<quote>They’re treated the same, those all have different scores but their values affect your overall risk/reward score for Amex. If that overall score reaches a certain threshold (which they’re still experimenting with what exactly it will be), you will get shut down. What they’re doing is based on the same ML models that are used at Amex for AML and fraud prevention. The shutdowns will happen in small waves because one giant hit would be a massive drain on resources and disruption to their overall business, including drawing negative public attention outside of the churning community. That would scare people who are profitable but think they’re winning. They are also incrementing shutdown waves because of the resource usage it takes to identify accounts and assign a score based on the new indicated risk factors. When shutdowns are triggered, they begin combing through the new dataset. Right now, the model only looks back two years.
The clawbacks on self-referrals were a pilot program to this week’s events, not an entirety separate event. The same model that identified self-referrers was used for these shutdowns.<end quote/comment on Doctor Of Credit website>
The Bluebird and Redcard shutdowns had a similar wave pattern. I only know one person who has survived with a Bluebird to this day, and that person actually still does some gc loading. I do think it's strange if Amex focuses on a retro data profile-- 2 years now and then even further back? Instead, you'd think they'd be battening down the hatches for all future gaming activities.
I also wonder if the number of accounts could be so large that "a giant hit would be a massive drain on resources." That statement would indicate a large number of customers, not a few heavy hitters.