Download PDF Open in New Tab

Table of Contents

Introduction (Slides 1-3)

  • 1. Title Slide
  • 2. Learning Objectives
  • 3. Why Synthetic Financial Data?

Foundations (Slides 4-6)

  • 4. Financial Time Series Properties
  • 5. Fat Tails & Volatility Clustering
  • 6. Generative Model Taxonomy

GANs (Slides 7-12)

  • 7. GAN Architecture
  • 8. GAN Training Dynamics
  • 9. Mode Collapse Problem
  • 10. WGAN & Gradient Penalty
  • 11. TimeGAN Architecture
  • 12. Conditional GANs

VAEs (Slides 13-15)

  • 13. VAE Architecture
  • 14. ELBO Derivation
  • 15. Latent Space Visualization

Diffusion (Slides 16-20)

  • 16. Diffusion Forward Process
  • 17. Reverse Diffusion
  • 18. SDE Formulation
  • 19. GBM-Diffusion for Finance
  • 20. Normalizing Flows

Private Credit (Slides 21-25)

  • 21. Private Credit Overview
  • 22. Loan-Level Simulation
  • 23. Fund-Level Simulation
  • 24. Correlation Structure
  • 25. Synthetic vs Real Comparison

Evaluation (Slides 26-28)

  • 26. Evaluation Metrics
  • 27. Model Comparison
  • 28. VaR & Stress Testing

Conclusion (Slides 29-30)

  • 29. Research Frontiers
  • 30. Conclusion & References

Key Mathematical Formulations

GAN Objective

minG maxD E[log D(x)] + E[log(1-D(G(z)))]

WGAN with Gradient Penalty

LGP = lambda E[(||grad D(x)||2 - 1)2]

VAE ELBO

L = Eq[log p(x|z)] - DKL(q(z|x)||p(z))

Diffusion Forward Process

q(xt|x0) = N(xt; sqrt(alphat)x0, (1-alphat)I)

Score-Based SDE

dx = f(x,t)dt + g(t)dW

Ornstein-Uhlenbeck

drt = theta(mu - rt)dt + sigma dWt

Presentation Details

Total Slides30
ThemeBeamer Madrid
Font Size8pt
Aspect Ratio16:9
Charts Included24
DateJanuary 2026