Decode Report: myzbsq — MSR Decode sample

Displayed vs. real PnL

We reconstruct profit from the complete on-chain fill log (not the profile number). For myzbsq the two agree almost exactly — the profit is genuine. The question isn't whether it's real; it's whether it's repeatable.

Polymarket displayed

$1,223,224

MSR Decode reconstructed

$1,218,186

99.6% — residual is fees

What actually happened

Every dollar came from one market, one side, over ~14 hours.

Market: "Will Argentina win on 2026-07-03?" (FIFA World Cup, Argentina vs. Cabo Verde). The only market this wallet had traded as of this July 4 snapshot.
The bet: bought the NO side — Argentina not winning — at an average of $0.15, when the market priced Argentina an ~85% favorite. A contrarian shot at ~6:1. This is a three-way moneyline settled on the 90-minute result, so a draw pays NO.
The result: the match was 1–1 after 90 minutes (Messi 29', Duarte 59'), so the market resolved NO — even though Argentina went on to win it 3–2 in extra time. The position ran $0.15 → $0.75 → $0.99; he sold all 1,623,065 shares into the move.
What that means: the $1.22M came from betting against a team that won the match. The edge, such as it was, sat in the 90-minute settlement rule — not in reading the game.

Cash-flow ledger — reconstructed from on-chain fills · legs shown rounded; the total is the exact reconstructed figure
Leg	Detail	Shares	USDC
BUY · NO	20 fills @ $0.14–0.16, Jul 3 (14 crossed the spread and paid fees)	1,623,065	−$249,281
SELL · NO	159 fills @ $0.75, Jul 3–4	623,064	+$467,298
SELL · NO	1 block @ $0.999, Jul 4	1,000,001	+$998,971
Rebates/reward	6 credits attributable to this position, settled Jul 4–9 (see Rebate attribution below)	—	+$1,198
Net realized	flat — bought 1,623,065, sold 1,623,065	0 held	+$1,218,186

Edge decomposition & classification

Directional skill vs. arbitrage: 100% directional. 0% post-resolution (nothing bought at ≥$0.99), 0% bilateral (never traded the YES side), 0% reward-farming. This is a clean, one-sided bet — not a bot.
Concentration: 1 market, 1 outcome, 1 day. No diversification, no track record, no second data point.
Classification: single-event contrarian bet — a directional wager on one match at longshot odds that resolved in his favor.

Replicability verdict

Do not copy this wallet.

The $1.22M is real, but it is n = 1. Whether it was a sharp read (a lineup or injury signal) or a longshot that simply landed, it is a single resolved binary event — the match is over and the position is closed. There is nothing live to mirror.

Following "myzbsq +$1.22M" copies an outcome, not an edge. At $0.15 entry, this bet loses its entire stake roughly 85% of the time the same situation repeats. One winning ticket is indistinguishable from skill — and on a sample of one, you cannot tell them apart. A copier arriving after the fact inherits all the tail risk and none of the (unproven) edge.

What the data can't tell us

Honest limits, stated up front:

Whether the entry was informed (injury/lineup news) or purely speculative — on-chain data shows the bet, not the reasoning.
Whether this wallet is one of several run by the same operator (no linked-cluster signal found in this market).
Maker vs. taker role on individual fills is not reliably exposed by the public API.

Method & confidence. Reconstructed from the complete Polymarket /activity fill log. (/trades returns the same 180 fills for this market, but only if you pass takerOnly=false. It defaults to true, which returns only the fills where the wallet crossed the spread — 15 of the 180 here, and 27 across the wallet's whole history.) Cash-flow = sells + redemptions + rebates − buys, using the fee-inclusive usdcSize of each fill, cross-checked against Polymarket's own displayed profit (agreement: 99.6%). Note the residual is a definitional gap, not measurement error: the displayed figure is struck before fees, ours after. On a like-for-like pre-fee basis the two agree to $11 on $1.22M. This is a decision-grade decode: the wallet is flat and single-market, so reconstruction is exact.

Rebate attribution. Rebates settle on Polymarket's schedule, after the trading that earns them, so this leg spans a wider window than the fills do. It is the six credits attributable to this position: a $4.88 REWARD during the trading itself (Jul 4, 00:00 UTC), then five posted between the position going flat (Jul 4, 00:11 UTC) and the wallet's next trade (Jul 12) — a gap containing no other trading, so every credit in it belongs to this position: MAKER_REBATE $877.54 · TAKER_REBATE $5.39, $250.00, $50.00, $10.00. Exact total $1,197.8189, shown rounded as $1,198 in the ledger above. All six are printed to the cent, so adding them exactly as shown gives $1,197.81 — a cent under the exact figure, because at least one is rounded down for display. The ledger uses the exact figure, which is what makes it tie: $1,216,987.9130 of trading plus $1,197.8189 of credits is $1,218,185.7319, printed as $1,218,186. (Note added 2026-07-16, correcting this row: it previously read $1,193 — the five post-flat credits only — which left the ledger $5 short of its own total. The total was always right; the row was not.)

Snapshot scope. This report decodes the wallet's trading as of July 4, 2026 — flat and single-market, which is what makes exact reconstruction possible. One class of cash item crosses that date: rebates settle after the trading that earns them, so the rebate leg includes credits posted through Jul 9 (rule and amounts above). The wallet has since traded a handful of additional markets; the July 30-day scorecard cycle shows a larger displayed figure ($1.5M, 5 markets), with 64% of volume still in this one match. The verdict does not change: the profit came from one resolved longshot.