Notes
What the record actually shows.
Cases from the pattern library, anonymised. Two of the six are negative: patterns that turned out to be a platform's own infrastructure rather than anything worth reporting. Those are the more useful half.
- Consumer protection
The harm markers were software
Every flag the consumer-protection indicator raised on its first live run belonged to an automated actor. That is the indicator working, not failing, and it says something uncomfortable about what the screen is currently able to find.
- AML/CFT
When a funding web turns out to be plumbing
Twice, our own account-linking produced a large, impressive, entirely fictitious network. Both times the network was the platform's infrastructure. We publish these because the negative cases are the more useful half of a pattern library.
- AML/CFT
One cluster, two patterns, one story
The accounts already flagged for self-trading were also building their positions exclusively out of small fills. Reported by indicator, that is two findings. Reported by participant, it is one.
- AML/CFT
New accounts, sizeable positions, and a deadline
Five accounts, all members of a cluster already flagged for self-trading, built positions roughly thirty-two hours before their markets settled. No single one of those three facts would have been worth a second look.
- AML/CFT
Fifteen seconds, ninety-five pence
The smallest finding we have recorded was worth under a pound. It matters because the shape of it does not change when you multiply the numbers by ten thousand.
- AML/CFT
Market making and wash trading look identical
Eight linked accounts produced a third of the volume in three related markets. That is either a liquidity operation working exactly as intended, or artificial volume, and public data cannot tell you which.
Every note is de-identified. No platform, operator, market or account is named, and no fixture is identifiable. The quantitative signatures are reproduced as recorded, because a signature with no platform attached is the part that transfers.