For journalists.

MSR Decode uses public on-chain records, venue-published records, official records, and other verifiable public sources appropriate to the specific investigation. Each published investigation identifies the relevant source set and material limitations. For a Polymarket wallet, that means activity as served by its public endpoints; other cases draw on public market data, official filings, or other verifiable public sources named in that case’s own workpaper.

• Wallet or account trading history, as served by the relevant venue’s public endpoints.
• Fill-level reconstruction and chronology, built from that same served history, market by market.
• Realized-PnL reconciliation against the venue’s own displayed profit figure, with the gap disclosed either way.
• Position timing, reconstructed against the market’s own clock.
• Concentration and pattern analysis across an account’s public activity.

Each of these is only as good as what the source served for that record in that window, and our own published work exists because that isn’t always the same thing as what actually happened — a completeness signal can read clean on a fetch that is mostly missing. Findings are published with the reconstruction behind them, and where a figure could not be tied to a reconstruction, the page carrying it says so.

Trading activity by itself does not establish real-world identity, common control, intent, possession of inside information, beneficial ownership, or off-chain communication.

• The real-world identity of an account’s owner.
• Common control between two or more accounts.
• Intent.
• Possession of inside information.
• Off-chain communications.
• Beneficial ownership.

MSR does not make these attributions without independent corroborating evidence sufficient to support them. Declining them by default is part of the method: a public trading record carries no name, no motive, and no transcript of a conversation, and no amount of pattern-similarity analysis manufactures one. Where a finding would require guessing at any of the six, MSR Decode does not publish a finding — it publishes what the record does and does not show, and stops there.

The wording rule findings are held to.

The rule: state what the evidence establishes, and state what it does not establish. Published findings are versioned, and material corrections are recorded publicly, dated. Read the research archive →

The surfaces that hold the numbers, the verdicts, and the corrections — check the figure there, not here.

• Scorecard — the published July 2026 leaderboard audit of the highest-earning wallets on Polymarket, scored against a published verdict rule, including every wallet we refused to score.
• Ledger — forward-tested outcomes on every wallet that earns a Real edge verdict, published at fixed checkpoints, gains and losses alike.
• Workpaper — the per-wallet reconstructions behind the scorecard, published against the venue’s own displayed figures.
• Record — the research archive: published research, material revisions and corrections, dated.
• Independence — in that page’s own words, a verdict here cannot be bought.

Named case files, with each page’s evidence, method and limitations stated on it.

• The weather-bot proof wallets →
• The BNB 5-minute wallet →
• The McDonald’s bet: a staged “$100,000 win” checked against the real market →
• The NYC Council’s staged-trade claim — what public records can test →
• Sample Decode Report: myzbsq →
• A trade-endpoint completeness defect behind apparent phantom positions →
• The Leaderboard Autopsy (archived — superseded research) →

Standalone research notes now originate at msrdecode.com/notes; from here on, social posts link back to the canonical version there.

For data reporters: the artifacts behind published figures.

The workpaper publishes the per-wallet reconstructions behind the scorecard against the venue’s own displayed figures. For the ledger, raw provenance — in the ledger’s own words, “full position dumps, content hashes, and each freeze's exact fetch” — is committed and available on request under the auditor-access terms stated on that page. Read the ledger’s auditor-access terms →

Where MSR’s analysis has assisted a reporter’s own investigation.

• Anna Gross, Financial Times. MSR Decode assisted Financial Times journalist Anna Gross with Polymarket fund-flow analysis for her reporting. View public acknowledgment →

The acknowledgment above is Anna Gross’s own public post. Assistance of this kind is analysis of public records; what a reporter publishes, and how, remains theirs.

Working on a story? Send us:

1. The claim. What are you trying to verify?
2. The evidence. Wallet, market, transaction, screenshot or source material.
3. Your deadline. We will tell you whether the public record can answer it.

Put your deadline in the subject line — press requests are triaged first. What comes back depends on what the record supports: a finding with its evidence and its limitation attached, or “could not verify,” said plainly rather than guessed.

The request can take four shapes: one specific factual claim checked on deadline; a fill-level reconstruction of a wallet or market; the data behind a pattern, with definitions and selection criteria; or a method check on analysis you already have. MSR reviews every note and replies personally, usually within a business day. No sales sequence.

• ceo@msrdecode.com — Press, factual verification →
• LinkedIn — Chris Park →

Press support and commercial work are kept separate.

Routine factual verification for journalists and researchers working on a story for publication is provided without charge. Commissioned research is separately scoped work — see Capabilities. In either case the wording rule above holds, and a negative or “insufficient” finding renders exactly like a positive one. Read the independence policy →