Stock Advisories

Inspiration

Traditional stock analysis relies heavily on price data and technical indicators, often ignoring alternative signals that reflect real-world interest and sentiment. We were inspired by how search behaviour and public attention can influence markets, and wanted to explore whether Google Trends data could provide meaningful insights into stock performance.

What it does

Stock Advisorys is a full-stack stock analysis platform that combines traditional market data with alternative data sources. Users input a stock ticker, and the system generates six related keywords using a neural network to capture context and sentiment around the stock.

It then retrieves Google Trends data for each keyword and combines this with historical stock data from yfinance. A second neural network processes these signals to produce a comprehensive analysis of the stock, highlighting momentum, sentiment, and potential market impact.

How we built it

The backend is built in Python using FastAPI, integrating yfinance for historical stock data and pytrends for Google Trends data. A neural network is used to generate related keywords for each stock, expanding the context beyond a single ticker.

We then extract features from the trend data (such as slope, volatility, and peak interest) and combine them with stock price data. A second neural network processes these combined features to generate insights and predictions.

The frontend is built using React with Vite and TailwindCSS, providing a clean, responsive dashboard with charts and visualisations for both trend data and stock performance.

Challenges we ran into

One of the main challenges was working with unreliable and rate-limited external data sources like Google Trends. Ensuring consistent data collection required caching and careful request handling.

Another challenge was integrating multiple stages of the pipeline, from keyword generation to feature extraction and final prediction, while keeping performance fast enough for a responsive UI.

We also had to balance model complexity with practicality, ensuring the system worked reliably within the constraints of a hackathon environment.

Accomplishments that we're proud of

We successfully built a multi-stage machine learning pipeline that combines alternative data with traditional financial data. The integration of keyword generation, trend analysis, and stock data into a single cohesive system is something we are particularly proud of.

We also created a clean and intuitive frontend that makes complex data easy to understand, allowing users to quickly explore and analyse different stocks.

What we learned

We learned how powerful alternative data can be when combined with traditional financial metrics, and how even simple models can extract useful signals when the right features are used.

We also gained experience in building end-to-end machine learning systems, handling real-world data limitations, and designing systems that are both technically sound and user-friendly.

What's next for Stock Advisories

In the future, we would like to improve the accuracy of the models by incorporating more data sources, such as news sentiment, social media activity, and earnings reports.

We also plan to enhance the modelling approach with more advanced architectures and backtesting to validate predictive performance.

Finally, we aim to expand the platform into a more complete financial analysis tool with portfolio tracking, alerts, and real-time signal generation.