Brazil’s Election Shock Hit Crypto Markets Before the Rest of the Market Woke Up. And It Might Help Predict Other Big Events
When polls closed in Brazil's runoff at 17:00 on Sunday 25 October, no Brazilian FX desk was open. The central bank's PTAX fixing would not print until Monday afternoon, and the stock exchange would not open until 10:00 on Monday, only to halt for half an hour soon after. The BRL stablecoin pools on public blockchains never closed. That makes the weekend a natural experiment: what do onchain FX markets do when a national shock lands while the traditional market is shut, and who actually trades it?
The short answers: the pools priced most of the move overnight, activity spiked to several times a normal week, and the spike came mainly from arbitrage and a few dozen active wallets. The small, occasional wallets that look like retail did show up, in larger numbers than usual, but they traded in both directions and carried a small share of the volume. There is no onchain sign of retail panic.
THE PRICE
The pools moved first
On Monday the PTAX rose 9.19%, from 5.0377 to 5.5007 reais per dollar. On Tuesday it gave back 3.0%, to 5.3346, and the pools were trading right on it by Tuesday afternoon. The main pools (BRLA/USDC and BRZ/USDT on Polygon) made most of their move overnight, before Brazil's market opened, and settled at roughly 75 to 80% of the official move.
Dollar price in reais: onchain pools vs the official rate (PTAX), runoff weekend, 23–27 Oct
The pools moved overnight, before Brazil's market opened
The pools jumped within hours of the polls closing and were near their Monday level long before B3 opened. Dashed stretches are overnight hours with too few swaps to price.
The first round, on 4 October, moved the other way and more slowly. The PTAX fell 4.55% on Monday 5 October. The pools started falling on Sunday night but had covered less than half of that move by Monday's open; most of it came on Monday morning, alongside the official market. By the end they had stopped at about the same fraction of the official move as in the runoff.
Dollar price in reais: onchain pools vs the official rate (PTAX), first-round weekend, 2–6 Oct
The move went the other way, and the pools made most of it on Monday
Pool price as a share of Monday's official move, both rounds, by hour from Sunday 00:00
Both times the pools went about three quarters of the way; in the runoff they got there before the open
0% is Friday's official rate and 100% is Monday's. The runoff line is near its final level before Monday's open; the first-round line gets there during Monday.
The clearest way to see how far the pools went is to compare them with the official rate while it was being set. The central bank builds the PTAX from four quotes taken between 10:00 and 13:15. During those hours on Monday, a dollar in the main pools cost about 1% less than the official rate: 105, 115 and 97 basis points below it in the three main pools. In a normal week the pools sit slightly above the PTAX, by about 0.4 to 0.5%. So the pools had already priced a very large move, but stopped short of the full 9.2% jump. After the first round the real strengthened instead, and the gap appeared on the other side (1.0 to 1.4% above the PTAX). Both times the onchain price went most of the way and left the last stretch to the official market.
Pool dollar price vs the official rate during Monday's fixing hours (10:00–13:15), in basis points
While the official rate was being set, the pools stopped short of it
Below zero, a dollar cost less on chain than officially. After the runoff the pools sat about 1% under the PTAX; after the first round, about 1% over it.
Two smaller details: the real strengthened about 1.5% onchain in the hours just before polls closed, and the thinner Celo wBRL pools lagged the Polygon pools by about 5% overnight, a reminder that "the onchain price" is several prices with very different depth.
THE ACTIVITY
Several times a normal week, and mostly in BRL
We compared Sunday 00:00 to Tuesday 09:00 (Brasília time, 57 hours) against the same window in six normal weeks, taking the median of each measure.
| 57 hours | Normal | 1st round | Runoff |
|---|---|---|---|
| Peer | 16,959 | 32,455 | 44,838 |
| Settlement | 54,536 | 65,662 | 69,953 |
| Swaps | 1,570 | 5,118 | 9,050 |
| Swap vol. | $0.49M | $2.03M | $3.59M |
| Swappers | 227 | 511 | 913 |
Peer transfers rose 2.6×, swaps 5.8× and swap volume 7.3×. Payments on the BRLA settlement rail (tokens minted to pay a counterparty and burned soon after) rose only 1.28×. That rail is business payments, and it barely noticed the election.
Onchain BRL activity, Sunday to Tuesday 09:00, as a multiple of a normal week
Swaps rose almost sixfold and swap volume sevenfold; settlement payments barely moved
As a control, Argentine and Colombian stablecoin transfers in the same window were at 1.02× and 0.78× their normal level. Mexican tokens rose 1.76×, which says some of the weekend's activity is broader than Brazil, but BRL's 2.6× stands well above it.
Stablecoin peer transfers by currency, Sunday to Tuesday 09:00, as a multiple of a normal week
BRL jumped most; Argentina and Colombia stayed flat
THE TRADERS
Who was trading?
We sorted the 9,050 runoff swaps by the wallet that signed them, and split the transfers into their parts. Three findings stand out.
1. The swaps were mostly arbitrage and a few dozen active wallets.
- Atomic arbitrage, where the same transaction buys and sells reais (a bot by construction): 2,278 swaps, 25% of the total, against about 5% in a normal week.
- Heavy wallets, with 50 or more swaps in the window: 29 addresses (one or two in a normal week) and 2,642 swaps.
- Regular wallets, with 4 to 49 swaps: 312 addresses and 3,552 swaps.
- Occasional wallets, with 1 to 3 swaps, the closest thing in the data to retail: 572 addresses, 4.6 times the usual number, but only 578 swaps (6%) and 7% of swap volume. Their median swap was US$ 97.
