Eth AI Forecaster pulls Ethereum's price history from Yahoo Finance via yfinance, going back to 2018, and uses a machine learning model to forecast the next likely price movement. Built with Python, pandas, and Jupyter notebooks.
Cryptocurrency price data is noisy and volatile, a model that overfits to short-term swings produces forecasts that look precise but mean nothing. The challenge was building a pipeline that could clean and structure years of price history, then produce a forecast simple enough to read at a glance.
Built a pandas-based pipeline to pull and structure historical ETH price data, charted the full trend from 2018 to present, then layered a forecasting model on top that plots predicted next-day movement directly onto the historical chart as distinct marked points, keeping the prediction visually separate from the historical data it's based on.
Data collection, exploratory analysis, forecasting workflow and visual communication of historical and predicted movement.
The notebook separates observed Ethereum history from the next-day forecast so model output is not confused with known market data.
This is an experimental learning project, not a trading system or financial recommendation. Historical behavior does not guarantee future performance.
Charts pulled directly from the project: historical price trend, the forecast table, and the day-ahead prediction overlay.
