Trade Integrity
How clean the trade flow looks at the microstructure level. In plain terms: how much of the volume is fake (wash trading), how concentrated the market is in a few hands, and whether flow is suspiciously one-sided.
How we score it
From a clean baseline, we screen the recent window for wash-trading proxies — the same wallet taking both sides, circular trading between linked wallets that end up flat, and single-wallet volume dominance — plus concentration measures over who holds and moves the market. Many independent participants and balanced flow raise the score; self-trading patterns, extreme one-sided pressure, or one wallet responsible for an outsized share of 24h volume lower it. The canonical academic method models trades as a directed multigraph and confirms a wash trade only when accounts round-trip back to roughly flat; concentration thresholds follow practitioner norms (top-10 over ~50%, or a high Gini, flag elevated risk).
How to read your score
Higher means cleaner order flow. A lower score flags microstructure patterns associated with manipulation — fake volume or concentration — without adjudicating anyone's intent. This axis is scored 0-100; the composite weights it at 20%.
No meaningful risk signal on this axis. Nothing here that warrants extra scrutiny before taking a position.
A clean read with only minor or ambiguous signals. Reasonable to proceed with normal due diligence.
A moderate signal worth a second look. Read the evidence on the market page before sizing a position.
A pronounced structural signal on this axis. Understand exactly what is driving it before participating.
A strong structural signal. This axis is the kind of thing a careful trader would resolve before trusting the market's price.
Signals we look for
Wash-trading proxies: circular trades between linked wallets that return to roughly zero net position.
The same wallet (or cluster) repeatedly taking both sides of the book.
One wallet responsible for an outsized share of recent 24h volume.
Extreme one-sided taker pressure relative to the resting book.
Top-of-book holder concentration (top-10 share or Gini) above practitioner risk thresholds.
References
The academic, industry, and regulatory sources that ground how this dimension is scored. Each note states what the source backs.
Detecting and Quantifying Wash Trading on Decentralized Cryptocurrency Exchanges
Victor & Weintraud, WWW '21 · 2021
The canonical wash-trade method: directed-multigraph SCC detection confirmed only when accounts round-trip to ~flat — exactly our wash proxy.
Crypto Wash Trading
Cong, Li, Tang & Yang, Management Science 69(11) · 2023
Distribution-shape detection (Benford, round-number clustering, power-law tail) plus the Gini/HHI/CR10 concentration indices we threshold on.
Polymarket wash-trading study (~25% of volume likely wash)
Sirolly et al. (Columbia), SSRN 5714122 · 2025
The live, on-prediction-market proof the wash signal fires — ~25% of Polymarket volume, peaking ~60% weekly, sports ~45%.
Statistical signals, not allegations. Every dimension measures a structural property of a market — the shape of its flow, the clarity of its question, the trustworthiness of its resolution source. A low score flags where a careful trader should look harder. It is never a claim that a specific account, or the market itself, has broken a rule.