You are building an automated trading agent that recommends how many units of an asset to buy or sell and at what price to maximize long-term returns. The agent will be continuously trained on streaming transaction data. Which type of machine learning algorithm is appropriate for this sequential decision-making problem with long-term reward optimization?

Reinforcement learning (RL) is appropriate because it is designed for sequential decision-making problems where an agent learns to perform actions in an environment to maximize a cumulative reward over time. In this scenario, the trading agent's actions (buying/selling assets) influence future states, and the goal is to maximize long-term returns, which aligns perfectly with RL's objective function. Supervised learning is incorrect because it requires labeled data (input-output pairs) to predict outcomes, not to learn optimal action sequences in an interactive environment. Unsupervised learning focuses on finding patterns or structures in unlabeled data, which isn't the primary goal here. Semi-supervised learning combines aspects of both but is still primarily focused on classification or regression with limited labels, not sequential decision-making for long-term reward.