Lesson 8 of 12 · All lessons in this module
- Fintech and the four Vs of Big Data
- Sources of Big Data and its challenges
- Financial data science and what makes financial data hard
- AI and how machine learning learns
- Splitting data for machine learning: size, leakage and time order
- Overfitting, underfitting and model fit
- Supervised, unsupervised and deep learning
- Reinforcement learning and the four learning approaches compared
- Inside a neural network: layers, backpropagation, GANs and VAEs
- Data science: processing and visualisation
- Text analytics, NLP and data science tools
- Large language models and generative AI in finance
This lesson is part of the 2027 curriculum. You are following the 2026 curriculum, where it is not taught in this form. Switch if you are sitting the exam under the 2027 curriculum.
Lesson 8 of 12 · 11 min
Reinforcement learning and the four learning approaches compared
Besides supervised, unsupervised and deep learning, machines can learn by trial and error from rewards and penalties (reinforcement learning), and each approach has typical uses, strengths and weaknesses the exam can ask about.
In short
- Reinforcement learning: the algorithm interacts with an environment, takes actions and learns from rewards or penalties; no labelled data are needed.
- Finance use: a trading strategy that learns when to buy, sell or hold by rewarding risk-adjusted performance; also portfolio optimisation.
- Supervised: accurate with enough labelled data and well understood, but needs large labelled datasets and may not generalise to entirely new scenarios.
- Unsupervised: finds hidden groupings and can reduce dimensionality, but results can be hard to interpret, may not be meaningful, and judging them is subjective.
- Reinforcement: good for sequential decisions in changing environments, but sensitive to the reward design and costly in computing power and training time.
- Deep learning: models complex non-linear patterns and extracts features automatically, but needs large data and computing power, is a black box and can overfit.
Unlock this lesson free for 7 days
Create a free account to get 7 days free: every lesson, video and flashcard, and the question bank. No card needed.