A CrowdIntel investigation is the output of a deliberate statistical process. It's not a tip, a rumour, or an accusation — it's a public record of a wallet or group of wallets whose performance clears our publication bar. This guide walks through how to read one.

The publication bar (what every investigation has cleared)

Before you read the page, know what it means that this page exists at all. An investigation is published only when:

Enough resolved bets to reject luck at standard significance levels.
Excess win rate vs the category's base rate — beating 50% by a lot, not a little.
Positive realized PnL — win rate inflated by heavy-favorite betting doesn't count.
Low p-value against the no-skill null hypothesis.
For cluster investigations: on-chain evidence that the wallets are linked — typically a shared funder or timing correlation that rejects coincidence.

The exact numbers are not public (we don't publish thresholds). What matters is that the investigation is there because the combined evidence passed the bar — not any one metric.

Page anatomy

Every investigation page has the same structure. In reading order from top to bottom:

1. Headline + context

What the investigation is called. For cluster cases: "[N]-wallet funding cluster on [category]." For single-wallet cases: "Wallet [addr] on [topic/category]." The context line tells you the scope (category, topic, time window).

2. Key stats band

Four numbers you care about:

Wallets — how many addresses are part of the case.
Total bets — combined resolved bets across all wallets in the investigation.
Total volume — combined USDC moved.
Win rate — shrinkage-adjusted rate on resolved bets.

Sometimes also:

p-value — the probability of observing this win rate if the wallets had no edge. Lower is stronger evidence.
Category edge — how much the wallets beat the market category's historical base rate.

3. Funder or coordination evidence

For cluster investigations, this section shows the funder address and the trail from funder → individual wallets. For scope-narrow investigations (single wallet on one topic), this section shows the scope definition and why it's narrower than the wallet's full history.

Click through on any wallet link to see that wallet's full dossier. Click through on the funder link to see every wallet it has ever funded.

4. Timeline

A chronological view of the investigation's trades. This is the most interesting section for pattern-matching:

Bunched-in-time entries — multiple wallets placing same-direction bets within minutes. Coordination signal.
Pre-news timing — entries that landed before public information moved the market. Information-advantage signal.
Consistent sizing — wallets entering with very similar trade sizes often turn out to be one operator.

5. Top markets

The markets this investigation has been active on, ranked by volume. This is what's useful if you're deciding whether to trust the cluster on a current market — check if they've traded this category before.

6. Summary (optional)

For longer-form cases, a written summary explains what CrowdIntel observed, what we rule out, and what we're less sure about. Read this last; it's the human layer on top of the numbers.

How to translate an investigation into a trading decision

Three questions to ask yourself after reading:

Q1: Is the investigation active on the market I'm looking at?

Cross-reference the "top markets" list against the market you're about to trade. If the cluster is currently on the other side of your intended position, that's a signal to either flip your view or size down significantly.

Q2: Has the cluster's prior pattern held?

An investigation is a summary of past performance. Check the dates. If the most recent trade was 6+ months ago, the operator may have stopped or been replaced. Recent activity matters.

Q3: Is my information independent of theirs?

If you're bullish on a market because a Bloomberg article ran yesterday, and the cluster took the same side three hours before Bloomberg published — you're not independent. They got the information first; you're reading their trade through a journalist. That's still useful, but it's not a confirmation of your thesis.

Common misreadings

What to do next

See which wallets are currently part of active investigations — /investigations
Check if a specific wallet you're tracking is in one — open its dossier at /whales/[address]
Understand our methodology at a higher level — /methodology
Learn the terminology — /glossary

How to read a CrowdIntel investigation