This document compares different forecasting models for daily stock prices: linear regression, Theil's incomplete method, and multilayer perceptron (MLP). Principal component analysis was used to reduce 4 stock price variables to 1 principal component, which was then used to predict closing prices. Linear regression and Theil's method produced similar results, with MAE around 110 and R-squared over 0.99. MLP had slightly higher error at 118 MAE. Overall, linear regression and Theil's method provided the best forecasts of closing stock prices based on this analysis of models and error metrics.