Ernest Chan – Generative AI for Asset Managers Workshop Recording

About Course

Workshop Overview

Day 1:

Exploring Generative AI and Large Language Models (LLMs) in Asset Management

• In depth look, into the applications of LLMs such as BARD, ChatGPT in the industry.
• Utilizing LLMs to develop trading approaches.
• Practical session; Converting data into signals for high frequency trading.

Day 2:

Advanced Strategies and Real World Uses

• Informal Conversation with Lisa Huang
• Discussion, on design and methods to manage risks associated with LLMs.
• Enhancing trading tactics with detailed sentiment analysis through LLMs.
• Hands on activity; Testing trading strategies and exploring how LLMs could transform asset management practices.

Workshop Outline

01 Exploring Big Language Models (BLMs) & Pre trained Generative Transformers (PGT)

• Getting familiar, with BLMs like BARD, ChatGPT and other advanced language models
• Common Uses of BLMs
• Understanding the functionality of BLMs
• Accessing BARD/PaLM online using their API

02 Developing Software

• Introduction to Prompt Design
• Creating software, for tasks like writing text summarizing content and more.
• Exploring few shot learning, with BARD
• Introduction to embeddings and their significance
• An overview of the BARD embeddings API. How it is utilized

03 Risks Linked with Language Models (LLMs)

• Recognizing risks associated with LLMs, including hallucinations, bias, consent and security.
• Strategies, for mitigating the risk of hallucinations, such as retrieval enhancement, prompt manipulation and self analysis.
• Techniques for identifying and managing hallucinations, including reinforcement learning based on feedback (RLHF) and model driven approaches.

04 Utilizing Language Models for Analyzing Federal Reserve Chairs Speeches

• Reasons for selecting the BARD family over LLMs.
• Assessment of BARDs performance.
• Enhancing performance through embeddings.
• Practical demonstration; evaluating sentiment scores on companies using embeddings.
• Test data; Video recordings of the Federal Reserve Chairs press conferences.
• Conducting an analysis of a trading strategy based on the sentiment analysis provided by an LLM.

05 Implementation of Language Models in Real world Scenarios

• Practices for deploying LLMs in production environments.
• Overview of models, like ChatGPT, BART, Cohere, Alpaca, etc.