Overview

Begini's scoring models transform the behavioural data captured during the assessment into a structured risk output. The model evaluates patterns across the entire session rather than scoring individual responses in isolation.

What the model analyses

The model is built on multiple layers of behavioural signal captured throughout the assessment:

• Decision-making patterns and risk-taking behaviour
• Response timing and variability across tasks
• Consistency of behaviour under different conditions
• Engagement and completion behaviour

These signals are combined to reduce noise from individual actions and produce a stable, comparable output.

The output: relative risk score

The score represents relative risk — it is designed to rank users within a population rather than produce a standalone absolute value. This means:

• Lower scores indicate more favourable behavioural patterns
• Higher scores indicate higher-risk signals
• The score is most meaningful when used to segment and compare applicants within your own portfolio

Risk bands

Users are placed into risk bands based on their score. Bands are designed to maintain clear separation between risk levels, so that lower-band groups consistently outperform higher-band groups on real-world outcomes such as repayment behaviour.

Calibration

Begini's models are calibrated against real-world portfolio performance data and refined over time. The scoring is designed to remain predictive as it scales across different markets and applicant populations.

Trait-level outputs

In addition to the overall score, the model produces trait-level outputs — component scores that indicate performance across different behavioural dimensions. These are included in the webhook payload and help explain why a user received a particular overall score.

What the model does not rely on

The scoring model does not rely on self-reported answers, single task outcomes, or demographic data. It focuses on observed behaviour across the full assessment journey, which supports both consistency and fairness.

Using scores in practice

Begini scores are designed to complement your existing decisioning framework, not replace it. Combine the risk score and trust indicators with your internal credit data and risk models for the best outcomes. See Understanding Assessment Results for how to interpret and apply the outputs.

Next steps

• Understanding Assessment Results — interpret and apply the risk score and trust indicators
• Fraud Detection & Trust Score — how session reliability is evaluated