For a long time it was believed that trying to forecast stock prices from stock markets was an absurd idea given the unpredictable nature of such  markets, and, indeed, many studies have suggested that. However with the rise of machine learning and data mining, new approaches for this problem were born and some of them achieved promising results. In the present study we implement and evaluate a predictive model that aims to forecast stock market prices based on a Temporal Association Rule Mining technique. The model is tested with brazilian stock market data and its results are compared to the traditional Buy and Hold strategy. The results here are also promising, since it is shown that the model is capable of overcoming the Buy and Hold baseline despite the fact that it makes a lot of wrong predictions.