Skip to content

Simulation Models Backlog

Candidate simulation approaches for stock market prediction, ordered roughly by implementation complexity and relevance to the current stack.

Prioritized

Model Status Notes
Monte Carlo Backlog 10k price path scenarios; output highest win-rate moves as probability distribution on top of RF
Bootstrap Resampling Backlog Resample historical returns instead of assuming a distribution; more realistic fat tails
Regime-Switching Backlog Bull/bear/sideways state transitions; integrates with existing GARCH + K-means vol regime
VaR (Value at Risk) Backlog Maximum expected loss at a given confidence level (e.g. 95%); pairs directly with existing GARCH and vol regime work
Parkinson Volatility Backlog Alternative vol estimator using high/low prices instead of close-to-close; easy addition alongside existing realized vol calculations

On Radar

Model Notes
Geometric Brownian Motion (GBM) Classic Black-Scholes path model; assumes log-normal returns and constant vol — fast but unrealistic; useful as a baseline comparison
Historical Scenario Replay Replay known crashes (2008, COVID) against current positions for stress testing
Agent-Based Model (ABM) Simulate thousands of traders with different strategies; captures herding, panic selling, microstructure effects — high complexity
Game Theory Simulation Model market as a strategic game between participants — experimental
Quantum Monte Carlo Used by some hedge funds; experimental and overkill for this stack