|
The latest version of RiskModel—available in new desktop and API versions—significantly advances the power, flexibility, and versatility of predictive analytics for loans and loan portfolios. Drawing on databases tracking 50-million mortgages and over $2 trillion in non-agency MBS and ABS, RiskModel uses Monte Carlo methodology to simulate future marketplace scenarios and predict a range of likely default, prepayment, or other outcomes.
On top of a unique ability to model prepayment/default risk, borrower behavior, delinquencies, interest rates, and housing prices simultaneously—the only way to model actual reality—the new RiskModel includes:
The new RiskModel incorporates true marketplace realities—even uncomfortable ones—to help originators, insurers, dealers, servicers, and investors build trustworthy foundations for future credit policies, loan loss reserves, bond pricing, marketplace analysis, and portfolio risk management.
| |
 |
|
Click to Enlarge |
| |
| |
Starting with OLEDB or ODBC-compatible data files*, RiskModel simulates future market scenarios and
creates reports that can either stand alone or power subsequent business applications.
*OLEDB = Object Linking and Embedding Database, ODBC-compatible = Open Database Connectivity |
|
|
|
| |
| |
|
|
|
|
|
| Setting loss/capital reserves |
 |
|
|
|
|
| Lowering costs |
|
|
|
|
|
Decreasing securitizationcoverage
in rating agency negotiations |
|
|
|
|
|
| Grooming portfolios (buy/sell decisions) |
|
|
|
|
|
| Strategic planning |
|
|
|
|
|
| Product evaluation and development |
|
|
|
|
|
| Pricing securities, B-pieces |
|
|
|
|
|
| Pricing MSRs, insurance, whole loans |
|
|
|
|
|
| Risk factor measurements |
|
|
|
|
|
Decreasing guaranty fees in
negotiations with GSEs |
|
|
|
|
|
|
|
 
A primary objective in designing the new RiskModel was to make it seamlessly integrate with client in-house systems—to extend their software capabilities to include automated predictive analytics. Since this required reengineering RiskModel from top-to-bottom, we also extended and streamlined its core analytics functionality.
RiskModel analytics enhancements:
- New subprime statistical model includes data through Q1 2006, improved prepayment/defaults controls and characteristics, refined parameters for ARMs, improved prepayment penalty impact measures, better prepayment and default estimates correlation with HPA, increased risk property types.
- New loss severity equations feature recalculated severities for various liquidations, ARM loans, investor properties, multi-family dwellings, loan type segments.
- New HPI simulation model offers historically accurate boom and bust cycles, accurate long-term HPI volatility for better default/loss distributions, longer economic capital horizons, historically linked returns in location, linked returns in other locations, Core Business Statistical Area (CBSA) definitions.
In addition to reengineered functionality, RiskModel simulations now reflect recent turmoil in the mortgage market accurately. For example, estimated subprime prepayment speeds factor in observed prepayment increases during 2003-2005, subprime loan characteristics include more precise and telling detail, transition equations mirror recent factual data, payment history variables offer greater insight and control.
RiskModel’s rich library of prime, alt-A, and nonprime statistical models has grown to include new subprime, Loss Given Default (LGD), and Housing Price Index (HPI) models that are specifically designed to cope with the current mortgage industry crisis. Sample changes:
- New subprime model includes ARM rate reset-specific parameters, redefined prepayment and default estimates, more recent data affecting default.
- New LGD model includes second lien statistical loss severity, redefined severity equations for first lien liquidations, additional 150,000 recent loss records.
- Enhanced Prime option ARM transition model affects prime option ARMs with 18-month or less resets, improving credit risk projections for this class of loans.
- New HPI simulation model includes general time dependence in returns, return as function of previous returns, contemporary dependence in returns across locations.
Extensive back-testing documenting the substantial performance improvements achieved using the new subprime, LGD, and HPI simulation models is available for review. Please contact the LoanPerformance representative for your geographic area.
The new RiskModel data conversion wizard allows you to run any OLEDB (Object Linking and Embedding Database) or ODBC (Open Database Connectivity)-compatible database file through RiskModel simulations, then reverse the process and output data in any desired reporting format. Other new features include:
- Data viewer added that allows review of source data prior to conversion.
