Inspiration
Are you a fundamental stock investor?
In the fast-paced world of stock investing, staying updated with the latest financial performances of companies is crucial. As fundamental stock investors ourselves, we understand the challenges many people are facing - keeping track of earnings releases amidst a busy life of work, study, or travel can be daunting. Missing out on these critical updates can sometimes lead to significant losses, especially when markets react before you get the chance to. That's where the idea for StockPulse was born.
What it does
StockPulse is crafted for retail investors who aren't working full-time in investment funds but are passionate about managing their investments wisely. Whether you're a seasoned investor or just starting, StockPulse offers you the insights you need without overwhelming you with information.
StockPulse is your vigilant financial assistant, designed to monitor earnings announcements from companies you're invested in. Upon detecting a new earnings release, StockPulse swiftly retrieves the report directly from the company's website, analyzes it, and sends you an alert about potential significant stock movements. This real-time information empowers you to make informed decisions swiftly.
How we built it
Our project seamlessly integrates frontend and backend components, leveraging a variety of scripts, packages, and APIs to deliver a comprehensive solution. Here's a breakdown of the technologies and methodologies we employed:
Frontend
The user interface of our application is powered by the ReactJS framework, utilizing HTML and CSS for structuring and styling, respectively. JavaScript plays a crucial role in dynamically managing frontend components and facilitating interaction with the backend. To ensure a smooth and continuous deployment process, we hosted the frontend on GitHub Pages and utilized GitHub Actions for CI/CD, enabling automatic updates and deployments.
Backend
Our backend infrastructure is built on the FastAPI framework, known for its high performance, and is served via a Uvicorn ASGI server. This combination offers a robust and scalable solution for our application's needs.
To gather the latest earnings and financial releases, we employed BeautifulSoup and PyPDF2 for scraping data from companies' investor relations websites, converting it into a text format for further processing. Additionally, historical financial data is sourced from Yahoo Finance using the yfinance library. Given Yahoo Finance's update latency of 2-3 days post-release, our direct scraping from investor relations sites ensures timely data retrieval, critical for our analysis.
We utilize OpenAI's GPT-3.5-turbo model to distill key financial insights from the scraped text data, providing succinct summaries of earnings releases. This processed information, combined with data from Yahoo Finance, feeds into our predictive model to assess potential significant stock price movements following an earnings announcement.
At the heart of our predictive capability is a custom-built neural network model, developed using TensorFlow. This model analyzes the amalgamated data to forecast potential shifts in stock prices, offering our users valuable insights to inform their investment decisions.
Challenges we ran into
One of the primary challenges we faced was sourcing the financial data necessary to train our model. Accessing comprehensive and accurate data often required premium API subscriptions, which posed a significant obstacle. Additionally, constructing a meaningful AI model within a constrained timeframe proved to be a daunting task. These challenges tested our resolve and pushed us to think creatively in our approach. The front-end team presented a significant challenge in bridging the connection between the back end and the front end. Establishing an environment outside of our usual framework was crucial to gaining a comprehensive understanding of how the back end operates. This process allowed us to integrate the systems effectively and ensure the webpage functioned seamlessly. It required meticulous attention to detail and considerable time to overcome these technical hurdles successfully
Accomplishments that we're proud of
Embarking on the StockPulse project was a journey of many firsts for us. It marked our inaugural venture into developing a live project as a team within a limited timeframe. For all of us, this was the first hackathon experience, introducing us to the exhilarating process of transforming an idea into reality alongside a newly formed team. This experience was a crucible of learning, from mastering project management skills to rapidly acquiring and applying new coding techniques.
What we learned
We take pride in not just the technical skills we've honed but also in the invaluable lessons learned in teamwork, perseverance, and innovation.
What's next for StockPulse
As we look to the horizon, we have identified several key enhancements to elevate StockPulse's capabilities:
- Enriching AI Analysis: Integrating additional factors into our AI's analytical framework to provide deeper insights.
- Model Optimization: Refining our AI model to achieve greater accuracy and reliability in its predictions.
- Comprehensive Monitoring: Extending our platform's capabilities to include alerts on news and events related to tracked companies, offering a more holistic view of potential impacts.
- Market Trend Detection: Analyzing real-time stock prices to identify emerging market trends, enabling users to make more informed decisions.
- Dashboard Implementation: Developing a user-friendly dashboard to offer users a comprehensive overview of their portfolio and individual stock performance.
Log in or sign up for Devpost to join the conversation.