Simulation Engines

Economy & Supply-Chain Simulators

Stock-outs cut 40% and service levels held at 97–98%, because every intervention was tested in simulation before deployment. Behind those numbers sit two industry-grade engines: a 23-million-parameter hierarchical MLP (H-MLP) simulating 155 economies and 45 sectors, and a 25-module deep-learning supply chain simulator running on 1.2 million IoT signals a day.

Challenge

Economic and supply-chain systems are deeply interconnected — economies, sectors, and logistics networks all move together. Conventional planning tools and standard forecasting models struggle to capture those interactions well enough to test an intervention before committing to it.

Approach

The economy engine is a 23M-parameter H-MLP simulating 155 economies and 45 sectors, evaluated over 12-quarter forecasts against VAR-X, DSGE, and Transformer benchmarks. The supply chain engine is a 25-module simulator ingesting 1.2 million IoT signals per day, modelling the chain end to end so interventions could be tested in simulation before deployment.

Outcome

The economy simulator beat the VAR-X, DSGE, and Transformer benchmarks by 30% over 12-quarter forecasts — results published as research. The supply chain simulator lifted service levels to 97–98%, cut stock-outs by 40%, reduced logistics cost by 11.2%, and improved forecast MAPE by 44%. The methodology is published in two books: Simulating the Entire World Economy with Multilayer Perceptrons (ISBN 979-828080999-4) and Creating a Simulation of an Entire International Supply Chain Using Multi-Layer Perceptrons (ISBN 979-828041476-1).

Stack & Methods

• Hierarchical multilayer perceptrons (H-MLP), 23M parameters
• 25-module deep neural network supply chain simulator
• 1.2 million IoT signals per day data pipeline
• Benchmarking vs VAR-X, DSGE, and Transformer models over 12-quarter forecasts
• Methodology published in two books (ISBN 979-828080999-4 and 979-828041476-1)