PROJECT 03 / Data analysis · Time-series forecasting
Stock Market Analysis & Prediction
Exploratory stock analysis and an LSTM forecasting experiment using historical prices for AAPL, GOOG, MSFT and AMZN.
01
The problem
Historical prices contain trends, volatility and relationships between stocks. This project explores those patterns and experiments with predicting a next-day closing price from recent observations.
02
My approach
Analyzed more than one year of historical stock data using Python, Pandas and NumPy. Explored moving averages, daily returns and cross-stock correlations to understand trend and risk patterns.
03
How it works
Built a two-layer stacked LSTM using a 60-day lookback window of MinMax-scaled closing prices and a 95/5 training and testing split.
04
What it demonstrates
A learning project spanning exploratory analysis, sequence preparation and neural-network time-series modeling.
05
Limitations & next steps
No forecasting accuracy or investment return is claimed. A stronger evaluation would document chronological splitting, training-only scaling, rolling backtests and comparison with a simple persistence baseline. These are proposed next steps. The project is for learning, not investment decisions.
KEEP EXPLORING
TruthLens AILoan Bias AuditorSri Balaji Earth Movers