Two days after Mexico cruised, Brazil walked into a very different night.
Across Polymarket’s three match markets, $21.05 million changed hands between 72,998 wallets over 97,628 trades. That is more than double the Mexico-South Africa opener. A Brazil game draws a crowd. A Brazil game that slips draws a stampede.
We recorded the order book every 10 seconds and rebuilt every on-chain fill. What follows is how a 58% favorite drifted into a draw, and the three completely different strategies that made money from it.
The marketplace at a glance
This is a tightly made market. A one-cent spread sits on a multi-million-dollar book, and buying dominates: 79% of trades are takers hitting resting orders.
It also looks much broader than it is. The average trade was just $216, but the top 20 wallets did 54% of the entire $21M. The top 80 did 95%. A few dozen accounts moved the market. Seventy-three thousand watched.
How the book reacted to the match
Figure 1 is an interactive line chart of implied win probability. Brazil opens at 58.5%, falls after Morocco’s 20th-minute goal, recovers on Brazil’s 32nd-minute equaliser, then drifts down as Draw climbs from 26.5% at kickoff to over 75% by the 85th minute, crossing above both teams and ultimately paying out.
Morocco scored in the 20th minute. Brazil’s win probability fell from 47.1% to 32.8% in a single 10-second snapshot. Morocco’s implied chance jumped from 24.4% to 39.3%. Twelve minutes later, Brazil equalised and the market whipped back: Brazil 28.5% to 45.4%, Morocco 41.5% to 22.1%.
Then something quieter took over.
With the score locked at 1-1, the draw line kept climbing. It went from 26.5% at kickoff to over 75% by the 85th minute, crossing above both teams along the way. The market did not need another goal. It repriced, tick by tick, on the absence of one.
| Event | Brazil win probability | Move |
|---|---|---|
| Morocco goal (20’) | 47.1% → 32.8% | −14.3 pts |
| Brazil goal (32’) | 28.5% → 45.4% | +16.9 pts |
| Full time (1–1) | → Draw 78.5% | Brazil 12.1% |
Volume and traders flooding in
Figure 2 is an interactive chart of per-three-minute trading volume in US dollars and cumulative unique wallets. The largest in-play spike is $1.67 million around the 63rd minute. Morocco’s goal window pulled in 2,665 first-time wallets. Cumulative wallets climb from about 17,000 at kickoff to 72,998 by full time.
The biggest in-play surge came deep in the second half: $1.67 million in a three-minute cluster around the 63rd minute, as traders collectively realized the draw was becoming the favorite and repositioned into it. (In Mexico-South Africa, the largest spike was pre-match. A swinging game moves the money live.)
Goals also pull in new participants. Morocco’s goal-minute window alone brought in 2,665 first-time wallets. The count climbed from about 17,000 at kickoff to 72,998 by full time.
By phase, the split was 30% pre-match, 60% in-play, and 10% after. The more uncertain the outcome, the longer the tape stays alive.
A machine’s market
Figure 3 is a bar chart of on-chain fills per minute. Baseline is a few hundred fills per minute. Spikes reach 1,514 at Morocco’s goal, peak at 1,942 just after Brazil’s equaliser, and surge again past 1,600 near full time.
Trade frequency tells you when the market panicked. Baseline was a few hundred fills a minute. It spiked to 1,514 at Morocco’s goal, peaked at 1,942 just after Brazil’s equaliser, and surged past 1,600 near full time as the draw settled.
Those vertical walls are not fans typing faster. They are algorithms reacting in the same sub-second window after every goal. The chaos on this chart is where the money below was made and lost.
The money: three ways to win
The market has resolved. Draw = Yes. Here is realized profit-and-loss for the largest participants. Wallet addresses link to PolygonScan so you can verify the figures on-chain.
The losers: one bet, one mistake
| Wallet | Loss | Trades | What they did |
|---|---|---|---|
| 0x0a55…f9be | −$348,664 | 418 | Brazil-Yes accumulator |
| 0xb6d6…be17 | −$293,177 | 1 | single $293k Brazil bet |
| 0x8cb4…1057 | −$290,000 | 75 | Brazil-Yes |
| 0x5375…aeea | −$264,414 | 253 | mostly Brazil-Yes |
| 0xf031…c80c | −$184,998 | 391 | bet against draw |
Everyone at the top of the loss table did the same thing: backed Brazil to win, and held to zero.
