Filter Assessment

Typical third-party filter assessments are notorious for their vagueness and superficiality. The results of the tested characteristics of the filter are presented as numbers, typically a percentage hit rate for a particular type of test, then compared with “peer” averages.

This is flawed. Filter behaviour is highly sensitive to the content – you expect the filter to respond differently to deleting a character from a very short name versus a long name with many words.

Our approach is different. The tests we use are specifically designed to take into account these variations in content and provide a much more detailed understanding of filter’s response. We provide clear and actionable conclusions enabling our customers to be aware of potential weaknesses and leverage strengths to improve efficiency.

Outcomes

• Comprehensive test report with detailed analysis and recommendations
• Test suite for future use

Optional packages

Business Rule / whitelist validation Review of false positive reduction rules to determine potential risk of masking and effectiveness

Enhanced efficiency analysis Focussing on patterns found in the “big hitters” from representative good guy data and sanctions entry hotspots

Policy alignment Using representative test cases to assess the extent to which the filter behaviour is aligned with policy

Get in touch via contact@deep-lake.comfor more information.

ISO 20022 Readiness

Following the recent European Central Bank announcement that it would delay the ISO 20022 migration of the Eurosystem to March 2023, now is an ideal time to look at how best to leverage this change to further their screening environments to identify the impact of this upcoming change. 

ISO 20022 messages provide additional, more structured and granular information than their FIN equivalents which creates new potential risks and opportunities for screening. Primary risks include failing to screen some fields, FIN translation issues, failure of business rules and high false positive rates. However, there are also opportunities to leverage the more structured format to high optimise the screening and improve efficiency. 

Outcomes

• Report including comparison performance of FIN vs ISO 20022 messages for both effectiveness and efficiency, covering structured and unstructured formats
• Re-usable test suite in FIN/ISO 20022 format

Optional packages

Guiding principles check Comparison of screening policy against SWIFT’s ‘Guiding principles for screening ISO 20022 payments’ (endorsed by The Wolfsberg Group and The Payment Market Practice Group.) 

Business rule validation Review and categorisation of existing business rules to support the development of a migration strategy 

Get in touch via contact@deep-lake.com for more information. 

Filter Tuning

There are two aspects of sanctions filters that are important when tuning: effectiveness and efficiency. Effectiveness is the filter’s ability to hit the right information in the transaction data, efficiency is filter’s ability to miss the wrong information. These are often treated as two independent features that need to be optimised, but this is a fallacy – they are two sides of the same coin. 

The usual process for filter tuning is to first verify that the risk is controlled (effectiveness) and then try to reduce the false positive rate (efficiency). The risk in adopting a sequential approach is that it fails to optimise the quality of the results. There is limited ability to control this at the “backend”. It is typically done by adjusting threshold (reducing volume but not improving quality) and/or implementing false positive reduction techniques (whitelists, rules etc. to supress certain patterns of alert). 

There is a better way. The biggest gain in efficiency will always come from eliminating screening scenarios rather than mitigating them. The common practice (in line with OFAC guidance) of not screening against weak aliases is the first step in this process. Much more can be done by further eliminating scenarios that could not realistically create useful alerts. 

Outcomes

• Screening scenario review – optimise screening strategy by eliminating unnecessary screening combinations
• Filter behavioural review – clearly define which filter behaviours are within your risk appetite
• Big hitter analysis – identification of false positive scenarios that have disproportionate impact on the volume of alerts
• Sanctions list hotspot analysis – identification of the sanctions list entries that have disproportionate impact on the volume of alerts

Optional packages

Policy alignment Using representative test cases to assess the extent to which the filter behaviour is aligned with policy 

Threshold response analysis Zero in on best filter settings by using statistical analysis of impact of parameter changes 

Get in touch via contact@deep-lake.com for more information. 

Test Suite Development

Continuous testing is the key to avoiding nasty surprises. The risks of untested screening environments grow over time. Changes to lists can result in new entries causing significant false positives or even worse – they can reveal previously unknown weaknesses in your screening setup. The addition of business rules or whitelist entries can also result in unintended masking of desired hits. 

Running testing on a regular basis eliminates these risks and can help you avoid potentially costly and time-consuming work when you do seek external validation.  

We work closely with our customers to build a tailored test suite that is aligned to their risk appetite. The goal is to equip you with something that is accurate, actionable and above all, simple to use. 

Outcomes

• Test Suite: Fully documented set of tests available in multiple file formats (e.g. FIN, ISO 20022) covering key aspects of screening behaviour and aligned to risk appetite 

Optional packages

Policy alignment Using representative test cases to assess the extent to which the filter behaviour is aligned with policy

Get in touch via contact@deep-lake.com for more information. 

Testing Best Practices Training

Testing sanctions screening environments is very often something that is only undertaken at critical points in time – usually as part of an annual independent assurance process or maybe in response to a new release of the filter.

Whilst useful and important, these tests are typically fairly non-specific and really only verify the screening environment at a superficial level.

Whether you are using existing tools or building your own test approach, we can help you build a best-in-class testing approach to harness the real power of testing in support of your business challenges.

Course modules

• Testing overview
• Exact testing
• Testing non-name attributes
• Fuzzy testing
• Results interpretation
• Results analysis
• Patterns of hits and misses

Additional modules

• Efficiency testing 
• Coverage testing 
• Testing name and addresses 
• PEP testing 
• Follow-up and drill-down testing 
• Threshold testing and tuning 

Get in touch via contact@deep-lake.com for more information. 

Policy Alignment

As part of OFAC’s FAQ Risk Matrix (taken from the 2014 FFIEC Bank Secrecy Act/Anti-Money Laundering Examination Manual), several red flags (high risk factors) are related to understanding of the screening environment and having a suitable policy to cover it.

Understanding how filter behaviour relates to regulatory obligations can be a complex problem. There is no one-size-fits-all approach to what and how you screen for sanctions violations. It depends on many factors including geo-presence, lines of business and risk appetite.

We can help simplify this process by providing a comprehensive framework as a start point and a clear and simple methodology. By conducting workshops and interviews with key personnel we make the process of describing how your policy is implemented straightforward.

Outcomes

• Documented policy covering list selection, screening matrix, filter behavioural traits and suppression strategy 

Optional packages

Test Suite Fully documented set of tests available in multiple file formats (e.g. FIN, ISO 20022) covering key aspects of screening behaviour and aligned to risk appetite

Get in touch via contact@deep-lake.com for more information.