Can DeFi Build a Better FX Market for Latin America?
Onchain FX promises a market that operates around the clock, settles on public infrastructure, and connects local currencies to dollar stablecoins without relying on a conventional exchange order book. But continuous access is only useful if the exchange rate is competitive. Across Brazil, Argentina, Colombia, and Mexico, the data show that some onchain markets trade close to official FX benchmarks while others maintain persistent premiums. The differences point to a more specific question: what determines whether an onchain FX market can offer a competitive rate?
THE PROMISE
An FX market built on open infrastructure
For people and businesses across Latin America, exchanging local currency for dollars is a recurring financial need. Dollar stablecoins offer a way to hold and transfer dollar-denominated value on infrastructure that operates across borders and outside conventional banking hours.
Onchain FX extends that model to the exchange itself. Automated market makers (AMMs) let users trade local-currency tokens against dollar stablecoins through shared liquidity pools. Prices are available continuously, transactions are publicly recorded, and the same assets can connect to other onchain applications.
The proposition is more than a different way to move dollars. It is a different way to organize the market in which local currencies become dollars: one that can be accessed at any hour and integrated directly into digital financial services.
That promise is testable. If onchain FX is to become a competitive alternative, the rates available in its pools should compare favourably with the relevant FX market, and remain competitive when users actually trade.
THE TENSION
A market that never closes still has to price dollars well
The cost of transferring a stablecoin does not tell us the cost of acquiring it. That depends on the exchange rate offered at the point of conversion, the liquidity available for a trade, and the mechanisms through which a local-currency token is issued and redeemed.
This study examines one part of that problem: the exchange rates implied by onchain swaps. It compares them with the official reference rates for Brazil (PTAX), Argentina (the central bank reference rate), Colombia (TRM), and Mexico (FIX). The difference is expressed as an onchain dollar premium. A positive premium means a dollar costs more local currency in the pool than at the reference rate; a negative premium means it costs less.
The dataset covers resolved swaps across nine blockchains from October 1, 2025, to October 11, 2026. To limit noise, a swap is included only when both sides are worth at least US$10, and a pool receives a daily price only when it records at least five qualifying swaps. Tokens with incomplete histories and undecoded swaps are excluded.
The broad result is clear: among pools with at least 1,000 qualifying swaps, none has a median price meaningfully below its official benchmark. The closest sits effectively at the reference rate.
Premium or discount: the window median per pool
Median premium over the official rate, with the 25th to 75th percentile of swaps; pools with at least 500 qualifying swaps
The median onchain dollar premium of every pool with at least 500 qualifying swaps, with the middle half of its swaps. No liquid pool sits meaningfully below its official rate.
That finding does not establish that onchain FX is more expensive than an exchange or an OTC desk; those executable alternatives are not measured here. It does establish that the observed onchain prices do not yet show a consistent discount to official FX benchmarks.
The more revealing result is how much those prices differ from one onchain market to another.
THE INVESTIGATION
Brazil: the price depends on the rail
Brazil offers the clearest comparison between different onchain FX designs. Mento's real on Celo trades close to PTAX, with a median premium of 2.7 basis points across 150,229 qualifying swaps. Half of those swaps fall between approximately 18 basis points below and 23 basis points above the benchmark.
The picture changes on pools using BRLA and BRZ, which are issued and redeemed through Brazilian banking channels. On Polygon, BRLA against USDC has a median premium of 39 basis points across 135,341 qualifying swaps. BRZ against USDT has a median premium of 48.5 basis points across 106,910 swaps. Other large pools show similarly elevated prices: 54.4 basis points for BRLA on Gnosis and 56.4 for BRZ on Base.
The difference is not readily explained by pool fees. Bank-redeemed pools charging 4–5 basis points have a median premium of about 44 basis points, while those charging 30 basis points have a median premium of about 57 basis points. Nor does the token alone explain it: BRLA and BRZ trade close to parity against each other.
The gap also persists when the comparison is restricted to hours when PTAX is being set. The bank-redeemed pools remain materially above the benchmark, while the Celo pool stays much closer. At weekends, the premiums on bank-redeemed pools increase further; the same pattern is not evident for Celo.
