Inspiration
Traders are required to perform technical analysis to gain deeper insight in the stock price of a company. To do this they rely on grunt work and outdated tools which makes the process slow and tedious, we aim to empower traders with a solution which leverages LLMs to do the grunt work for them, so that traders can focus on more important tasks. The llm does not give advice, it just does what it is told to do.
What it does
The user can upload reports filled by the company (quarterly, annual, financial reports). Once the documents have been uploaded, the user can ask the llm all the technical details from which the trader can gain an advantage.
How we built it
We built this by locally running the llm, so that we can get a faster response time and can have a much better control over the information that we can feed to it. The backend is built in flask, and the front end is built using reactjs. For the llm workflow we are testing out different open source models such as Llama 3.1, DeepSeek R1.
Challenges we ran into
LLM are a new technology so it was a given that we would need to do a lot of research of our own to find the most optimal way to leverage this latest technology, this was our biggest problem.
Accomplishments that we're proud of
We are proud that the system we have developed is built using open source and locally run llm.
What we learned
The biggest learning has been being able to create the software we aspire to use.
What's next for LLM powered technical analysis
For future scope we would like to add more features to the platform and perform calculations directly in the system
Log in or sign up for Devpost to join the conversation.