary Authority: Press Releases35

    HKMA launches "ART" framework to promote responsible AI adoption in the banking sector

    The HKMA launched the "ART" framework on October 15, centered on the core principles of Accountability, Resilience, and Trust, to support the banking sector in responsibly adopting artificial intelligence. The framework includes the release of the latest regulatory guidelines on responsible AI adoption, the joint establishment of an AI training data sharing platform, and the advancement of agentic AI application testing. Concurrently, the HKMA and Cyberport co-hosted a two-day "Finance + AI Forum," marking the successful completion of the second phase of the Generative AI Sandbox technical testing.

    Read original
14 OctWed
  2. Moontower63

    Key Takeaways from the Interview with LTCM Co-Founder Victor Haghani: Position Sizing Matters More Than Instrument Selection

    Moontower host Kris Abdelmessih breaks down an interview with LTCM co-founder Victor Haghani, arguing that investment decisions consist of two parts: instrument selection and position sizing, and that LTCM's failure stemmed from flawed position sizing rather than poor instrument selection.

    Read original
  3. Anthropic News20

    Anthropic Opens Singapore Office and Signs Memorandum of Understanding with the Singapore Government

    Anthropic opened an office in Singapore, its fifth in the Asia-Pacific region, and signed a memorandum of understanding with Singapore's Ministry of Digital Development and Information. The two sides will cooperate in areas such as AI safety, model evaluation, and responsible AI implementation. Singapore is one of Anthropic's most active global markets, and Claude has been deployed in multiple local sectors including government, finance, and technology.

    Read original
  4. Google DeepMind33

    How Google DeepMind Is Developing Gemini 4 Argon Responsibly and Safely

    Google DeepMind is advancing the widespread deployment of Gemini 4 Argon while expanding the coverage of trusted cyber defenders in the Fairwind Program. The model solved all 12 scenarios in end-to-end simulated cyberattack tests, a significant improvement from 5 in the previous generation. Its refusal rate for violating requests nearly doubled compared to the previous generation, and violations accounted for only 2% of attempts in coding benchmark tests.

    Read original
  5. Google DeepMind43

    Google DeepMind Releases Gemini 4 Argon: Built for Solving Highly Challenging Complex Problems

    Google DeepMind officially released the frontier large model Gemini 4 Argon on October 13, 2026, available globally to developers, enterprises, researchers, and consumers. The model ranks first on the Vals Index benchmark, with input token pricing at $2 per million and output tokens at $10 per million, with a 95% discount for cached input tokens. It has already been deployed in multiple practical applications across engineering optimization, quantum chemistry, cybersecurity, and other fields.

    Read original
  6. Hong Kong Monetary Authority: Press Releases32

    HKMA Welcomes Official Launch of the Chongqing-Hong Kong Cross-Border Data Verification Platform

    The Hong Kong Monetary Authority (HKMA) announced on October 14 the official launch of the Chongqing-Hong Kong Cross-Border Data Verification Platform (CQ-HK DVP), jointly promoted by the HKMA and relevant Chongqing authorities. Leveraging blockchain technology and hash-based data encoding, the platform enables verification of customer-submitted data without cross-border transmission or storage of original documents. Initially focused on cross-border financial scenarios, it supports verification of northbound and southbound data flows for banks in Chongqing and Hong Kong, serving a range of banking services such as account opening and loan applications for customers in both regions.

    Read original
12 OctMon
  1. Alpha Architect68

    LLM Replication Pipeline Examines 1,328 Top Empirical Finance Papers: Reproducible In-Sample but Substantially Decayed Out-of-Sample

    Alpha Architect interprets Dmitriy Muravyev's 2026 working paper, which uses an in-house LLM replication pipeline to reconstruct the core results of 1,328 papers published in the Journal of Finance, Journal of Financial Economics, and Review of Financial Studies between 2000 and 2020, of which 1,005 were eligible for subsequent out-of-sample testing.

    Read original
  2. CFA Institute Research and Policy Center: Publications41

    Where Have All the Portfolio Managers' LLMs Gone?

    Most portfolio managers and analysts in the investment industry have not yet adopted large language models in their daily work, with only a few well-resourced large institutions having implemented them. The article notes that LLMs can improve investment decision hit rates and long-term productivity, and that current barriers to adoption stem primarily from non-technical factors such as organizational processes, incentive structures, and psychological safety.

    Read original
9 OctFri
  1. Aswath Damodaran：Musings on Markets51

    Damodaran: AI Capital Expenditure Effects Emerge, US Stocks Remain Supported Despite Rising Treasury Yields in September

    New York University Professor Aswath Damodaran reviewed the performance of US stocks in September 2026: the 10-year US Treasury yield rose from 4.75% to 5.29% during the month, an increase of 54 basis points, placing it in the top 10% of 770 monthly changes since 1962. Despite this, the total market capitalization of US stocks still increased by $2.5 trillion during the month, with nearly $1.5 trillion of that increase coming from the technology sector.

    Read original
8 OctThu
  1. Anthropic News30

    Anthropic releases 2026 usage policy update

    Anthropic has released an update to its usage policy for 2026, which will take effect on November 12. This update mostly clarifies existing rules, adds integrated specialized provisions for deceptive activities, refines restrictions related to elections, weapons, and surveillance, specifies requirements for high-risk use cases, and adds prohibitions against malicious behavior targeting models. The full policy is available on the corresponding page.

