NASDAQ:DLO DLocal Limited

ISIN: KYG290181018
Information TechnologyEmerging-Markets PaymentsCross-Border Financial InfrastructureHigh-Growth
NASDAQ · Montevideo, Uruguay · Market cap ~US$4.2bn · Reports in USD under IFRS · Beta 0.87 Analysis Status: Donatien Pick
Currency: dLocal trades in US dollars on NASDAQ and reports in US dollars under IFRS, despite earning across Latin America, Africa and Asia. Every figure in this report is in US dollars and no currency conversion is performed anywhere — see §4 for why that makes the standard foreign-currency valuation problem inapplicable here, and §5 for how foreign exchange nonetheless drives the business.
$14.17
−6.4% vs last report ($15.14)
16 Aug 2026 · Signal v6

What changed since 2026-07-31

The quarter arrived, and it split the verdict. dLocal reported Q2 FY2026 after the close on 2026-08-13: total payment volume nearly doubled to US$17.7bn, revenue beat consensus by roughly 8%, and full-year guidance was raised on both volume and gross profit. The shares opened up at US$15.28 — and closed at US$14.17, down 4.0% on 2.58x average volume. The market read past the headline to the take rate, and so does this report.

DISCLAIMER: This is a quantitative framework for educational purposes only. It is not financial advice. Always do your own research and consult a licensed financial advisor before making investment decisions.

DLocal Limited

dLocal is the payments plumbing that lets a global company get paid in places where getting paid is hard. When someone in Brazil, Nigeria, Egypt or Argentina pays for a streaming subscription, a ride-hail trip or a plane ticket, the merchant on the other end is usually a large international business with no local bank licence and no way to accept a Pix transfer, a boleto or a domestic debit card. dLocal provides one integration that reaches over forty emerging markets and hundreds of local payment methods, collects the money locally, converts it, and settles it across the border — and does the same in reverse for payouts. Its genuine advantage is not technology but permission: the portfolio of payment and money-transmission licences it has assembled across dozens of jurisdictions, each slow, costly and politically awkward to obtain. Founded in 2016 and headquartered in Montevideo, Uruguay, it is profitable, generates cash, and carries no meaningful net debt.

HorizonSignalComposite ScoreConfidenceKey Driver
Short-term (1–3 mo)HOLD6662%Buy on confirmation — valuation path open, tape against it
Medium-term (6–12 mo)BUY6962%Attractive anchor + accelerating volume
Long-term (3–5 yr)BUY6962%Licence moat and cash generation, discounted for the take rate
Next update: 2026-08-31 — default +14d (no impactful dated event inside the window; next earnings 2026-11-11). 2026-08-30 is a Sunday, rolled forward to the next trading day.
Table of Contents
1Five-Pillar Scorecard2Hard Gates & Do-Not-Buy Status3Pillar Detail: Business Quality4Pillar Detail: Valuation Attractiveness5Pillar Detail: Underlying Drivers6Pillar Detail: Economic Alignment7Pillar Detail: Entry/Exit Timing8Economic Event Risk9Multi-Timeframe Technical Analysis10Price Chart (6-Month Daily)11Scenario Summary12Entry / Exit Rules13Position Sizing Context14Calibration Snapshot15Data Sources & Methodology
1

Five-Pillar Scorecard

Five independent scores — each 0–100 with its own confidence. The three fundamental pillars (Quality / Valuation / Timing) set the base BUY/HOLD/SELL via the Decision Matrix; the two context pillars (Underlying Drivers, Economic Alignment) then amplify a BUY to STRONG BUY or a SELL to STRONG SELL when both corroborate.