BRL/USD swaps by type of sender, Sunday to Tuesday 09:00
The surge was mostly arbitrage and a few dozen active wallets
About 93% of swaps came through a router or aggregator rather than directly from the signing wallet, against 72% normally. That is consistent with automated execution.
2. The big wallets sold reais. The small ones didn't panic. In a calm market, roughly half of all swaps sell reais and half buy them. When the results came in on Sunday night, the busiest wallets turned hard: about three in every four of their swaps sold reais for dollars (74% for heavy wallets, 66% for regular ones). The small, occasional wallets kept doing what they always do, buying and selling in roughly equal measure (44% to 49% selling) from Sunday afternoon right through Monday. A panic looks like everyone running for the same exit. Here the small wallets stayed put. The selling, and the price move that came with it, came from the active traders.
Share of runoff swaps that sold reais for dollars, by phase and type of wallet
The big wallets sold reais as the count came in; the small ones stayed balanced
3. More transfers, but not smaller ones. Transfer counts rose 1.6 times and the average ticket fell by a third, which could read as many people moving small amounts. Taking the transfers apart says otherwise:
- Swap legs. Every DEX swap also moves tokens between the trader and the pool, and those movements appear as ordinary transfers. They went from about 3,200 to 17,500 (5.5×), all of it the trading described above.
- Other peer transfers (wallet to wallet, exchange deposits and withdrawals) went from 13,800 to 26,000 (1.9×). Their median rose to R$ 484 from R$ 172, and the top ten senders' share fell to 35% from 53%: more people, sending more each.
- The settlement rail (BRLA minted to pay a counterparty and burned soon after, mostly business payments) rose only 1.3×, from 54,500 to 70,000 payments, and its volume was flat at about R$ 240M. It barely noticed the election, and it has been growing steadily for weeks anyway.
| Sun to Tue 09:00 | Normal | Runoff |
|---|---|---|
| Swap legs count | 3,178 | 17,464 |
| volume | R$4M | R$30M |
| average | R$1,208 | R$1,709 |
| Other peer count | 13,762 | 25,993 |
| volume | R$338M | R$395M |
| average | R$24,555 | R$15,191 |
| Settlement count | 54,536 | 69,953 |
| volume | R$247M | R$236M |
| average | R$4,525 | R$3,371 |
Where the extra transfers came from. XRPL BBRL (about 1,400 runoff transfers) is not split and is left out.
| Other peer transfers | Normal | Runoff |
|---|---|---|
| Median transfer | R$172 | R$484 |
| Under R$100 | 44% | 37% |
| Top-10 share | 53% | 35% |
| Senders | 1,931 | 3,272 |
The average ticket fell because thousands of small swap legs joined the count and because the largest BRLA flows paused during Monday morning, not because people sent less.
THE CIRCUIT BREAKER
A faint mark on chain
B3 halted equities at 09:57 on Monday, with the Ibovespa down sharply, and reopened at about 10:38 after an auction. It closed down 13.88%.
On chain, the halt left only a faint mark. BRLA transfer counts held at 1.6 to 1.9 times normal throughout, though BRLA volume dropped in the halt hour (R$ 15M against R$ 70M in a normal Monday 10:00 hour) and stayed low until midday; R$ 162M then moved in the 16:00 hour. BRZ ran at four to five times normal all day. BBRL on the XRP Ledger had no peer transfers from 08:24 to 16:32, but that silence began before the halt and covered the whole trading session, so it says more about how BBRL is operated than about the circuit breaker.
BRL stablecoin transfers per hour, every token and chain, vs a normal week
Choose the event, the measure and the token; the circuit breaker barely shows
The market that couldn't wait
At 17:00 on Sunday the polls closed, and the people who trade the real for a living had nowhere to go. Brazil's FX desks were shut, the PTAX wouldn't print until Monday afternoon, and B3 wouldn't open for another 17 hours. The stablecoin pools were the only Brazilian real market still open.
A few dozen of those traders used it. As the count came in, wallets that trade every day turned into heavy sellers of reais: three in four of their swaps that evening went from reais into dollars. Selling into a pool does one mechanical thing. It pushes that pool's price of the dollar above the price in the next pool, and above the price on every other venue that still quotes the real. Each of those gaps is free money for whoever closes it first.
That is what the bots are for. Atomic arbitrage, where a single transaction buys reais in one place and sells them in another, went from about one swap in twenty in a normal week to nearly one in three between Sunday night and Monday's open. Every one of those trades took the sellers' move and copied it into another pool. Within hours the pools were no longer pricing a few traders' fear, but something close to a market-wide price, with no desk open anywhere in the country.
The ordinary user was mostly a spectator. Small, occasional wallets did show up, almost five times as many as usual, but they bought and sold in equal measure and accounted for about 6% of swaps. The weekend's surge in transfers was the trading itself, a wider circle of people moving money between wallets, and the settlement rail's steady growth, not a run.
By the time B3 opened on Monday, the job was mostly done. In the hours before the open, the main pools stood at 5.43 reais per dollar, 84% of the way from Friday's PTAX to the 5.50 the central bank would fix that afternoon. Sellers set the direction, bots spread it, and the result landed close to where the official market priced the real once it could.
That makes the onchain real an interesting experiment in forecasting, as long as it's read carefully. It got the direction right after both rounds. It did not get the size right both times: after the first round, the pools had covered only 43% of Monday's move before the open. And a market of $3.6 million over a weekend can be moved by a handful of wallets. But on the night of the runoff, a small, open market answered the question everyone in Brazil was asking about 17 hours before anyone else could.
That is worth watching beyond Brazil. Elections, central-bank surprises and other shocks that land on a weekend or overnight all leave traditional FX markets shut while onchain pools keep trading. If the pattern holds across more events, the first price for a currency after a big surprise may increasingly be set on chain, and read there first.