We build market integrity surveillance for prediction markets.
Find out what a market can be shown to have done, and what its own monitoring missed, before anyone else does.
We measure from outside, using data that is already public, and map every finding to the obligation it engages. An operator ends up with evidence rather than an assertion. A supervisor ends up with measurement rather than self-certification.
12
indicators, each mapped to a duty
3
obligation families covered
4
capability layers above detection
1
indicator we publish as unable to see
The thesis
Prediction markets are becoming information infrastructure. Nobody independent is watching them.
Money and attention are moving into platforms whose prices are starting to be quoted as evidence. The supervision around them has not caught up. There is no shortage of data: activity settles onto a public record that anyone can read. What is missing is anyone outside the platforms reading it.
01
The observation gap
Trading on these platforms settles onto a public ledger, in full, permanently. Almost nobody reads it. The operator sees its own book and no one else's, and no independent party is reconstructing what actually happened from the outside.
02
The attribution gap
Detection alone cannot finish the sentence. Authorised market making and wash trading produce the same record: the same accounts, the same two-sided volume, the same interlocking pattern. Better rules do not separate them, because what separates them is not in the data.
03
The assurance gap
An operator reporting on its own integrity is marking its own work, and an in-house function reporting on the platform it belongs to has the same problem more quietly. Nobody in the picture can produce a measurement a sceptical third party would accept.
The attestation problem
The same behaviour, three classifications.
What changes the classification is not better detection. It is whether the operator has declared the accounts, and under what mandate. That answer exists in exactly one place, and it is not in the data.
| Observed | Behaviour | Declared | Classification |
|---|---|---|---|
| Cluster A | Continuous two-sided quoting across 14 markets | DeclaredMatches mandate | Routine market making. Recorded as control evidence. |
| Cluster B | Continuous two-sided quoting across 14 markets | DeclaredDiverges from mandate | Reconciliation exception. Activity exceeds the declared scope. |
| Cluster C | Continuous two-sided quoting across 14 markets | Not declaredNo mandate on record | Unexplained internal volume. Review under AML conditions. |
The behaviour column is identical in all three rows. What changes the classification is not better detection, it is whether the operator has declared the accounts and under what mandate.
Specimen figures. Not live data, and not drawn from any platform.
Capability
Detection is the sensor. Four layers sit above it.
Detection is the foundation and the part a competent team could rebuild. Each layer above it needs something the one below does not have, and the upper layers are structurally impossible from inside a single book.
01
Detection
Twelve indicators, run continuously against the public settlement record, each mapped to the obligation it engages.
Is anything happening on this platform that a duty attaches to?
02
Reconciliation
Check what an operator declares against what independent observation actually shows.
Is the busiest participant on this platform the one the operator says it is?
03
Benchmarking
Measure one platform against a baseline drawn from several, rather than against nothing.
Is this normal for the sector, or unusual?
04
Supervision
Check whether an operator's own monitoring caught what independent observation detected.
Is the compliance function actually working?
05
Assurance
Independent integrity reporting across platforms, produced by a party with no stake in the result.
Can this be shown to a third party and believed?
Honest limits
We publish what we cannot see.
Three of the twelve indicators are qualified and one is screening only. The alternative is worse than saying nothing: an operator told their platform is clean, by a system that could not have seen the problem, has been given false assurance against a duty they remain accountable for.
- 01Wash tradingObservableThe public settlement record supports this directly.
- 02Linked walletsObservableThe public settlement record supports this directly.
- 06StructuringObservableThe public settlement record supports this directly.
- 07Fund flow and layeringPartly observableDetected, but a manual check or a second source is needed.
- 12New-wallet positionObservableThe public settlement record supports this directly.
- 03Pre-event spikePartly observableDetected, but a manual check or a second source is needed.
- 04Coordinated positioningObservableThe public settlement record supports this directly.
- 08SpoofingScreening onlyThe evidence largely does not reach the public record.
- 09Cross-platform divergencePartly observableDetected, but a manual check or a second source is needed.
- 11Resolution-window riskObservableThe public settlement record supports this directly.
- 05Position concentrationObservableThe public settlement record supports this directly.
- 10Markers of harmObservableThe public settlement record supports this directly.
Three of the twelve are qualified and one is screening only. We publish this because the alternative is an operator told their platform is clean by a system that could not have seen the problem.
Specimen figures. Not live data, and not drawn from any platform.
Every market that matters ends up with an independent observer. The only question is whether it arrives before the scandal or after it.
Find out where you stand before you are asked.
The data is already public, so a first run needs no integration and commits neither side. What comes back is what a market can be shown to have done, what the operator's own monitoring would have had to catch, and where the evidence runs out. That holds whether you supervise licensed operators, run the framework several brands sit on, or hold an operating licence yourself.