SETTLED.

Prediction-market claims, checked against resolution data.

Whale file: Dishonest-Bloom — $3,507,478 down across 87 large resolved bets

$6,393,161 staked at an average entry of 63¢. Win rate 44.8% against a price-implied 63.0%. The skill test says: bleeding.

The claimLarge-bet wallet Dishonest-Bloom (0x2a2c…9bc1) — profiled from its full resolved record, not a screenshot. on-chain record
87Resolved large bets
$6,393,161Staked
−$3,507,478Honest net P&L
-3.9σSkill z-score

The short version

This wallet placed 87 large resolved BUYs ($6,393,161 total) at an average entry of 63¢, between 2026-03-09 and 2026-05-08. Scored honestly — each bet at the price it actually paid — the record is −$3,507,478 (−54.9% on stake).

The skill test: at these entry prices the market itself predicted 63.0% winners; the wallet delivered 44.8%. That difference is -3.9 standard deviations — outside the band luck can explain on this sample.

The style is in-play favorite chasing at maximum size, concentrated in two months of American college sports. All six of the wallet's biggest swings are losses, all on college football, college basketball, or NBA games. The catastrophe band is 80–90¢: $2.01M staked on heavy favorites that won only 43% of the time — a 44-point gap between price and outcome that no closing-line bettor produces by accident. The signature trade is $1.74M on TCU against Duke at 87¢, a total loss. Two other entries were recorded at effectively $1.00 — paying par for near-certainties mid-game — and both lost everything, which is what buying a live favorite seconds before a collapse looks like in the data.

A win rate of 44.8% against a price-implied 63.0% over 87 bets is not variance; at −3.9σ it is a measurable anti-edge. The most plausible mechanism is the oldest one: chasing live favorites after adverse moves, doubling into positions that feel safe at 85–99¢ exactly when the game state says otherwise, at sizes that guarantee the mistakes dominate. Every price band above 60¢ is negative. Whatever this wallet believes it is doing, its own entry prices predicted 55 winners; it delivered 39. The market was not wrong about these games — the wallet was, systematically, for $3.5M.

Technical appendix

Exhibit A Record by entry-price band

Honest net by 10¢ price band. Where a wallet's edge lives (or leaks) is usually visible here.

BandBetsWRStakedNet
10–20¢20%$39,231−$39,231
20–30¢425%$51,926−$507
30–40¢617%$366,710−$318,782
40–50¢1331%$590,769−$314,002
50–60¢1553%$731,275$44,708
60–70¢1856%$668,015−$89,327
70–80¢757%$429,651−$135,766
80–90¢743%$2,012,496−$1,893,815
90–100¢1553%$1,503,089−$760,757

Exhibit B Largest resolved swings

MarketEntryStakeNet
TCU Horned Frogs vs. Duke Blue Devils87¢$1,740,000−$1,740,000
Troy Trojans vs. Nebraska Cornhuskers100¢$405,760−$405,760
South Florida Bulls vs. Louisville Cardinals35¢$245,000−$245,000
North Dakota State Bison vs. Michigan State Spartans100¢$205,637−$205,637
Iowa Hawkeyes vs. Clemson Tigers45¢$200,598−$200,598
Suns vs. Spurs80¢$160,434−$160,434

Exhibit C Method note

BUY-side resolved bets only, scored as stake × (1−p)/p on a win and −stake on a loss at the recorded entry price p. Platform P&L columns and leaderboard figures are not used. The z-score compares actual wins to the price-implied expectation Σp with variance Σp(1−p) — i.e., the wallet is graded against the exact prices it chose to pay. Large-bet feeds under-sample small trades, so this is the wallet's large-bet record, not necessarily its whole book.

Source under review: https://polymarket.com/profile/0x2a2c53bd278c04da9962fcf96490e17f3dfb9bc1