Insider edge is uncopyable: the 2026 Polymarket prosecutions and what copiers should learn

The surest way to lose money copying a Polymarket "winner" is to copy one whose edge was information you'll never have. 2026's insider-trading prosecutions show why: an information-driven record can look elite right up until the morning of the arrest — and an information edge is the single least copyable kind there is. It's tied to one event, it ends the instant the trader is charged, and it's invisible to you beforehand. The durable lesson for due diligence isn't "the leaderboard is full of insiders" — the evidence says it mostly isn't — it's that a systematic, repeatable edge is worth copying and an episodic, one-event spike is not.

What happened on Polymarket in 2026?

Two criminal cases and a Congressional probe put insider trading on prediction markets into the headlines. All of the following are allegations; the charges are unproven and at least one defendant has pleaded not guilty.

• The Maduro-raid case. A federal indictment unsealed on April 24, 2026 charges U.S. Army Master Sergeant Gannon Ken Van Dyke. Prosecutors allege he used classified information about a U.S. military operation to capture Venezuela's Nicolás Maduro ("Operation Absolute Resolve") to place roughly $33,000 of bets on related Polymarket markets in late December 2025 and January 2026, then profited about $409,881 when those markets resolved "YES." The indictment also alleges he later asked Polymarket to delete his account. He is charged with Commodity Exchange Act violations, wire fraud, and an unlawful monetary transaction, and has pleaded not guilty.
• The "Year in Search" case. In late May 2026, prosecutors in the Southern District of New York charged Michele Spagnuolo, a Google information-security engineer who allegedly traded under the Polymarket account "AlphaRaccoon." The complaint alleges he used nonpublic Google information to bet on Google's "Year in Search 2025" results and profited about $1.2 million, and charges him with commodities fraud, wire fraud, and money laundering.
• The Congressional probe. On May 22, 2026, House Oversight Chair James Comer opened an investigation into Polymarket and Kalshi over concerns that government employees may be trading on nonpublic information, citing 80-plus suspiciously timed bets — including wagers placed hours before undisclosed U.S. and Israeli military action against Iran — and asking both platforms how they verify identity, enforce geographic restrictions, and flag anomalous trading.

You don't need to take a view on any individual's guilt to draw the copy-trading lesson. The lesson is about the shape of the edge, not the verdict.

Why can't you copy an information edge?

Because an information edge fails every test that makes an edge worth following. It is the opposite of the systematic, repeatable record a copier actually needs.

• It's episodic. An information edge resolves one event. The Maduro markets paid once; the "Year in Search" markets paid once. There is no next trade with the same advantage, so the headline return is a one-time print, not a process you can ride.
• It's legally fragile. The edge ends the moment the trader is caught. A record built on an information advantage doesn't decay gracefully — it stops, and the trader may be charged, frozen, or gone.
• It's invisible until it isn't. Before any arrest, an information-driven wallet looks like a brilliant call. A copier pointing a bot at it sees a clean, profitable record and no warning label, because "traded on information you don't have" is not a number in the PnL column.
• The account itself can vanish. The Van Dyke indictment alleges he asked Polymarket to delete his account after the operation. A wallet you're copying can go dark exactly when its record is most embarrassing — taking the only audit trail with it.

Put plainly: copying an information edge means inheriting all of its fragility and none of its information. You arrive late, after the event has resolved, holding a position whose entire rationale already paid out for someone else.

Aren't most top wallets just insiders, then?

No — and it matters to say so. The most rigorous study of the platform, the working paper Who Wins and Who Loses in Prediction Markets? (Akey, Grégoire, Harvie, Martineau, 2026), examined the behaviour of Polymarket's largest winners and concluded that insider trading is unlikely to explain their performance. Sensational prosecutions are real, but they are not the median story.

The everyday reasons copying loses money are more boring and more common: a farming-flagged record built to attract copiers, and the structural disadvantage of arriving as a late taker after the edge has already moved the price. Insider edge is a vivid hazard worth understanding, but it sits on top of those two — it doesn't replace them. Treat it as one species of a larger problem: uncopyable edge.

How does due diligence tell systematic from episodic?

By looking at where and how often the edge shows up, not at the size of the biggest win. A systematic record climbs a little across many independent markets — a point or two at a time, to roughly +20 units of cumulative profit in the illustration below; an episodic one sits near zero across dozens of markets and then jumps almost the whole way — about +18 in a single step — on one event, and that jump is doing all the work. Both can finish at about the same cumulative number; only one of them has anything left to repeat.

This is exactly what the Copy Score's edge-authenticity factor is built to read — whether profit looks like a repeatable process or a single lucky (or informed) event dressed up as skill. A wallet whose entire edge lives in one market or one week scores poorly on authenticity no matter how large the headline number is. The same instinct underpins reading a record like a quant: weight consistency and edge location over the biggest print.

What should a copier actually do?

Don't chase event-spike records, and don't assume a clean profit history is a safe one.

1. Discount one-event wonders. If a wallet's record is one enormous resolved market and a flat line everywhere else, treat the headline as unrepeatable — there's no process there to copy.
2. Weight edge location and consistency. A steady edge spread across many independent markets survives the next event; a concentrated one may not. Read the Copy Verdict for the sub-score breakdown, not just the top-line number.
3. Set an alert for out-of-character wins. A sudden, oversized win on a market unlike anything the wallet usually trades is worth a second look. An alert on a score or farming-flag change reaches you before you copy into a record that's about to stop working.

The wallets worth copying are the unglamorous ones with a repeatable edge — and only 1.3% of active wallets clear every test. An information edge, however large, is on the wrong side of that line.

CopyGrade is analysis-only — it never executes trades or holds funds — and a Copy Score is a documented research opinion about copy-trading risk, not a statement of fact about any trader, an accusation of wrongdoing, or financial advice.