On its first live run the markers-of-harm indicator raised seven flags. All seven belonged to actors we already knew were automated or part of a linked cluster. None was a plausible individual in difficulty.
The mechanism is worth understanding. A cluster quoting both sides of a market produces runs of rising stakes as a by-product of how it works, and rising stakes after losses is precisely the escalation pattern the indicator is built to notice. On a second platform the flagged accounts had a median gap between trades of zero seconds: same-second bursts, where the automation hint fired on its own.
A screen for consumer harm that is looking at the operator's own automated flow is not yet looking at consumers.
Signature
Separating automation from distress, which is the whole task for this indicator.
- Automation: trade spacing
- Median gap under about 90 seconds, and metronomic
- Automation: session shape
- Short bursts, two-sided, account sits in a counterparty cluster
- Escalate instead: spacing
- Irregular, human
- Escalate instead: direction
- One-directional stake escalation sustained across hours
- Escalate instead: context
- Chasing after real drawdowns, long spans without rest, not a cluster member
The practical conclusion loops back to attestation. Genuine consumer-harm monitoring needs the automated population identified and filtered out first, and only the operator can supply that. Until then this indicator is doing something useful but narrower than its name suggests, and saying so is part of reporting it honestly.