16. Quiz

What is feature engineering in the context of reinforcement learning for trading?
SOLUTION: Selecting, modifying, and creating new features from raw financial data.

Why is high-quality feature engineering important in reinforcement learning for trading?
SOLUTION: It enhances the learning efficiency and predictive power of the model.

What special consideration must be taken into account when defining state spaces in financial markets?
SOLUTION: Non-stationarity of the data.

What historical data is needed to calculate the 5-day moving average (SMA)?
SOLUTION: Data from the current day and previous 4 days.

What is the purpose of feature vectors in the context of reinforcement learning and financial markets?
SOLUTION: To provide a numerical representation of training features at a given time step.

Why is normalization important when constructing feature vectors for reinforcement learning models?
SOLUTION: To prevent features from being distorted by their scales and to stabilize learning.

What is the primary goal of normalization in data preprocessing for AI models?
SOLUTION: To adjust feature values to a common scale without distorting differences.

What happens if rolling normalization is not used and a new data period shows significantly different values from previous periods?
SOLUTION: The model might incorrectly perceive the new data as having more importance, affecting its performance.

What are the three possible actions in a financial market RL trading agent's action space?
SOLUTION: Buy, Sell, and Hold

What is a primary consideration when defining the granularity of actions in the action space?
SOLUTION: The level of control and the size of the action space.