Business Quality

66
Good, marked down on the take rate
conf 72%

Valuation Attractiveness

78
Attractive (0.475× warranted)
conf 70%

Entry/Exit Timing

61
Improving, but only just
conf 62%

Underlying Drivers

70
Tailwind (volume 84 / dollar 45)
conf 65%

Economic Alignment

62
Trend-Following · pressure Neutral
conf 62%
2

Hard Gates & Do-Not-Buy Status

Binary safety checks — any TRIGGERED gate is a hard cap regardless of the scores above; CAUTION gates are sizing notes.
Gate 1 — Financial Distress
Clear. Cash US$794.9m against financial liabilities of US$64.6m at 30 June 2026; interest coverage 8.1x; four consecutive quarters of positive free cash flow. The US$150m facility signed 2026-08-12 leaves the company in net corporate cash.
Gate 2 — Earnings Event Risk
Clear. The binary event has just resolved: Q2 reported 2026-08-13. The next print is 2026-11-11, eighty-seven days out — far outside the 14-day window.
Gate 3 — Valuation Ceiling
Clear on all four arms. Price US$14.17 is below the highest analyst target of US$21.00; the forward multiple sits near decile 3 of its own range, not the top 5%; EV/Revenue is 2.85x against a 20x trigger (struck on the corporate-cash enterprise value of ~US$3.87bn used in §4, not the provider figure); and the anchor band is Attractive (0.475), with the actual 13.15x far below the 33x Information Technology guardrail.
Gate 4 — Accounting / Dilution
Clear. Share-based compensation is 1.6% of revenue against a 25% trigger. The diluted share count is falling, not rising, under the buyback. Non-operating income is 5.7% of trailing net income, well inside the 30% backstop.
⚠️
Gate 5 — Regulatory / Binary Event
CAUTION, not triggered. There is no pending FDA-style decision, takeover or antitrust action, so the gate does not fire. But dLocal operates under payment licences in over forty emerging-market jurisdictions, several with volatile capital-control and FX regimes — this is a chronic elevated risk rather than a dated binary event, and it is why the overall gate state is caution rather than clear.
No gate is triggered and no Do-Not-Buy trigger fires. Two were checked explicitly because they would be easy to get wrong. Trigger 4 (insider selling): director and co-founder Sebastián Kanovich sold 25,700 Class A shares on 2026-07-07 — a single insider, under a pre-arranged Rule 10b5-1 plan, retaining 11.6m convertible Class B shares. The trigger needs three-plus C-suite insiders selling outside automatic plans. It does not fire. Trigger 5 (structural business-model threat): take-rate compression is serious enough to have cut the Quality score, but it is margin erosion, not an existential threat — gross profit still grew 29% and operating profit 15%. It belongs in the Bear case and the exit rules, which is where §11 and §12 put it, not in a hard prohibition.
3

Pillar Detail: Business Quality

What the business is worth owning — scored on gross profit rather than revenue, because for a payments processor revenue includes the rails it buys on the merchant's behalf. Seven sub-signals, a five-dimension moat scorecard derived from named competitors, the sector benchmark, and the ROIC and capital-allocation record.
Business Quality — Pillar Score
A genuinely good business selling its growth at a falling price
66
confidence 72% · High-Growth · moat 60 · benchmark 58

Lifecycle: High-Growth — and an unusually profitable one. dLocal grew revenue 56% year-over-year in the second quarter of 2026 while earning a 16.1% operating margin and converting 125% of net income into adjusted free cash flow. That combination is rare. It is also the reason this report scores the business on gross profit rather than revenue throughout.

