Prediction-Market Integrity Research.
MSR Decode is building an empirical map of where prediction-market integrity fails, where it survives scrutiny, and where the available evidence cannot yet determine the answer — tested systematically across defined market universes, not one wallet or one market at a time.
Why this sits apart from case work
An individual forensic investigation answers one bounded question. It often surfaces a market-structure mechanism that recurs beyond the case that found it.
The cartel-arrest markets case is a working example: the wallet pair behind two contracts whose displayed volume was generated at minimal net economic cost recurred on a third contract, and was found on a fourth we had missed, plus one unrelated market, once we went looking. That recurrence is what integrity research is for — taking an observation from one case and testing whether it holds, generalizes, or breaks down across a defined universe of markets, rather than leaving it as a single anecdote.
Public research is not a side activity here. It develops reusable methods, infrastructure and specialist market knowledge that strengthen commissioned investigations, and it is how MSR accumulates prediction-market-specific expertise that a generic blockchain-forensics or conventional market-microstructure background does not carry on its own.
How the work fits together
Three activities, each feeding the next.
Forensic Investigations
Bespoke reconstruction of a wallet, market, account cohort or claim — the primary commercial work today, and the source of most integrity-research questions.
Integrity Research
Individual observations tested across a defined market universe, so an isolated finding can become generalizable evidence — or fail to reproduce at scale. Reusable analytical methods and accumulated domain knowledge.
Institutional Intelligence
The logical extension of a mature research program — monitoring, structured datasets, settlement-risk classifications, market-quality metrics. Not yet built; named here so the trajectory is legible, not implied to exist today.
Research domains
Eight areas MSR tests systematically. Each domain names the class of question, not a completed finding. Where a domain already has a related case link below, that is a single case observation — real, dated, and scoped to one wallet, market or event — not yet a systematic integrity test across a defined universe; see “How a systematic test will be published” below for that distinction.
No systematic integrity test has yet been published under this framework. Case observations are shown separately above and should not be read as universe-level findings.
Settlement & Resolution
Benchmarks, resolution sources, settlement disputes, and external influence over how a contract resolves.
Two of 24 markets resolved on the Chainlink BNB/USD data stream these contracts settle on, opposite to the direction Binance’s own spot price had moved during the window. The wallet’s single largest loss came from one of them: the wallet traded in the direction Binance spot had moved, yet the market resolved the opposite way under the Chainlink benchmark.
Source · MSR-CASE-03Contract Design
Participant control over outcome-defining terms, incentive structure, ambiguity, and market-structure choices baked into a contract.
Price Formation
Liquidity, how information gets incorporated into price, price impact, and thin-market effects.
A late-window timing signal against a quote that, during the observed session, did not reprice for the shrinking time left on the clock.
Source · MSR-CASE-03Market Data & Disclosure
Volume, PnL, API semantics, completeness, and how displayed figures relate to the underlying activity.
150 of 150 flagged positions reconciled after restoring omitted maker-side fills the default view had dropped.
Source · Trade-endpoint case fileA public PnL example did not survive complete economic reconstruction — the flagship exhibit’s cash flows contradict its displayed profit metric.
Source · MSR-CASE-05In the scoped cartel-event case, $117,153 of displayed trading notional was generated through repeated complementary transactions at $171.71 of combined net economic cost.
Source · MSR-CASE-06Trading Conduct
Timing, concentration, coordination, circular or offsetting activity, and incentive-driven trading behavior.
No pre-public positioning established across the two tested disclosure events.
Source · MSR-CASE-06Concentrated repeated trading observed between the same counterparties, generating displayed volume at minimal net economic cost. Common control is not established.
Source · MSR-CASE-06Surveillance & Controls
Detection methods, reconstruction techniques, false-positive rates, and how effective a given control actually is.
Financial Mechanics
Fees, reward programs, collateral, redemptions, and complementary or offsetting positions.
A complementary-pair position, occurring once in the sample, altered the session’s displayed trading economics.
Source · MSR-CASE-03Cross-Market Dependencies
External benchmarks, correlated markets on other venues, and recursive dependencies between contracts.
How a systematic test will be published
Every completed integrity test — a domain tested across a defined universe, not a single case — will publish in this fixed structure, the same discipline the methodology applies to commissioned work. No test has completed under this structure yet; it is published now so the standard is set before the first result is.
- Domain
- Which of the eight research domains this test sits in.
- Hypothesis
- The falsifiable question being tested.
- Universe
- The defined set of markets, accounts or events tested — and how it was selected.
- Test
- The method applied across the universe.
- Result
- What survived, using the same status vocabulary as every MSR finding, plus “Integrity deficit identified” where the test genuinely concerns an integrity failure rather than an ordinary forensic question — see the independence policy.
- Replication
- Whether the result holds on an independent slice of the universe.
- Materiality
- How economically significant the result is.
- Limitations
- What the test cannot conclude.
- Status
- Published / updated / reproduced / corrected.
Have an integrity question?
Integrity research and commissioned work run on the same standard.