In November, we launched tools that give you more control and transparency, from catching discrepancies between what applicants say and do to choosing exactly which accounts power your models while staying compliant.
Detect Differences in Self-Reported and Bank Data with New Attributes
Catch discrepancies between what applicants say and what they do with a new category of risk signals that cross-reference self-reported data against actual bank behavior.
- We've launched 54 new attributes that detect a gap between applicant claims and financial reality
- Example: Instantly detect when someone reports $100,000 income but bank data shows only $70,000
- Simply send the self-reported data from your application process (such as income, rent, employment), and we'll automatically compute these discrepancy signals
- Read about how to integrate self-reported data here, and view the new attributes here.
Control Compliance & Optimize Models with Attributes Report API
Stay compliant and boost model performance by controlling exactly which accounts and attributes power your underwriting decisions.
- Specify which borrower accounts to use via Account Presets, ensuring you only leverage data you have permission to underwrite against
- Compare attribute performance across different account combinations to identify new sources of model lift
- Select only the specific attributes you need for your underwriting process
- View the documentation here
Identify Institutions in Transactions
Unlock institution-level risk signals by knowing which bank, or credit union each transaction and balance comes from.
- We now capture and store the financial institution identifier for transactions and balance data
- Build risk signals based on institution type
- Identify patterns associated with specific financial institutions
New Attributes: Raw Balance Attributes
Fifty-four new raw balance attributes provide transparency into balance trends and signals for fraud and risk models.
- Balances are calculated using only actual uploaded data. For example, missing days remain NULL instead of being estimated by our Balance End-Of-Day Model
- See exactly how many days of real balance data exist (e.g., days with double-digit balances, days with negative balances) to assess data completeness
- View the new attributes here
.avif)
