Real-time risk scoring
Risk scoring condenses the risk of a customer or transaction into a single measurable value to streamline decision-making. TruvaLI maintains this score not as a static label, but as a live value recalculated with every new event.
Why scores cannot be static
In many institutions, customer risk is determined during onboarding and reviewed annually. However, risk changes alongside customer behavior. A customer who appears low-risk at onboarding might exhibit a completely different profile six months later.
TruvaLI updates the score on an event basis. When a new transaction, a new device, a profile change, or a screening match occurs, the score is recalculated based on the current reality.
What is scored?
- Customer score: Onboarding data, screening results, behavioral history, and corporate ownership structure for business clients.
- Transaction score: Amount, time, channel, counterparty, and comparison against the customer's own baseline.
- Counterparty score: Recipient history, relationship network, and blacklist status.
These three scores feed into each other: a transaction with a high-risk counterparty affects both the transaction and customer scores.
Blending financial and non-financial signals
The distinguishing feature of the score is that it does not look at money movement alone. The company's activity code, trade name, incorporation year, transaction time, and device history also enter the equation.
This helps capture cases that do not trigger amount thresholds: a POS transaction occurring at 03:00, even if the amount is low, increases the score because it contradicts industry expectations.
What is the score used for?
- Decision thresholds: Transactions above a certain score are rejected, flagged for monitoring, or routed for additional verification.
- Customer onboarding: A high score triggers escalation to a video call.
- Investigation priority: The case queue is sorted by score, allocating limited time to the riskiest cases.
- Periodic review: Customers requiring enhanced due diligence are identified based on their scores.
Institutions set the weights
The weight of each signal is determined by the institution's risk policy: this is not a change that needs to be requested from a software team. Weight and threshold changes can also be tested on historical data before going live, allowing you to see in advance how much a threshold change will expand the investigation queue.
What it delivers to the institution
- A real-time, zero-latency assessment that keeps risk updated with every event.
- An explainable record of which signals make up the score.
- A common metric for onboarding, transaction, and investigation decisions.
- A demonstrable rating system expected by the risk-based approach.
Common questions
- How often is the risk score updated?
- It is updated on an event basis, not periodically. When a new transaction, device change, profile update, or screening match occurs, the score is recalculated instantly.
- Can we show auditors how the score is calculated?
- Yes. The signals contributing to the score and their weights are recorded: the reason behind a decision can be traced directly within the case. This is crucial for documenting your risk-based approach.
- Can we customize the scoring model ourselves?
- Yes. Signal weights and decision thresholds are adjusted according to the institution's risk policy, and changes can be tested on historical data before going live.