Investment in stocks has recently become a trend among the public. The problem that often occurs in stocks is that it is difficult to predict the increase or decrease in stock prices. In July 2019, Bank Mandiri experienced a system failure that caused customer balances to increase or decrease drastically. Incidental to the emergence of various opinions on social media and news media that spread the information. Based on this incident, a system is needed to see the relationship between sentiment and the company’s stock price. This research implements the Backpropagation Neural Network method to analyze sentiment and Word2vec feature expansion is used to find similarity words and reduce vocabulary mismatches in textual data. The result of this researchs, using feature expansion in the classification algorithm can improve performance up to 0.7% with an accuracy value of 72.58%. The sentiment analysis results on stock prices show that the sentiment value and stock prices have a very weak relationship.