[ad_1]
Balyasny Asset Management is working on building the AI equivalent of a senior analyst by combining several different AI tools, the hedge fund’s head of applied AI told Business Insider.
The $21 billion hedge fund is building its own improved version of ChatGPT, making it available to all employees in June 2023. Balyasny is working on incorporating something called BAM ChatGPT into all of Balyasny’s internal and third-party data sets. . The tool is built using OpenAI’s API, hosted on Microsoft Azure, and currently has 10 different sources available, including transcripts, sales and sell-side commentary, and broker research.
The goal is to create a collection of bots that can proactively push relevant information to PMs and other business teams, such as breaking news about related companies or reporting discrepancies in company disclosures.
“We’re very focused on proactive insights. How do we actually deploy these capabilities when we’re actually doing analytics? In some cases, there are questions that people should be asking. before you know something,” said director Charlie Flanagan. About applied AI, he told BI. “We’re pretty focused on transitioning from junior to senior analysts right now,” Flanagan said of agent capabilities.
Charlie Flanagan, Head of Applied AI, Balyasny Asset Management
Baryasny Asset Management
Hedge funds are hoping to take advantage of the AI wave sweeping Wall Street. Many companies are leveraging AI to improve productivity. Software Engineer at Man Group Two Sigma researchers use it for coding. machine learning Analyze hundreds of market variables. On the other hand, Bridgewater Training AI investors It will be conducting transactions with the customer’s money.
Use AI to analyze data in new ways and automate tasks for analysts
AI is already changing the way Baryasny’s investment team works, helping analysts discover nuggets of information across the fund’s hundreds of data sets, making research easier and saving time by automating time-consuming tasks. doing.
In one case, the head of research for one of BAM’s investment strategies and his team built a set of AI bots to reduce the time it takes to compile a detailed monthly analysis of regularly occurring market events. According to Flanagan, the process he cut from two days to 30 minutes.
Analyzing regulatory filings such as 10-Ks to determine differences in risk and legal disclosures is another area ripe for AI. Currently, analysts use specialized software to perform such analysis, which Flanagan said is a tedious and time-consuming process. Instead, AI automatically identifies discrepancies in a particular company’s disclosures as soon as the filing is published, and sends that information to the analyst responsible for that company via Slack or email. can be actively pushed. The bot can also be prepared to answer any follow-up questions that analysts may have.
and an in-house company responsible for creating the so-called Morning Note, which provides updates in areas such as what happened overnight, what’s relevant, and what news and other data sources are saying. There are many teams. “Once that data is available, we can do it better, faster and faster for people,” Flanagan said.
These agents are not designed to completely replace analysts, but only represent about 10% of their workload. Flanagan said he hopes AI will enable analysts to become editors of these analyzes and summaries, rather than creators.
Flanagan’s team has grown to 10 people in recent months as Balyasny looks to integrate AI into the work of his employees. This team includes researchers, data scientists, and engineers from Wall Street firms such as Citadel and Goldman Sachs and from Silicon Valley.
BAM ChatGPT Details
Rather than having one generic chatbot that can perform a variety of tasks, Balyasny’s strategy is to build highly specialized “agents.” An agent is essentially a bot with very specific tasks and privileges, some of which can be combined to get a richer output. The team divides the work into a number of very specific subtasks, whether it’s an analysis or a sector overview, builds separate solutions for each, and then combines them. This is “the key to achieving productivity gains,” says Flanagan. .
Looking ahead, Flanagan encourages analysts and PM teams to build their own “agents” using the AI building blocks created by the Applied AI team. Some investment teams have already started building their own agents, such as comparing comments from third-party broker surveys about company challenges with what company executives talk about those challenges on earnings calls. Flanagan said they are discovering “properties that lead to
So far, BAM ChatGPT has been deployed by about 80% across Balyasny’s approximately 2,000 employees, Flanagan added.
Balyasny is also working on improving the model’s ability to understand numbers, graphs, and charts that current OpenAI GPT models aren’t sophisticated enough to compute.
“As you can imagine, we have a huge corpus of broker research and all kinds of other reports available in PDF and other formats. Much of the valuable information there is available in tabular and visual formats. It’s hidden within the formality,” Flanagan said.
[ad_2]
Source link
