This study focuses on applying reinforcement learning techniques in real time trading. We first briefly introduce the concept of reinforcement learning, definition of a reward function, and review previous studies as foundations on why reinforcement learning can work, specifically in the setting of financial trading. We demonstrate that it is possible to apply reinforcement learning and output valid and simple profitable trading strategy in a daily setting (one trade a day), and show an example of intraday trading with reinforcement learning. We use a modified Q-learning algorithm in this scenario to optimize trading result. We also interpret the output policy of reinforcement learning, and illustrate that reinforcement learning output is not completely void of economic sense.