Stock Insights

Home Page
Sources
Some News Snippets

Inspiration

The inspiration for this app stemmed from observing how businesses are profoundly influenced by macroeconomics, business policies, and the surrounding environment. Equity investors diligently track news related to these factors. While rudimentary news tagging solutions exist in the market, they rely solely on keyword occurrence. With the advent of vector databases and embeddings, a more sophisticated approach has become feasible. Our project leverages the contextual information of stocks, going beyond simple keywords, to accurately tag news articles to specific stocks.

What it does

NewsMapper, revolutionizes the way equity investors stay informed. It goes beyond traditional news tagging methods by utilizing advanced vector databases and embeddings. NewsMapper seamlessly maps news articles to specific stocks based on the rich context captured in their latest earnings calls. This enables us to tag news content with precise metadata, providing investors with a comprehensive view of news related to their favorite stocks.

How we built it

We built NewsMapper by leveraging RSS feeds from leading news publishers, specifically focusing on their business news content. Through a two-step process, we map the news articles to relevant stocks. Firstly, we capture the context of each stock by analyzing its earnings calls and utilizing Pinecone vector databases and OpenAI embeddings to tag the most similar stocks. Secondly, we pass this highly relevant stock content to OpenAI chat-completion model, employing retrieval chains in LangChain to understand the impact of the news on the stock as positive or negative.

Challenges we ran into

We encountered a few challenges while developing StockInsights. Besides the expected data pre-processing and cleaning hurdles, we faced issues with OpenAI due to their system overload, which temporarily impacted our operations. Additionally, as newcomers to LangChain and Pinecone tools, we had a learning curve in effectively utilising their capabilities

Accomplishments that we're proud of

We are proud of working on this novel idea and what we have achieved within a short timeframe of just one week. A stable web-app, with a user-friendly experience, having worked on some cutting-edge technologies in the backend with seamless integration is something we're proud of.

What we learned

Familiarizing ourselves with new technologies like LangChain and Pinecone was a valuable learning experience. In our development journey, we gained insights into enhancing NewsMapper with a focus on user-centric design for upcoming versions.

What's next for Stock Insights

Our future plans for NewsMapper include offering a way for end-users to specify their own news feeds and build user-centric features like alerts & watchlists. For StockInsights as a product, we want to incorporate features that enable investors to analyze qualitative data from various non-financial resources alongside financial data. We want to build the product towards empowering investors with powerful yet easy-to-use tools for informed decision-making.