Reinforcement learning is the third type of ML. It aims to choose the action that yields the highest reward, given a set of input data that describes a context or environment. It is both dynamic and interactive: the stream of positive and negative rewards impacts the algorithm's learning, and actions taken now may influence both the environment and future rewards. 

The trade-off between the exploitation of a course of action that has been learned to yield a certain reward and the exploration of new actions that may increase the reward in the future gives rise to a trial-and-error approach. Reinforcement learning optimizes the agent's learning using dynamical systems theory and, in particular, the optimal control of Markov decision processes with incomplete information.

Reinforcement learning differs from supervised learning, where the available training data lays out both the context and the correct decision for the algorithm. It is tailored to interactive settings where the outcomes only become available over time and learning has to proceed in an online or continuous fashion as the agent acquires new experience. However, some of the most notable progress in Artificial Intelligence (AI) involves reinforcement that uses deep learning to approximate functional relationships between actions, environments, and future rewards. It also differs from unsupervised learning because feedback of the consequences will be available, albeit with a delay. 

Reinforcement learning is particularly suitable for algorithmic trading because the concept of a return-maximizing agent in an uncertain, dynamic environment has much in common with an investor or a trading strategy that interacts with financial markets. This approach has been successfully applied to game-playing agents, most prominently to the game of Go, but also to complex video games. It is also used in robotics—for example, self-driving cars—or to personalize services such as website offerings based on user interaction. We will introduce reinforcement learning approaches to building an algorithmic trading strategy in Chapter 21, Reinforcement Learning.