The AI component helps interpret what the chart is showing by analyzing the overall trend, the stock's position relative to VWAP, trading volume patterns, recent highs and lows, and any saved support or resistance zones. In addition, a Python based multi model prediction system can capture chart screenshots, send them to multiple AI models for analysis, and combine their insights for a prediction report. Instructions on how to use are in the readme file
Removed all Render related deployment files and configuration from the project including render.yaml, Render specific requirements, and documentation references. Replaced Render deployment setup with a local Python server setup only. Fixed Vercel deployment issue by updating server.py to expose a proper FastAPI app entrypoint and restructuring dependencies to match Vercel runtime requirements. Pushed updated changes to main branch and triggered redeployment.
I successfully fixed the deployment issue on Render and got the backend service running in production. The environment is now building cleanly, dependencies are installing correctly, and the server is live and accessible through the production URL. The application is stable after deployment and ready for testing or further integration.
It should work on the URL NOW
Before participating in the HackTime challenge, I had the initial idea of creating a stock analysis platform. On June 5, I officially started developing StocksAI Market Desk, a web based platform focused on U.S. equities.Since then, I have built a functional trading interface that includes real time stock data, interactive charts, watchlists, replay functionality, technical indicators such as EMA and VWAP, and custom drawing tools for support and resistance zones, trend lines, and notes. All of this is designed to give a clear and structured view of market movement.The interface includes several key features. A ticker search and live info bar provides real time price data, including open, high, low, close, volume, and percentage change, along with a live candle timer for precise interval tracking. The chart system supports multiple timeframes and includes annotation tools such as crosshair, trend lines, support and resistance boxes, and text notes. Users can also overlay comparison tickers and toggle technical indicators like EMA 9, EMA 21, and VWAP.The platform also includes market replay functionality for backtesting, allowing users to step through historical price action bar by bar. A price alert system is also integrated, enabling conditions based on price movement above or below selected levels. In addition, a quick watchlist provides fast access to major stocks and ETFs such as SPY, QQQ, AAPL, NVDA, MSFT, and TSLA.A built in analysis layer is included in the interface, where the system generates structured insights based on price action, trend direction, and indicator alignment. This is paired with a manual analyze trigger for deeper updates when needed.I used Codex extensively to help develop the charting system, user interface, and overall application structure. I also began building the foundation for a system that processes and compares outputs from multiple external analysis providers through API integration. Additionally, I created the framework for connecting external services such as OpenAI, Gemini, and Anthropic through API keys. However, these integrations are not yet functional, and the analysis and prediction features are still under active development. At this stage, the project serves as a working prototype and foundation for a more advanced trading platform that combines technical analysis, market visualization, and multi source market insights. In addition, I explored FinImpulse as a potential data infrastructure layer for the project. It provides API based access to financial market data, allowing applications to retrieve structured stock information such as real time prices, indicators, and other market metrics. I set up the developer configuration area to prepare for future integration, which will allow the platform to pull reliable market data directly into the system and later feed it into the analysis pipeline once the features are fully implemented.
Today the 29 min will be logged because I improved the stock search experience. Now when a user types a ticker, a dropdown with matching results appears, and selecting an option automatically triggers the search and loads the chart without needing to press the search button.