Together, these results point to the conversion and redemption mechanism as a plausible explanation for the difference between rails. A token whose price depends on an offchain conversion and redemption process may carry costs or risks that an oracle-priced, onchain-collateralized design handles differently. The data are consistent with this interpretation, but do not establish causality.
The onchain dollar premium over the official rate, by rail
Daily median premium on days the official rate was published (dots) and its 28-day persistent level (line)
Daily premium on days the official rate was published. Mento's real on Celo holds PTAX; BRZ on Polygon, redeemed through a bank, holds a level above it that rises in June and July. Argentina's and Colombia's main pools are shown for comparison; the Colombian pool goes quiet after June 2026.
Brazil also shows why a competitive median price is not the whole story. The market spans 18 pools across seven chains, and larger trades tend to receive worse prices than smaller ones. More liquidity and better routing may improve execution around a pool's prevailing price. They may not, by themselves, remove a premium shared across the largest bank-redeemed pools.
Brazil: most swaps price the dollar a little above PTAX, with a long tail higher up
Swaps per 10 bps of premium over PTAX, all 68 pools on 7 chains together
Every Brazilian swap by its premium over PTAX, from all 68 pools with at least one qualifying swap, on seven chains. The median swap pays 32 basis points over PTAX, and about one swap in twenty pays more than 240.
The Brazilian result is not simply that some pools are expensive. It is that different onchain FX rails can produce systematically different prices for the same currency.
Argentina: which dollar is the benchmark?
Argentina makes the comparison more complicated. Across four pools on three chains, the onchain dollar trades at a median 4.8% to 5.5% above the central bank's reference rate. Measured against that benchmark, the premium is substantial.
But Argentina does not have a single dollar price. The official rate coexists with financial-market rates such as MEP and CCL, the informal blue dollar, and the crypto dollar quoted on local exchanges. Against those alternatives, the same onchain price looks quite different.
On days with comparable observations, the median onchain gap is +4.9% against the central bank reference rate, +2.6% against MEP, and +2.2% against the blue dollar. Against CCL, the onchain price is 0.8% lower; against the crypto-dollar quote, it is 0.6% lower.
The onchain price therefore sits closer to CCL and the crypto dollar than to the official rate. Its daily movements also track MEP, CCL, and the blue dollar more closely than the crypto-dollar series, while following official-rate changes only partially and with a delay.
Argentina: the chain prices the crypto dollar, not the official one
Pesos per dollar on days the BCRA published its reference rate: the wARS/USDC.e pool's daily median implied rate beside the official and parallel rates
Pesos per dollar on central-bank publication days: the onchain pool's level sits with the crypto and CCL dollars, well above the official reference rate. The crypto-dollar series starts in late November 2025, after a frozen stretch at its source.
This changes the interpretation of the headline premium. The gap to the official rate is real, but it does not mean the pool is charging a 5% premium over every alternative way to access dollars. Nor does proximity to another benchmark prove that a user can execute more cheaply: the crypto-dollar series is a quoted rate, not a record of completed transactions, and the pool itself has a measurable spread between dollar buyers and sellers.
The sample also calls for caution. The daily onchain observation comes from a single pool on a single chain, and only around 6% of that pool's swaps meet the US$10 threshold used in this analysis. It describes that pool, not the entire Argentine onchain FX market.
Argentina's lesson is that competitiveness depends on the market being compared. An official fixing can be a useful reference, but it is not always the relevant price for a user seeking dollars.
Colombia and Mexico: close to the benchmark is not the same as scalable
Colombia and Mexico show that onchain FX pools can trade close to official reference rates. They also show why a median price cannot be treated as proof of a durable, widely accessible market.
Mento's Colombian peso on Celo has a median premium of 0.7 basis points against TRM, effectively at the official rate. But activity in that pool falls sharply after March 2026 and is nearly absent from July onward.
Other Colombian peso pools on Polygon also trade close to TRM over the period, with median premiums of −3.5 and +12.9 basis points. Their prices fluctuate more from month to month, and their activity changes over time. The data do not establish whether trading moved from one token to another.
