STRAAS Vault Guard

Features Engineering

Configure the core logic of the different risk factors of the features engineering engine.

Sum of weights must equal 100%. Current total: %

An error occurred while saving the configuration

Configuration saved successfully

Factory Reset Confirmation

This will revert all risk factor weights, and also the features configurations to the default suggested values. This action cannot be undone.

New weights will be:

Channel Risk: 25%
Geolocation Risk: 20%
Velocity Risk: 15%
Device Risk: 15%
Merchant Risk: 10%
Amount Risk: 8%
Time Risk: 7%

Amount Risk

Detects the risk probability of the transaction amount.

Update Weight

8.0%

Velocity Risk

Detects the risk probability of the transaction frequency in one hour duration.

Update Weight

15.0%

Time Risk

Flag transactions happening at unusual hours of the day.

Update Weight

7.0%

Geolocation Risk

Detects the risk probability of both issuer and acquirer locations.

Update Weight

20.0%

Merchant Risk

Evaluate risk based on the type of the merchant and its reputation.

Update Weight

10.0%

Channel Risk

Detects the risk probability based on the type of transaction.

Update Weight

25.0%

Device Risk

Evaluates the risk probability of the device used for the transaction.

Update Weight

15.0%

Velocity Risk Engine

Transaction Frequency Limits

Set the maximum allowed transactions per hour before flagging.

transactions/hour

Max Frequency

Logic Explanation

This engine analyzes the transactions frequency during the latest hour to determine probability of fraud.

Changes apply immediately to new transactions.

Time Risk Engine

Time-of-Day Risk Probabilities

Adjust risk levels for each specific time window. Higher values mean more risk.

Deep Night (00:00 - 05:00) 80%

Early Morning (05:00 - 08:00) 60%

Business Hours (08:00 - 20:00) 15%

Late Evening (20:00 - 24:00) 40%

Logic Explanation

This engine analyzes the time of day when transactions occur to assess fraud probability. Transactions during unusual hours are flagged with higher risk scores.

Changes apply immediately to new transactions.

Geo Risk Engine

Max Acceptable Distance (km)

Set the maximum distance between issuer and acquirer locations before flagging as suspicious.

Max Acceptable Distance

5000 km

Logic Explanation

This engine evaluates the geographical distance between the issuer and acquirer locations to detect potential fraud.

Transactions exceeding the maximum acceptable distance are flagged with higher risk scores. Changes apply immediately to new transactions.

Merchant Risk Engine

MCC Risk Categories

Assign base risk probabilities to merchant categories.

Essentials & Public Services 10%

Retail & Consumer Goods 50%

Travel & Hospitality 50%

Professional Services 50%

High-Risk & Financial 90%

Logic Explanation

This engine evaluates transactions based on the merchant category, assigning different risk probabilities to different types of businesses.

Higher risk categories like financial services are assigned higher base probabilities, while essential services have lower risk scores. Changes apply immediately to new transactions.

Channel Risk Engine

Channel Risk Probabilities

Base risk assigned to each transaction entry point.

E-Commerce 20%

Point of Sale 20%

ATM Withdrawal 20%

Wire Transfer 40%

3DS Missing Penalty

Additional risk added if 3DS authentication is missing

Penalty Score 70%

Logic Explanation

This engine evaluates transactions based on the channel through which they are initiated, assigning different base risk probabilities to each entry point.

Channels with less security (like wire transfers) are assigned higher base risks. Additionally, transactions without 3DS authentication receive an extra penalty score. Changes apply immediately to new transactions.

Device Risk Engine

Device Age Risk Probabilities

Assign risk probabilities based on device age.

New Device (less than 7 days) 25%

Young Device (7 to 30 days) 15%

Old Device (more than 30 days) 10%

VPN Detected Penalty

Probability added when IP is masked with VPN

Penalty Score 30%

Logic Explanation

This engine evaluates transactions based on the device characteristics and network signals. Newer devices may indicate account takeover attempts, while older devices suggest established user patterns.

Additionally, VPN detection adds a penalty score as masked IP addresses are often used to hide fraudulent activity. Changes apply immediately to new transactions.

Amount Risk Engine

Balance Depletion Curve

Define how much of an account balance drain is considered risky.

The risk curve is determined by statistical analysis of depletion patterns.

Logic Explanation

This engine evaluates transactions based on the amount relative to the account balance. The risk increases as the transaction amount approaches or exceeds the available balance.

The formula calculates risk by considering both the transaction amount relative to the old balance and the proportion of balance remaining after the transaction. Higher depletion rates result in higher risk scores. Changes apply immediately to new transactions.