Is that wallet’s record evidence, or luck?

A high win rate can still be luck

24 wins from 25 events at 90 cents average is a 96% win rate, and the verdict is still consistent with luck. Those prices already predicted about 22.5 wins, so beating them by 1.5 proves little. That is why this page does not rank wallets by profit or win rate. Type a record with no address at the skill-or-luck calculator.

How the check works

Every graded position is compared with the price the wallet paid for it. A position entered at 30 cents that resolves in the wallet’s favor counts as a win the market priced at 30 percent, so the price itself sets the bar each position had to clear. The check sums wins and losses against those stated odds and asks how far the record sits from what its own prices predict. Under our independence rules each set of mutually exclusive outcomes is one event, so a race with a hundred candidates counts once at most, and a wallet that bet on more than one of its candidates has that race left out of the test. Separate yes-or-no markets each count, even on one subject, except a repeating series such as one price threshold asked for several dates, which we count as one event. The confidence interval is sized by the number of resolved events.

What the checker covers

OVERROUND scores fills on Polymarket’s CTF Exchange and NegRisk markets, the classes that form graded prediction categories. Up or Down markets, the five minute crypto price series, are excluded by policy. An entire series collapses to a single event under our independence rules, and its outcomes track short horizon price noise rather than judgment.

The verdict also counts only positions on a single market that the wallet held on one side through resolution. Where a wallet held both sides of a market, or sold any of its position before resolution, the position is a trade rather than a forecast. A pair bought at 40 cents a side profits whichever way the market resolves, and scoring that against a single entry price would credit arbitrage as accuracy. Positions on several markets we count as one event, such as No on two candidates in the same race, are left out too. The test grades each event as one bet at the average price paid, and for such a position that average need not match its chance of profiting. For wallets at the top of a category board, these positions still appear in that category’s graded history, which subscribers see. The full policy is on the methodology page.

Polymarket Perps positions are matched off-chain and never pass through the exchange contracts the pipeline reads, so they are not part of a wallet’s graded record. A wallet that trades mostly perps will show a thinner record here than its Polymarket activity suggests.

The three verdicts

• Distinguishable. The record’s excess wins clear the interval that chance produces at its entry prices, at 95% confidence. This describes the resolved record only.
• Consistent with luck. The excess wins sit inside the range chance produces. Profitable wallets land here regularly, especially on thin evidence.
• Insufficient evidence. Too few independent resolved events in the scored classes to run the test either way.

OVERROUND holds its own work to the same standard. Every position it has alerted is graded in public on the receipts ledger, and the aggregate calibration numbers behind the method are on the statistics page. If the wallet you are checking is one you might mirror, the copy-trading guide covers what a record has to survive before it means anything.

What does the wallet checker do?

You paste a Polymarket wallet address and the checker tests that wallet's graded record against the prices it paid. Every graded position is compared with its own entry price, so the question being tested is whether the record beat the odds the market stated at entry. The answer comes back as one of three verdicts with a 95% confidence interval, plus a coverage summary and an evidence range.

Is this a P&L or profit rank?

No. Raw profit and win rate reward bankroll and easy prices. This check asks whether the record beat the prices the wallet actually paid, over resolved events. A wallet can be profitable and still be consistent with luck.

What does “consistent with luck” mean?

It means the wallet's excess wins sit inside the range that chance alone produces at those entry prices. A record can be profitable and still be consistent with luck. Favorable prices, a hot streak, or a few large positions can produce profit without demonstrated skill. The verdict reports what the evidence supports at 95% confidence, and more resolved events can move it either way.

Why are Up or Down markets excluded?

OVERROUND scores fills on Polymarket's CTF Exchange and NegRisk markets, the classes that form graded prediction categories. Up or Down markets, the five minute crypto price series, are excluded by policy. An entire series collapses to a single event under our independence rules, and its outcomes track short horizon price noise rather than judgment. The full policy is on the methodology page.

Which positions does the verdict leave out?

The verdict counts only positions on a single market that the wallet held on one side through resolution. A position on both sides of a market, or one the wallet sold any of before resolution, is a trade rather than a forecast, and scoring it against a single entry price would credit arbitrage as accuracy. Positions on several markets we count as one event, such as No on two candidates in the same race, are left out too. The test grades each event as one bet at the average price paid, and for such a position that average need not match its chance of profiting. Bets at 2.5 cents or less are not scored, so a position counts as one market held on one side when its only other bets, on the other side or on another market in the same event, were such bets, for example a sale at 97.5 cents or more. The full policy is on the methodology page.

Are Polymarket Perps positions graded?

No. Polymarket Perps positions are matched off-chain and never pass through the exchange contracts the pipeline reads, so they are not part of a wallet's graded record. A wallet that trades mostly perps will show a thinner record here than its Polymarket activity suggests. The checker covers resolved CTF Exchange and NegRisk markets only.

Why does the checker report ranges instead of exact numbers?

The free checker reports buckets, like 100 to 1,000 resolved events, because the exact graded record is part of the subscription. The buckets are honest at their boundaries. A wallet reported at 100 to 1,000 events has at least 100 and fewer than 1,000, and the verdict itself is computed from the full underlying record before any bucketing.

What does a subscription add?

A subscription shows the wallet's skill score and its rank in every category it trades. For a wallet at the top of a category board, it also shows that category's calibration diagram and graded history, one row per resolved event. It also includes a daily digest of qualifying moves by the wallets you follow, for any of them the alert feed tracks (the highest-ranked forecasters in each category). The checker's verdict stays free either way.