Dynamic Model Behavior and You: Unlocking Your Monitoring Potential

April 26, 2022
By 

Subscribe to our Blog!

Please fill out the form below:

What if you and your team could build flexible, customizable models that flag alerts without writing a single line of code? Well, with Unit21’s new Dynamic Model Builder, you can. 

We get that predefined scenarios, typologies, and patterns can only get you so far, especially when you have needs as unique as some of our clients. If you want to iterate and deploy models that catch evolving fraud trends quickly, reduce false positives, flag truly suspicious activity, and overall level up your risk and compliance practices, Unit21 can deliver.

Use Unit21 Templates for Faster Behavioral Monitoring

Building from scratch is a thing of the past with Unit21’s Dynamic Model templates. They help you stay current with the latest trends and are easy to tweak to meet your business’ specific thresholds. Unit21’s platform makes developing new model behaviors faster than ever with easy-to-populate modules that can be tailored to your monitoring needs.

If you don’t see a template that suits you at the moment, don’t worry. We’re working with outside risk and compliance consultants to deliver new, in-demand templates in Q2.

Dynamic Model Builder Example: Transaction Volume Historical Deviation

One of our most popular models is a sample we have set up to monitor Transaction Volume Historical Deviation. We’ll use this as an example to dive deeper into our Dynamic Models.

Variables & Trigger Conditions

The basis of any Dynamic Model starts with Variables and Trigger Conditions

Variables are where you can specify exactly what you want to monitor. For example, the number of BTC transfers in the past day, or the average amount withdrawn in the past month.

Trigger Conditions are what actually cause the rule to generate an alert. This is where the models are super flexible, powerful, and laser-focused. For example, you can flag an entity if the number of BTC transfers in the past day is more than 1 million—or the average amount withdrawn in the past month is more than $50,000.

How does the template work in this example?

Based on the criteria shown, this template will flag potential account deviation for any entities that had an average daily total $ transactions in the past week greater than the average daily total $ transactions in the past month.

With other platforms, you might have to reach out to an internal team to make tweaks to better suit your monitoring needs, or even send a request to a professional services team. With Unit21, you have complete control and are free to make any changes as you see fit.

Why you might make updates to this template

If you’ve seen a lot of false positives with the basic template, you’ll likely need to modify it further. An example is setting a new threshold to reduce the number of false positives. 

For example, we may want to make sure the average volume in the past week is actually at least 2x that average. Making this update is as simple as a few edits to the trigger condition, and boom: you’ve successfully altered the model! 

As shown above, in this example, we’ll multiply the average for the past 31 days by 2 and then make sure that the past week’s daily average transaction volume is greater than that) ( Variable A > 2 * Variable B)

This is just a peek at what kind of no-code customization is possible with Unit21. Comparisons and behavior models can get infinitely more specific and complex to give you precisely what you need. For instance, you can add more variables, look at various time ranges, and add a few more trigger conditions to fine-tune what you’re looking for. 

Some common things we see customers doing with Dynamic Model Behavior: 

  • Comparing an individual’s past transaction volumes with current transaction volumes to account for stolen cards or accounts
  • Looking at sign-up bonuses for accounts that are then followed by a long period of dormancy to prevent customers from taking advantage of bonuses 
  • Comparing custom data fields on transactions and events to add additional data signals (e.g., comparing the number of transactions that had a high-risk value vs. a medium risk value).

To learn more about working with variables and triggers and customizing Unit21 for your specific needs, contact your Customer Success Manager or get in touch with us

Download Transaction Monitoring Product Guide

By: Hana Bendy, Product Manager & David Cooper, Tech Lead

Cooper is a Tech Lead on the Detection Modeling team at Unit21. He values himself by enabling others to do their best work. He is also easily distracted by dogs and mountains.
Hana is a Product Manager on the Detection Modeling team at Unit21. She finds joy in working with customers and hearing their perspectives. She also enjoys traveling to remote beaches, playing tennis, and taking dance classes.

Unit21 is dedicated to helping our customers empower their teams to make data-driven decisions in the fight against financial crime. Discover why customers switch to Unit21.

Getting started is easy

See first-hand how Unit21
can help bolster your risk & compliance operations
SUBSCRIBE

Subscribe to our Blog!

Please fill out the form below: