Models

REVEAL THE VALUE AND MANAGE THE RISK

At AD&Co, our models reveal the value, as well as manage the risk in your mortgage loan and related mortgage-backed security (MBS) portfolios. Our innovative analytics help to evaluate borrower behavior by forecasting prepayment, delinquency, default and loss probabilities. In addition, by reacting dynamically to changes in economic and market conditions, our models deliver more accurate forecasts for use in the pricing, valuation, and risk assessment of loans and securities.

Interested in a product demo?

LDM Logo LoanDynamics (LDM) is our flagship prepayment and credit model. It comes in several mortgage versions depending on asset class: Agency LDM and Agency LDM+; Non-Agency LDM, and Multifamily LDM.

RiskValDynamics logoRiskValDynamics is our flexible valuation engine. It can evaluate mortgage backed-securities, combining LoanDynamics, our suite of MacroDynamics models, and our original and highly efficient cash-flow engines.

MacroDynamics includes the following models:

HomePriceDynamics utilizes various models to forecast home prices. Simulating home prices consistent with Office of Federal Housing Enterprise Oversight (OFHEO) historical data, it produces Monte Carlo analyses of projected housing prices forecasted to Metropolitan Statistical Area (MSA) levels. It employs one model using a “geographical localizer” to look at housing price index (HPI) forecasts simulated to MSA levels and uses a second, stochastic home price appreciation (HPA) model, to determine how default options and capital structures should be valued.

InterestRateDynamics combines four models: a choice of three single-factor models or a two-factor model. All models are free of arbitrage with each model instantly calibrated to swaps (a LIBOR swap or Treasury curve) and a matrix of at-the-money swaptions. Each model can be run time-dependent or at constant volatility that is completely controlled by the model.

UnemploymentDynamics will be our latest macro-economic model focused on unemployment statistics and trends. We anticipate launching this model by year-end 2020.