- Data analyzer matches specified source data with an existing template.
- Column mapping enables advanced formatting to the source data.
- Data translation allows non-resident source data to be converted.
- QA checks/messages define errant data and actions to correct.
Conversion processing also allows conditional mapping as part of the user-defined QA checks (for example, “delay mapping until the QA checks are executed and use conditional parameters to map source data to RiskModel input field”).
The RiskModel Desktop 4.0 graphical user interface (GUI) has been completely reengineered to extend and simplify user control over its full range of sophisticated predictive analytics. Some of the more significant changes:
- Setup supports multiple open projects, simplified setup process, easy error correction, new copy-and-paste capability, new data explorer function.
- Simulation controls simplified and clarified, multiple jobs possible in one project, online report access, maximum error specification capability.
- Reports can be grouped by portfolio or loan level, over-time reports permit sub-reporting, customizeable output processing for all reporting types.
- Data inputs and handling functions expanded and clarified, conversion from/to any OLEDB-compliant format, supports single and multiple input conversions.
- HPI expanded, simplified controls of housing price inputs, geographical areas defined by CBSA codes, regional volatility parameters, vector selection added.
- Segmentation now separate utility giving more control and allowing multiple conditions for segment, ability to generate specified-segment outputs.
This sampling of updated GUI-controlled RiskModel Desktop 4.0 capabilities does not capture the full effect of its much-enhanced usability. The new GUI is not a surface upgrade—it represents extensive user testing and a commitment to reengineer everything necessary.
RiskModel Enterprise 2.0 is an API engine that can be embedded in mortgage processing and trading systems to analyze huge volumes of securities and bonds, evaluate them in light of the latest housing price, interest rate, default and loss severity rate fluctuations, then produce realistic, accurate results in time to capitalize on immediate trading opportunities.
RiskModel Enterprise users enjoy significant benefits:
- Faster decisions RiskModel Enterprise API enables the inclusion of sophisticated risk analytics in evaluating thousands of securities at automation-level speeds, giving traders invaluable, real-time perspective on the likely accuracy of current bond pricing.
- Better-quality decisions RiskModel Enterprise API improves the quality of the risk analytics process, transforming a multiple-step, manpower-intensive, easily-skewed procedure into a tightly integrated, consistently reliable, easily managed predictive process.
- Consistent application of proprietary view RiskModel Enterprise API can be set to apply unique client perspective on housing prices, interest rates, consumer confidence, other parameters for projecting roll-rates, prepayments, defaults and loss severity.
The RiskModel Enterprise API engine inputs predictive data to enterprise system components like mortgage securities waterfall and option-adjusted spread-pricing applications. RiskModel's performance projections—incorporating client perspective on key industry trends—become unequivocal valuation parameters, significantly improving the trustworthiness of subsequent bond analytics and pricing decisions.
RiskModel can be a challenge to set up and operate. If you lack the resources or time to run RiskModel effectively, there is an alternative—RiskModel Professional Services. Using your data and following your specific parameters, we can provide finished RiskModel analytics that enable you to make better, more profitable decisions.
RiskModel Professional Services benefits include:
- Forecasting accuracy like leading originators, aggregators, servicers, investors
- Third-party opinion of current information influencing your business decisions
- Conservation of insufficient staff and/or machine resources
- Easily customizeable frequency and other parameters to suit your needs.
- Cost-effective shortcut to important answers that you need fast
- Easy-to-budget pricing on by-engagement basis
With a minimal commitment of time and resources, you can receive full-blown RiskModel predictive analytics and incorporate them into your internal decision-making process. By leaving the operational nuts and bolts questions to our experts, you can quickly benefit from RiskModel prepayment, delinquency, default and cash flow projections customized to your own specific requirements.
RiskModel Professional Services is available to both new users and existing licensees.
Identify yourself (name, company, email only) on the accompanying form and download a PDF of the RiskModel product brochure (242KB).
 Technorati  Del.icio.us  Calculated Risk  The Big Picture  Housing Bubble  Seeking Alpha
To learn more about RiskModel, please email us or
give us a call at the appropriate number for your geographic area.
|
|