In Mexico, the main pools have median premiums within seven basis points of FIX. One pool on Avalanche sits higher, at about 30 basis points. The available record is small, and coverage of the MXNB token is incomplete across some chains.
Colombia and Mexico: swaps cluster around the official rate
Swaps per 10 bps of premium over the official rate, all pools of each currency together; each panel has its own scales
Every Colombian and Mexican swap by its premium over TRM and FIX, all pools of each currency together. Both centre close to the official rate: the median Colombian swap sits 1 basis point above TRM, the median Mexican swap 10 above FIX.
These results show that some pools can maintain prices close to official benchmarks. They do not show that those prices are available for every transaction size, that the pools can absorb larger orders without substantial price impact, or that the same liquidity will remain available over time.
A pool's median is a summary of observed trades, not an executable quote for a particular user. A market can look competitive at the median and still offer poor execution for a larger order, or become inactive altogether.
THE OPPORTUNITY
Better onchain FX requires more than deeper pools
The evidence points to two distinct challenges. The first is the prevailing price: in Brazil, the most active bank-redeemed pools consistently trade above PTAX, regardless of their fee tier. The second is execution: prices can deteriorate as trade sizes increase, smaller pools are more dispersed, and activity can disappear.
Those problems are related, but they are not interchangeable.
Improving conversion and redemption may be the more direct way to address persistent premiums. Brazil's differences across rails suggest that the route between an onchain token and conventional currency markets deserves as much attention as the pool itself. Faster or more flexible redemption, or a design that anchors the price through onchain collateral and an oracle, could reduce some of the constraints. Each approach also introduces costs and risks: oracle-based pricing depends on the quality of the reference and available reserves, while someone must bear the risk of maintaining the price outside local market hours.
Deeper, actively managed liquidity may improve execution. Concentrated liquidity and professional market makers could provide more depth near the prevailing price and reduce the price impact of larger trades. But market makers must manage their own hedging and redemption costs. The higher weekend premiums observed in Brazil are consistent with those costs mattering, although the data do not prove that they are the cause.
Routing can help traders find better available prices, but it cannot eliminate a market-wide premium. Brazil's pools are distributed across chains and tokens, and the smaller pools tend to show greater price dispersion. An aggregator could route trades toward better-priced liquidity. It cannot create liquidity where none exists, and cross-chain routing introduces costs and delays that this study does not measure.
Incentives can attract liquidity, but persistence matters. Colombia and Mexico show how quickly activity can change. The dataset does not record liquidity-incentive programmes, so it cannot establish why particular pools lost activity or whether incentives would have sustained them. The relevant test is whether trading demand and market-making economics can support liquidity after temporary incentives end.
Larger trades may require a hybrid execution model. Above US$10,000, the observed pools increasingly depart from their own retail price. Combining AMMs with professional quoting, request-for-quote systems, or intent-based execution may be more effective for larger orders than relying on a single pool. That would bring professional counterparties back into the execution process while retaining the benefits of onchain settlement.
These are directions to test, not proven solutions. The next step is to measure pool depth and price impact at meaningful trade sizes, compare executable onchain quotes with exchange and OTC alternatives, and include the fees and costs of entering and exiting each market. Without those comparisons, benchmark premiums can describe where prices sit, but cannot establish the cheapest route available to a user.
So, can DeFi build a better FX market for Latin America?
Yes, in specific cases, but the evidence does not yet support a general claim that onchain FX is a cheaper alternative across the region.
Brazil shows that a market can trade close to its official benchmark through one rail while maintaining a persistent premium through others. Argentina shows that the apparent cost depends on which dollar serves as the reference. Colombia and Mexico show that benchmark proximity alone says little about the persistence of liquidity or the prices available for larger trades.
The opportunity is therefore not simply to put local currencies and dollar stablecoins in more pools. It is to build markets in which conversion and redemption mechanisms support competitive prices, liquidity remains available when users need it, and execution holds up at the sizes they trade.
Onchain FX already provides continuous, transparent access to local-currency markets. Whether that infrastructure can translate into consistently better execution remains a question for measured comparison, not an assumption that follows from being onchain.