Baseball & Glove on Baseball Field
Spring training for Major League Baseball is long over, the regular season is winding down, and playoffs will soon be upon us. Over the course of a long season, the teams that spent their time during spring training actually training and preparing frequently find that it makes the difference between being a playoff contender and an also-ran.
In a similar fashion, efforts to thoroughly validate an anti-money laundering system can pay off in the long run.
How do you achieve a strong AML system? Here are four areas to work on to get you closer to post-season form.

Have a personalized game plan

What kind of manager would try to run his team like any other team, without regard for the specific strengths and weaknesses of each player? A fired manager, that’s who. Every single baseball team requires its own, unique approach. The players learn to work with each other, play to their strengths, and compensate for their vulnerabilities. The same goes for AML monitoring.
Too often financial institutions try to go with the vague, generic monitoring parameters that came with the system instead of taking the time to tune it to the institution’s unique risk factors. There is no single list of rules and parameters that will work for everybody. It is important to take a look at the types of activities that merit attention in your area, and whether those types of activities are being addressed by your AML system. But if you skipped spring training, it’s not too late. Now might be a good time to put your AML system through a training session to work out anything it might be missing.

Work on your batting average

A rule that generates a significant number of alerts and that rarely result in a SAR is the equivalent of a batter who swings hard at everything, but strikes out most the time and hasn’t hit a homer since the Reagan administration. While you don’t want your AML system to over-compensate and stop swinging entirely, there is usually room to tighten up that strike zone to improve the ratio of alerts to SARs. On the other hand, if a rule generates only a few alerts, but many of those alerts are escalated, you may want to lower those rule parameters. Either way, reviewing the stats and conducting analysis and documenting revisions shows regulators that you are pursuing an efficient AML System.

Practice, practice, practice

The first baseman fields the ball, spins, and throws to the pitcher covering first for what should be the final out. Only the pitcher, who wasn’t paying attention during training, was slow to leave the pitcher’s mound. So the ball sails into the dugout, runners on second and third score, and the game is over.
Sometimes one little glitch in the system messes up the whole game. The same is true for an AML monitoring system. The potential for a tiny flaw in the parameter settings may result in missed alerts or false positives. And often these flaws are missed because internal personnel didn’t understand them or have access to the information behind the alert. It is important that analysts have sufficient knowledge of what causes alerts, and that if they find transactions that triggered alerts but shouldn’t have, they do a little digging to find out why. That means training. It also means documentation of the effort to identify and correct the errors, which also promotes model accuracy and efficiency.

Hit to all parts of the field

The AML monitoring needs to be tested on both a macro and micro basis. That means looking at thresholds for all transactions as well as those for specific customers or members who may warrant exceptions to given thresholds. What’s more, when rule thresholds are raised or lowered, it is important to compare the number of generated alerts and output ratios to what you got from the old settings. If a test of a higher threshold results in 100 fewer alerts generated in a given period, but your review of those 100 alerts that fell below the new threshold reveals no activity that should be escalated, then that suggests that the new threshold level may be valid. Conducting and Documenting this type of sandboxing on both a macro and micro basis promotes and demonstrates the merits of adjustments made to the model’s inherent risk-detection capacity and overall system efficiency.
While these types of efforts may be time-consuming on the front end, much like a successful spring season, these are the kinds of efforts that can help make sure your financial institution is still playing ball in October.