Trade Shield Models- Payment Risk and Default Risk -How our predictive models work

Created by Amy Sara Price, Modified on Mon, 11 May at 12:54 PM by Amy Sara Price

What these models do

Trade Shield uses two predictive models to assess every customer: Payment Risk and Default Risk. These models don’t just collect data — they weigh multiple signals against each other to build an accurate picture of how a customer is likely to behave.

The result is a risk band for each measure, which then feeds into the Risk Adjustment Framework (RAF) to determine an actual credit limit recommendation.

Payment Risk

Definition

The likelihood that a customer will pay you late or miss a scheduled payment. It looks at how the customer trades with you now — their behaviour, consistency, and ability to pay within your current terms.

What the model considers

Age of the company
Number of directors, their age, and tenure
Location and industry of the business
Economic conditions
Number of judgements against the business
How the customer pays you and other suppliers
Historic ageing data and payment patterns
Statutory data and any negative credit events
Recency of default events
Size of the credit history

This measure is also used to calculate the Potential Credit Capacity (before risk adjustment) — the maximum amount a customer could reasonably spend with you based on their payment behaviour and available capacity across other suppliers.

Payment Risk Bands

Band	What it means
Low	Consistently pays on time
Reduced	Small chance of late payments
Medium	Some late payment history
Elevated	Frequent late payments
High	High chance of late payments
Very High	Persistent late payments and serious concerns

Default Risk

Definition

The likelihood that a customer will fall into protracted or full default on a future payment.

Default Risk considers the same model inputs as Payment Risk, with one key difference — it focuses specifically on how the customer treats their obligations with all suppliers, not just you. This replaces the need for traditional trade references by providing a data-driven view of the customer’s behaviour across their entire credit footprint.

Default Risk Bands

Band	What it means
Low	Very unlikely to default
Reduced	Low chance of default
Medium	Some indicators of default risk
Elevated	Notable warning signs
High	Strong risk indicators
Very High	High probability of default

Why these signals matter — how the model thinks

The model doesn’t just collect data points — it weighs them against each other. Here’s the logic behind the key signals:

Age of the business

Newer businesses are statistically more likely to close — not because they’re badly run, but because the early years are hard. Undercapitalisation, market uncertainty, and cash flow pressure all play a role. The longer a business has been trading, the more evidence there is that it can survive. Age equals stability.

Number of directors, their age, and tenure

A business with one director is inherently more exposed — if that person is unable to run the business, the business may not survive. The age of directors matters too: very young directors may lack experience; very senior directors raise succession questions. Tenure tells us about stability — frequent director changes can signal instability, even if the business looks fine on paper.

How they pay you and other suppliers

This is arguably the most important signal. A business can look perfect on paper — right age, right directors, no judgements — but if they consistently pay their suppliers late, that behaviour tells you more than anything else. The model weights actual payment behaviour heavily because it reflects what’s really happening in the business, not just what looks good in a credit report.

Judgements, statutory data, and negative credit events

These are hard flags. A judgement means a supplier or creditor had to take legal action to get paid. The model looks at how many there are, how recent they are, and the pattern over time.

Recency matters

A default five years ago with clean behaviour since is very different from a default six months ago. The model considers not just whether something happened, but when — because recent events are stronger predictors of future behaviour.

Credit Strategy Guide

Once both scores are produced, the Credit Strategy Matrix tells you how to respond — whether to Grow, Maintain, or Reduce your exposure to that customer.

Payment ↓ / Default →	Low	Reduced	Medium	Elevated	High	Very High
Low	Grow	Grow	Grow	Grow	Maintain	Reduce
Reduced	Grow	Grow	Grow	Grow	Maintain	Reduce
Medium	Grow	Grow	Grow	Maintain	Maintain	Reduce
Elevated	Grow	Grow	Maintain	Maintain	Reduce	Reduce
High	Maintain	Maintain	Maintain	Reduce	Reduce	Reduce
Very High	Maintain	Maintain	Reduce	Reduce	Reduce	Reduce

How the risk score becomes a credit limit

Once the model has produced a Payment Risk and Default Risk band, those scores are applied to the Risk Adjustment Framework (RAM).

The RAM takes both scores and translates them into a percentage of the customer’s requested credit limit that you should actually offer. This is how a risk band becomes a rand amount.

How it works

Every customer starts with an asking amount — how much credit they want from you. The RAM looks at where they sit in the risk matrix and applies a percentage adjustment to that asking amount:

Low-risk customers receive a higher percentage of their asking amount
High-risk customers receive a lower — or even negative — adjustment

The order always matters: First the model → then the RAM. The model tells you what the risk is. The RAM tells you what to do about it.

Configuring the RAM by customer type

The RAM can be configured differently depending on your credit strategy. For example, if you are on a drive to acquire new customers, you can apply a more aggressive framework for new buyers while keeping a more conservative one for existing accounts. Two separate frameworks can run simultaneously.