Methodology
A short description of the current simulation model and its implementation.
Simulation
Each run generates 10,000 Monte Carlo paths. Innovations are sampled from the seven-dimensional historical innovation data using multivariate kernel density estimation with component bandwidths based on Silverman's rule of thumb.
Market factors
The model evolves volatility, the Treasury spread, the BAA corporate bond rate, earnings, and the stock valuation measure. These factors enter the return equations for U.S. stocks, international stocks, and corporate bonds.
Portfolio
The stock portfolio combines U.S. and international returns. It is then combined with corporate bonds. The bond share changes linearly from the selected starting allocation to the selected ending allocation.
Cash flows
The simulator supports contributions and withdrawals at annual, quarterly, or monthly frequency. The initial cash flow can increase or decrease annually.
Results
The interface reports portfolio survival, mean and median final wealth, the 10th and 90th percentiles, and selected wealth paths ranked by final wealth.
Implementation check
The separated Python engine was compared with the existing implementation using the original 97 × 7 innovation data. Full wealth-path outputs matched exactly for withdrawal, contribution, and no-flow test scenarios under identical random seeds.