Neural Network Option Pricing

This site documents a Machine Learning for Finance project on neural network surrogates for European option pricing under Black-Scholes dynamics.

The project investigates the following question:

Can a feed-forward neural network accurately approximate the Black-Scholes pricing function for European call options?

The documentation is organized into two complementary parts:

• narrative documentation, which explains the mathematical and experimental choices;

• API documentation, which is generated from the Python docstrings.

Project Documentation

• Project Overview
  • Main Components
  • Final Deliverables
  • Repository Layout
• Methodology
  • Black-Scholes Dynamics
  • European Call Pricing
  • Supervised Learning Task
  • Monte Carlo Benchmark
  • Controlled Noisy Targets
  • SVR Baseline
• Experiment Pipeline
  • Generated Artifacts
  • Runtime Benchmark
  • Auxiliary Experiment Scripts
  • Fast Smoke Test
  • Intermediate Experiment
• Experiments
  • Diagnostic Plots
  • Final Experiment
  • Additional Experiments
  • Future Extensions
• Reproducibility
  • Environment
  • Final Experiment
  • Tracked Results
  • Validation Commands
  • Reproducibility Notes
• Command Reference
  • Main Experiment
  • Runtime Benchmark
  • SVR Benchmark
  • Noisy-Target Neural-Network Experiment
  • Noisy-Target SVR Benchmark
  • Report Figures
  • Testing and Documentation
• Testing and Validation
  • Current Test Coverage
  • Testing Philosophy
• API Reference
  • nn_option_pricing.black_scholes
  • nn_option_pricing.config
  • nn_option_pricing.dataset
  • nn_option_pricing.evaluation
  • nn_option_pricing.model
  • nn_option_pricing.monte_carlo
  • nn_option_pricing.noise
  • nn_option_pricing.noisy_experiment
  • nn_option_pricing.noisy_svr_experiment
  • nn_option_pricing.pipeline
  • nn_option_pricing.plots
  • nn_option_pricing.svr
  • nn_option_pricing.training