Home

Neural Networks for Finance - Student Wiki

Welcome to the comprehensive study guide for the Neural Networks for Finance course. This BSc-level lecture series introduces feedforward neural networks with practical applications in finance and stock prediction.

Course Overview

This course covers the fundamentals of neural networks from their biological inspiration to modern applications in finance. You will learn:

  • How artificial neurons model biological decision-making
  • The mathematics behind perceptrons and multi-layer networks
  • Training algorithms including backpropagation
  • Practical considerations for financial applications

Total Duration: 8 lectures (~45 minutes each)

Prerequisites: Basic calculus (derivatives), linear algebra (vectors, matrices), introductory statistics

Lecture Navigation

# Lecture Topics Slides
1 History and Biological Inspiration McCulloch-Pitts, Hebb, biological neurons 18
2 Perceptron Fundamentals Architecture, weights, decision boundaries 32
3 MLP Architecture Hidden layers, forward propagation 32
4 Activation and Loss Functions Sigmoid, ReLU, MSE, cross-entropy 23
5 Gradient Descent and Backpropagation Optimization, chain rule, error propagation 38
6 Training Dynamics and Regularization Overfitting, dropout, early stopping 37
7 Financial Applications Walk-forward validation, case study 27
8 Modern Networks and Future CNNs, RNNs, Transformers, ethics 17
Additional Resources: Glossary Reading List

How to Use This Wiki

For Each Lecture

Each lecture page contains:

  1. Learning Objectives - What you should understand after completing the lecture
  2. Prerequisites - Which lectures to complete first
  3. Key Concepts - Detailed explanations with examples
  4. Key Formulas - Mathematical formulations with intuitive explanations
  5. Charts and Visualizations - Key figures from the slides
  6. Practice Questions - Self-assessment with answers
  7. Reading List - Papers and resources for deeper study

Week 1-2: Lectures 1-2 (Foundations)

  • Understand biological inspiration
  • Master the single perceptron

Week 3-4: Lectures 3-4 (Architecture)

  • Learn multi-layer networks
  • Understand activation and loss functions

Week 5-6: Lectures 5-6 (Training)

  • Master gradient descent and backpropagation
  • Learn regularization techniques

Week 7-8: Lectures 7-8 (Applications)

  • Apply knowledge to finance
  • Understand modern developments

Key Formulas Quick Reference

The Perceptron

y = f(sum(w_i * x_i) + b)

Where:

  • x_i = input features
  • w_i = learned weights
  • b = bias term
  • f = activation function

Gradient Descent Update Rule

w_new = w_old - learning_rate * gradient

Backpropagation (Chain Rule)

dL/dw = dL/dy * dy/dz * dz/dw

Common Activation Functions

| Function | Formula | Derivative | |———-|———|————| | Sigmoid | 1/(1+e^(-x)) | sigmoid(x)*(1-sigmoid(x)) | | tanh | (e^x - e^(-x))/(e^x + e^(-x)) | 1 - tanh^2(x) | | ReLU | max(0, x) | 1 if x > 0, else 0 |

Reading List

Foundational Papers

  • McCulloch & Pitts (1943) - “A Logical Calculus of Ideas Immanent in Nervous Activity”
  • Rosenblatt (1958) - “The Perceptron: A Probabilistic Model”
  • Rumelhart, Hinton & Williams (1986) - “Learning Representations by Back-propagating Errors”

Textbooks

Finance-Specific

  • Lopez de Prado - Advances in Financial Machine Learning
  • Heaton et al. - “Deep Learning for Finance: Deep Portfolios”

Video Resources

  • 3Blue1Brown - Neural Networks series
  • Stanford CS231n - Convolutional Neural Networks for Visual Recognition

Course Materials

Slide PDFs

Download the compiled lecture PDFs from the repository.

Chart Code

All visualizations are available as Python scripts with full source code. Each chart folder contains:

  • Python script (.py)
  • Generated PDF and PNG
  • QR code linking to the code

Contact and Support

This course is part of the Digital Finance curriculum at FHGR.

Repository: Digital-AI-Finance/neural-networks-introduction

QuantLet Mirror: QuantLet/neural-networks-introduction

Last updated: December 2025

↑