Shield score

The shield score is a core metric used to quantify the potential fraud risk of a transaction. The score is generated by Antom based on a general machine learning model and is applicable across most industries and transaction scenarios. If you are using the Antom Shield Premium version, a customized model will be activated to further enhance scoring accuracy.

The shield score is calculated in real time by integrating data from multiple dimensions, primarily including the following factors:

• Transaction amount and frequency
• Payment method and historical transaction records
• Global risk intelligence network information
• Buyer behavior patterns
• Device fingerprint and geographic location

The accuracy of the shield score depends on the completeness and quality of payment data. Missing key fields, irregular formatting, or data errors can affect the model’s ability to accurately assess transaction behavior and may lead to biased scoring results. Please ensure that all necessary information is provided to improve the effectiveness of risk assessment.

For the data fields required by Antom Shield, please refer to Enhance risk assessment quality. For detailed field definitions, see the Antom API documentation.

Shield score reference

Currently, the Antom card payments merchant accounts have the risk scoring feature enabled by default. You can log in to the Antom Dashboard, navigate to Shield > Risk insight to view detailed shield scores for each transaction. Every transaction receives a shield score ranging from 1 to 100, which quantifies the potential fraud risk level. According to default settings, transactions with a score of 85 or above are classified as high-risk by the model and should be given special attention.

Notes:

• The shield score is generated by a machine learning model and is intended to serve as a reference indicator, not as a definitive measure of actual risk. We recommend that you consider your business context and other relevant information to make a comprehensive assessment when using the shield score.
• Due to data security requirements in certain regions, shield scores for some transactions may not be fully displayed. We are continuously improving data coverage, and you are encouraged to use additional assessment dimensions and supplementary information to enhance the accuracy of risk evaluation.

Shield score settings

Antom Shield uses an intelligent risk scoring model to intercept suspicious transactions in real time while minimizing interference with legitimate payments. The following configuration modes are supported to meet different business needs:

Default setting

The default threshold for shield score interception is 85 points, meaning the system will automatically block transactions with a shield score higher than 85. This default value balances security and transaction success rate, making it suitable for most standard business scenarios. If a transaction blocked due to a shield score is confirmed to have been initiated by a legitimate buyer, you can use the allowlist management feature to add the corresponding buyer or payment instrument to the allowlist.

Notes:

• Adding an object to the allowlist will not reprocess or reinitiate a transaction that has already been blocked. This operation prevents the system from intercepting future transactions from the same buyer or payment instrument.
• If high-risk transaction-related entities are mistakenly added to the allowlist, subsequent transactions may lead to potential financial losses, please proceed with caution when managing your allowlist.

Custom configuration Antom Shield Pro

If you are using the Antom Shield Pro version, you can quickly and flexibly adjust the shield score interception threshold based on your business characteristics. The configurable range is 70 to 90 points. When adjusting the threshold, the system will display the following real-time prediction metrics to help you make informed decisions:

• Estimated decline rate: The projected proportion of transactions that will be intercepted under the new threshold
• Estimated number of intercepted Transactions: The projected number of transactions that would be declined
• Estimated amount of intercepted Transactions: The projected total value of intercepted transactions

In addition, the shield score is included as a risk variable in the rule configuration to optimize risk management and control processes. For example, when the shield score exceeds a certain threshold, specific actions can be automatically triggered—such as initiating 3D Secure (3DS) verification or rejecting the transaction. For details on implementation and configuration procedures, please refer to Rule.

You can access the shield score settings page in two ways:

• Direct configuration path: Navigate to Shield > Smart engine > Score, then click Edit score in the upper right corner. Adjust the threshold using a slider or by manually entering a value.
• Rule configuration path: The shield score interception logic operates as a rejection rule. Go to Smart engine > Rules, locate the Rejection rules, and edit or update it as needed.

Best practices for shield score adjustment

By setting appropriate shield score thresholds, you can strike a balance between improving transaction approval rates and strengthening security protection. The table below illustrates the specific impacts of different adjustment approaches for your reference: