kirst-dk/stockpilot-ai

StockPilot AI is a portfolio manager for tokenized stocks built on top of a blockchain network called Mantle. Tokenized equities are digital tokens on a blockchain that represent real-world stocks, for example, holding a token called SPYx is equivalent to holding exposure to the S&P 500 ETF. The project was built for a hackathon focused on AI and real-world asset tokenization. The way it works is that a user picks one of three investment strategies, Balanced Growth, Momentum, or Value Investing, and an AI agent takes over from there. The agent fetches live prices for supported tokenized stocks (SPYx, NVDAx, AAPLx, TSLAx, MSFTx, AMZNx), runs the chosen strategy to decide what to buy, sell, or rebalance, and then uses an AI model to validate and refine those recommendations before executing trades on-chain. Every action, buy, sell, rebalance, is recorded permanently in a smart contract on Mantle, so the agent's track record is fully verifiable. Trades are executed either via the xStocks atomic settlement system (which mints or redeems tokens at live market prices) or through the Fluxion decentralised exchange, which finds the best available price across different liquidity pools. The three strategies differ in risk tolerance: Balanced Growth is conservative with automatic stop-loss and take-profit levels, Momentum follows price trends, Value Investing buys on price dips and trims on extended rallies. The tech stack includes a Python backend with FastAPI, smart contracts written in Solidity, and a Next.js frontend dashboard showing portfolio state, market data, AI recommendations, and trade history. The full README is longer than what was provided.