SteamPulse

Inspiration

This app is suitable for any day trader, regardless of their level of financial or trading expertise. By providing clear, actionable insights and a simple interface, we make it easy for both beginners and experienced traders to leverage market predictions. Our goal is to empower users to make data-driven decisions without requiring a deep background in finance.

What it does

Our app leverages machine learning to analyze stock market data and provide real-time predictions on future price movements. It simplifies complex financial data, making it accessible and actionable for day traders. Users receive insights, such as anticipated price trends and volatility levels, which they can use to make informed trading decisions. Designed for both novice and experienced traders, the app removes the need for in-depth financial knowledge by offering an intuitive interface and clear recommendations based on predictive analytics.

How we built it

We started by gathering historical stock market data and preprocessing it to ensure consistency. Using a Random Forest Regressor as our baseline model, we tested its performance in predicting price changes. Then, we integrated XGBoost to refine our predictions, taking advantage of its gradient boosting capabilities for better handling of complex patterns. Our workflow involved multiple iterations of model training, testing, and backtesting to fine-tune performance and minimize prediction error.

Challenges we ran into

One of the biggest challenges was dealing with a gap in our data from 01/2023 to 01/2024. This lack of continuity caused issues during backtesting in 2024, as the model struggled to account for trends from the missing year. Additionally, tuning the models for optimal performance required extensive experimentation, and managing large datasets without overfitting posed further difficulties. Despite these hurdles, we managed to develop a solution that provides valuable insights for traders in volatile markets.

Accomplishments that we're proud of

We’re proud to have built an accessible app that empowers both novice and experienced traders with reliable stock predictions. Through dedicated 24-hour teamwork, we achieved high accuracy using Random Forest and XGBoost, and we tackled challenges like handling missing data to maintain model integrity. Our collaborative effort produced an intuitive tool that enables informed trading decisions without requiring deep financial knowledge.

What We Learned

We deepened our knowledge of machine learning and financial forecasting, gaining insights into trading terms and methods like price trends and volatility. Working with Random Forest and XGBoost models taught us the importance of data preprocessing, feature engineering, and parameter tuning for accuracy. Our 24-hour team effort highlighted the value of collaboration and time management in overcoming technical challenges and building an accessible, user-friendly app.

What's Next for StockPulse AI

Currently, StockPulse AI provides daily stock change insights and calculates daily gains on Tesla investments throughout 2024. This capability lays a strong foundation for expanding into a comprehensive trading assistant. By refining our model and adding support for more stocks, we could offer personalized investment insights across multiple sectors, helping users diversify and manage portfolios effectively. Future updates could include features like intraday predictions, real-time alerts, and risk assessment tools, making StockPulse AI an invaluable tool for day traders and long-term investors alike. Ultimately, StockPulse AI has the potential to evolve into a robust financial platform that empowers users to make informed, data-driven investment decisions.

Built With

• alpha-vantage
• amazon-web-services
• firebase
• github
• numpy
• pandas
• plotly
• scikit-learn
• sqlite
• streamlit
• technologies-used-python
• xgboost
• yahoo-finance