Five accounts built buy positions between roughly $880 and $1,750, each assembled from five to sixteen separate fills, about thirty-two hours before the markets they were in resolved. Every one of the five belonged to a cluster that the self-trading indicator had already flagged.
The finding is the composite. New exposure, linked accounts and an imminent settlement are each ordinary alone, and specific together.
Signature
The composite, not any single component.
- Observed account age
- Under 1 day at the time the position was taken
- Position size
- $500 to $2,000, built from 5 to 16 fills in one session
- Time to settlement
- 24 to 48 hours
- Cluster status
- Account belongs to a cluster concurrently flagged for self-trading
The honest reading of this case is that it is a demonstration of cross-indicator correlation rather than a finding in its own right. It also settled a methodology question: an indicator that depends on knowing how old an account is cannot be trusted until a platform has been observed for longer than the age threshold it tests against.
What this cannot show
- Under the production thresholds this returned nothing at all. We report that, rather than presenting the exploratory result as though it were the configured one.
- Nothing here establishes what any participant knew or intended.