    Read original
7 OctWed
  1. Google DeepMind62

    Google DeepMind Releases Open-Source Lightweight Multimodal Embedding Model EmbeddingGemma 2

    Google DeepMind released EmbeddingGemma 2 on October 6, 2026, based on the Gemma 4 architecture, open-sourced under the Apache 2.0 license, with a total parameter count of 740 million, capable of mapping text, code, images, audio, and video into a unified embedding space.

    Key Takeaways:Learn about EmbeddingGemma 2's modular parameter configuration, vector truncation rules, and on-device memory usage for building local multimodal retrieval and RAG pipelines.

    Read original
1 OctThu
  1. Federal Reserve: Speeches48

    Waller: The Data Version of Godzilla x Kong — FRED vs. AI

    Federal Reserve Governor Christopher Waller reviewed the history of FRED in a speech, calling it the most valuable public good built by the Federal Reserve. FRED now hosts over 850,000 data series and reached 18 million unique users in 2025. He also noted that AI introduces risks such as incorrect data citations and interpretive bias. FRED traffic is currently growing at an annual rate of 150%, with roughly half of visits coming from AI agents. FRED has launched the FRED MCP Connector, allowing users to retrieve data through their preferred AI agents.

    Read original
  2. Google DeepMind34

    Google DeepMind Unveils Gemini 4 Argon: A New Era of Frontier Intelligence

    Google DeepMind announced the launch of the frontier large model Gemini 4 Argon on September 30, 2026. The model is designed for complex long-cycle workflows such as software engineering, legal and financial enterprise knowledge work, and cybersecurity defense, and is currently being phased in to trusted cyber defenders through the Fairwind Program.

    Read original
30 SeptWed
  1. Federal Reserve: Speeches43

    Fed Governor Waller: Payments in the Age of AI Agents

    Fed Governor Waller delivered a speech at the Sibos conference, focusing on the current state and potential impact of AI technology applications in the payments sector. He noted that AI can improve the efficiency of compliance screening for cross-border payments, optimize payment routing and foreign exchange conversion decisions, and also analyzed the two mainstream models and development prospects of agentic commerce driven by AI agents.

    Read original
27 SeptSun
  1. Moontower62

    Moontower Derives the Optimal Kelly Bet Size from First Principles and Discusses AI-Generated Content and 'Thickness'

    Moontower author Kris Abdelmessih released this newsletter, first sharing Adam Mastroianni's article on 'thickness', distinguishing between deep works that require attention to extract value and content-free AI-generated 'slop', then step-by-step deriving the optimal Kelly bet size formula from first principles.

    Read original
10 SeptThu
  1. AQR: Research29

    The core competitiveness of quantitative investment in the AI era lies in human creativity and emotional intelligence

    AQR co-founder and dean of Northwestern University’s Kellogg School of Management co-authored an article pointing out that AI can automate most analytical and technical capabilities in quantitative investment, making creativity and emotional intelligence enduring competitive advantages that are difficult to replace. AQR is leveraging the Quanta Academy, launched in 2015, to expand relevant capability development, while Kellogg School of Management also promotes students’ simultaneous development of analytical skills and soft skills through programs such as MBAi. Research shows that the impact of AI on empathy is not fixed, and depends on tool design choices and organizational culture building.

    Read original
21 AugFri
  1. Aswath Damodaran：Musings on Markets59

    AI's Coming-of-Age Moment: From Hype and Vision to Business Questions

    NYU valuation professor Aswath Damodaran published a lengthy article proposing a four-stage framework for revolutionary change, arguing that AI is currently entering its coming-of-age phase, shifting from investment and infrastructure building to commercial implementation. The article notes that since the release of ChatGPT in November 2022, the Mag Seven contributed 45% of the total increase in U.S. stock market capitalization from 2022 to 2025. Six leading AI companies have cumulatively spent $1.7 trillion in capital expenditures, while the current upper bound of total revenue from AI products and services is only about $250 billion.

    Read original
4 AugTue
  1. Bridgewater: Research & Insights47

    Bridgewater's Thoughts on the Most Important Policy Decision of Our Lifetime

    Bridgewater has published an AI policy commentary arguing that AI is the most impactful technological change of our time, and that policy must be enacted immediately to balance its productivity benefits against risks of social disruption and safety incidents. Bridgewater proposes that US AI policy must prioritize societal resilience to technological shocks and establish a robust AI safety regulatory framework; otherwise, AI development will be forced to stall due to political unsustainability or major incidents.

    Read original
29 JulWed
  1. Model Context Protocol Specification: GitHub Releases19

    Model Context Protocol 2026-07-28 RC Version Released

    Model Context Protocol has released the 2026-07-28 Release Candidate (RC) version. The current specification is provided as a draft, and detailed changes can be found in the corresponding draft changelog. This specification is not yet finalized and may still be adjusted between the RC and official release. Each SDK will adapt to this version at its own pace, and older versions of the specification may continue to be used for an unspecified period.

    Read original