A draw is the most expensive result in the book because both win-sides lose at once. The single biggest loss, −$348,664, is larger than Mexico-South Africa’s entire top five combined. The second wallet lost $293,178 in one transaction: one click on the favorite, held to nothing.
The winners
| Wallet | Profit | Trades | What they did |
|---|---|---|---|
| 0xfe78…0319 | +$183,915 | 326 | Brazil-No + Draw-Yes |
| 0xf751…fec8 | +$124,094 | 5 | bought draw early, held |
| 0x2005…75ea | +$103,564 | 3,019 | arbitrage bot |
| 0x1b47…d814 | +$88,726 | 100 | bet Brazil wouldn't win |
| 0xe907…cff6 | +$42,108 | 239 | backed the draw |
The winners split into three completely different playbooks.
Way 1: the conviction trader. Buy the cheapest truth, then hold it. The wallet at +$124,094 made its money in five trades. It bought the draw at 26¢, 38 minutes before kickoff, sat through the entire match without flinching, and sold at 99.9¢ at the end. It found the most underpriced outcome before a ball was kicked and simply held.
Way 2: the scalpers. Ride the swings, get out before they flip. The odds moved so violently that hundreds of wallets traded both sides. 504 wallets round-tripped Morocco-Yes (buy at 0.26 right after the 20’ goal, sell at 0.41 by the 26’, out before the equaliser). 442 round-tripped Brazil-Yes (buy at 0.28 while losing 0-1, sell at 0.45 into the equaliser six minutes later).
A handful of dedicated scalpers worked both swings for roughly $20k and $9.5k. But scalping is the small-money game here. The in-play book is too thin to push size through. The best single-swing scalper made $7,662. Most made $400 to $2,000. Speed caught pennies. Conviction caught the dollars.
Way 3: the arbitrage bot. Win no matter who scores. The most active winner (+$103,564) never bet on a result at all. It fired 3,019 trades, buying both the Yes and No of all three markets continuously through the match.
In any market, Yes + No must settle to exactly $1. When the book reprices violently, the two sides briefly fall out of sync. For a few seconds you might buy Brazil-Yes at 0.31 and Brazil-No at 0.67 (total 0.98), then redeem the pair for a guaranteed $1.00. Bank 2¢ risk-free. Do that 3,000 times across every goal and every wobble, and the pennies become $103,564.
The bot does not care who wins. It harvests dislocation, not direction. That is why its activity tracks the goal spikes on the frequency chart. The chaos that wiped out Brazil backers was its raw material.
The pattern that only shows across games
The same wallets keep appearing:
| Wallet | Mexico–SA | Brazil–Morocco | Combined |
|---|---|---|---|
0xfe787d…0319 | +$62,040 | +$183,915 | +$245,955 |
0xe9076a…cff6 | +$72,936 | +$42,108 | +$115,044 |
0xf0318c…c80c | −$53,043 | −$184,998 | −$238,041 |
The biggest winner of this match is up nearly a quarter-million across two games, betting against favorites and toward draws. The third wallet is down nearly as much, having faded the eventual outcome both times. The same hands win. The same hands lose.
The takeaway
Brazil-Morocco was the mirror image of the Mexico blowout. A draw punishes the obvious bet (back the favorite) more brutally than almost any other result. It rewarded three different kinds of edge: conviction (price the underpriced outcome early and hold), speed (scalp the swings, but only for pennies), and infrastructure (an arb bot that turns the market’s own chaos into risk-free profit).
Tens of thousands of fans showed up to watch Brazil. A few dozen machines and sharps quietly ran the market underneath them. Win or lose, they are the same names every match.
Order-book and win-probability figures from our 10-second recording. Volume, wallets, trade frequency, and scalper/arb-bot reconstructions from the complete on-chain OrderFilled log. Profit-and-loss from Polymarket’s resolved values for the highest-volume wallets. Reconstructed and analysed by TiltDesk.