Frequently Asked Questions

TARA™, the Transparent Autonomous Risk Algorithm delivers an unbiased look into collateral-level risk for commercial real estate debt and CMBS. Utilizing machine-learning to distill 30+ years of data and more accurately forecast future performance.

TARA™ is the first AI assisted risk analysis service for CRE, giving you a superior way to instantly assess a portfolio, compare pools, and triage teamwork effort. It integrates with current team workflows, direct data feeds, Excel add-in, data provider integration, or web application.

Ai SPARK spent 5+ years of development with a team of researchers, programmers, data scientists, modelers, and investment professionals to create TARA™. Out of the box, TARA™ brings your team the power of Artificial Intelligence.

Keep using your model. TARA™ is different, being the first AI assisted analyst helper that supports CRE CLO’s, SASB, and conduit. TARA™ is inherently more dynamic and is used alongside or in conjunction to regular credit risk models.

First and foremost, TARA™ is different than a “model”. Models tend to behave linearly when reacting to user inputs and are backwards looking. TARA™ is more like an analyst in that how it reacts to user inputs will be reliant on the particulars of a collateral loan. It is as if TARA™ provides an individualized “model” for each collateral loan that it looks at and predicts future performance.

Second, a model is fixed, having been built upon the statistical relationships/regressions that were observed in the data. TARA™ continuously brings in new information and learns from new remittance.

Yes and no. Like almost all AI under the very strictest sense of the word, it is not. Ai SPARK chooses which data to expose to TARA™ in the data procurement process. Ai SPARK sets the learning architecture of TARA™ at the very beginning of a major version release, thereby addressing issues noticed in one or more of the testing phases, but that is where the line is drawn.

Ai SPARK never tells TARA™ what to think of a particular field, never injects its bias into the opinion formulation process, and limits itself strictly to creating, improving, or procuring the data fields that TARA™ looks at, after which TARA™ learns autonomously of any intervention.

TARA™ is a deep learning neural network using mathematics and algorithms from the fields of Artificial Intelligence and Machine Learning. Ai SPARK designs the learning architecture that TARA™ uses to learn and formulate itself.

TARA™ incorporates 1,000’s of data points broken into several categories; submarket, economic, geographic, property, performance, status, and loan. From there TARA™ learns which data most frequently contributes to historical loan losses and uses that information to predict future performance.

TARA™ handles the heavy computational work and frees people to perform the deeper dives into collateral, and gather knowledge which can then be incorporated and combined with the power of TARA™. Under all economic conditions, use TARA™ to optimize portfolios or opportunities in the market.

TARA™ is used by CRE professionals at leading due diligence providers, B-piece buyers, investment grade buyers, researchers, insurance companies, and top-tier asset managers.

For details on the stress scenarios that we apply "out of the box", please contact us.

TARA™ has experienced the dotcom bubble, 9-11, the Great Financial Crisis, a market liquidity crisis in the wake of the Great Financial Crisis, and the Coronavirus. These events keep building its long-term memory. We therefore do not explicitly treat or adjust TARA™ for business cycles outside of the creation of “SPARKS”; indices that we create to address particular shortcomings in the dataset (e.g., the fact that CMBS history has largely been in a declining interest rate environment).

Ai SPARK provides the actual results alongside TARA™’s predictions, for the entirety of CMBS history. Thus, enabling a user to view test performance for themselves under any and all cuts of the data.

Unlike man-made credit risk models the key drivers of TARA™ results will change for each collateral loan. The inputs that drive TARA™ can be viewed using the "Illumination" feature.

It means that TARA™ has no expectation that the collateral loan will take a significant loss.

Yes. Loss expectation within the next 12 months, next 36 months, and lifetime of the collateral are all provided in the Bond Cash Flow section of the tara_sheet, in addition to being evident in the CDR’s in INTEX calc, and the INTEX Global Asset Level outputs produced by TARA™.In version 3.0+, loss timing is expressed (in months) for each distressed collateral loan.

Any fields within the TARA™ tara_sheet are available to change and adjust. This is where users apply their overrides and “Ask TARA” for updated results. The "Illumination" feature will graphically show which fields are most impactful for each collateral loan.

TARA™ provides a loss estimate for each collateral loan (in addition to CDR vectors which can be used in a cash flow engine). TARA™ also provides a letter rating grade tied to the loss estimate. The track record or prevailing performance of each letter grade is viewable in the ‘committee docs’ report. TARA™ ranks the TARA™ sheet by loans which are most impactful to any specific portfolio – providing a priority list for deep dives on loan collateral. TARA™ shows the user which fields are driving its ratings. This allows users to change those field values and ask TARA™ to update collateral results considering individual qualitative input.

As a deal seasons, TARA™ will react and provide end to end loss guidance from new issue to surveillance. The more track records a loan has as it moves through its life, the stronger TARA™ will be in its opinion.

The principles, testers, and advisors behind the AI have over 90 years’ experience in the CRE industry and some of the best track records in model creation through the Great Financial Crisis and subsequent recovery. TARA™ was built to provide CRE participants with an unbiased view into collateral credit risk, with humility and experience to know its limitations—to know where to hand off to users—to provide the most accurate loss guidance.

The technology excels at compiling various data sources into a cohesive loss guidance. TARA™ is able to continuously reinvent itself in an effort to provide the most accurate guidance possible for CRE debt. Even if you have the same data, TARA™ offers the best way of pulling it all together into a single answer for each collateral loan.

No. TARA™ can be used without an INTEX subscription through Ai SPARK.