Read the gross-profit line, not the revenue line. For a cross-border payments processor, reported revenue includes the cost of the local rails, card schemes and settlement it buys on the merchant's behalf. Gross profit is the analogue of net revenue — the money dLocal actually keeps. The gap between the two is where this quarter's story lives: revenue grew +56%, gross profit grew +29%, and operating profit grew +15%. Scoring the top line would flatter the business by 27 percentage points.
Sub-signalReading (Q2 FY2026, quarter ended 30 Jun 2026)Score
Gross-profit trajectoryUS$127.2m, +29% YoY (+23% constant currency) — a record, and an annualised run-rate above US$500m. But decelerating: +37% FY2025, +38% Q4 FY2025, +40% Q1 FY2026, +29% now.66
Profitability vs peersOperating margin 16.1% (Q2 FY2025: 21.7%). Operating profit over gross profit 50% — down 6pp YoY, but up 6pp on the quarter from 44%.55
Cash generationAdjusted FCF US$68.5m, +41% YoY; conversion of net income 125%. Capex US$9.9m = 2.5% of revenue.85
Balance-sheet healthCash US$794.9m (of which US$369.1m corporate); equity US$540.5m; financial liabilities US$64.6m. Interest coverage 8.1x. Net corporate cash even after the new US$150m facility.82
Take-rate trend (sector-specific)Gross profit over TPV 0.72%, from 0.84% last quarter and 1.07% a year ago. On a full-year basis: 1.15% (FY2024) → 0.99% (FY2025) → 0.72% now. Multi-year, and accelerating.22
Retention (sector-specific)Net revenue retention 153%, the fifth consecutive quarter above 140%. TPV retention 188%. Merchants are expanding, not leaving.92
Geographic diversification (sector-specific)Seven named markets across three continents — Brazil, Mexico, Argentina, Egypt, Nigeria, South Africa, Other Africa & Asia. This quarter Brazil and Argentina were up while Mexico and Africa/Asia were down, which is diversification visibly working. Scored on qualitative disclosure: the quarterly release publishes no revenue-by-country split.70
INDUSTRY BENCHMARK: EM Payments — TPV Growth + Take-Rate Stability
TPV growth: +92% YoY (exceptional — the fastest since Q1 2022, seventh straight quarter above 50%)
Take-rate stability: failing — gross-profit take rate down 33% year-over-year
Rating: ONE STRONG (the framework's 50–84 band, not the 85–100 "both strong" band)
Benchmark Score: 58/100
Context: this is the single most diagnostic pairing in the sector, and it is deliberately a pair. Volume growth at a collapsing take rate is not growth — it is share bought with margin. dLocal is scoring maximum marks on one half and close to zero on the other.

Competitive Moat Scorecard

Pricing power
32
Derived, not asserted: the company's own FY2026 guidance note concedes "some structural volume-based discounting expected", and this quarter names Mexico's "large Tier 0 merchants hitting higher volume pricing tier". Take rate down 33% YoY is the arithmetic of that concession.
Network effects
62
Genuinely two-sided — one integration reaches 40-plus countries and hundreds of local payment methods, and each new method makes the platform worth more to merchants. But the effect is not strong enough to defend price at the top of the merchant pyramid.
Switching costs
76
The hardest evidence in the report: NRR 153% and TPV retention 188%, five straight quarters above 140%. Held below 90 because retention is measured in volume, and some of that volume is retained by conceding price — so it partly overstates the lock-in.
Cost advantage
56
Licence and settlement scale is real, but cost of services grew +73% against revenue's +56% this quarter — unit costs rising faster than the top line. Nigeria carried a one-off cost increase and Mexico "cost pressure".
Intangible assets
72
The strongest and most durable asset: payment and money-transmission authorisations across 40-plus emerging-market jurisdictions. Slow, expensive and politically awkward to replicate — this is what a well-funded incumbent cannot simply buy.

Moat score: 60/100 — the average of the five. The shape matters more than the number: a strong licence-and-retention moat wrapped around a pricing layer that is visibly eroding.

Competitive Environment

Share trajectory: gaining volume, conceding price. Threat level: elevated.
This is the hypothesis the numbers have to answer, so it is scored from evidence rather than assertion. dLocal is winning volume — TPV nearly doubled and NRR is 153%, which is not what losing share looks like. But it is winning it at a falling unit price, and the company says so itself. The honest label is not last quarter's "gaining"; it is gaining volume while conceding price.

Management frames the discounting as aggregation theory — scale earns better FX liquidity and conversion data, which wins more volume, which earns more scale. That is a coherent flywheel if the take rate settles. It is commoditisation if it does not. The distinction is testable within two quarters, and §12 sets the test.
RivalWhere it pressesShare / moat-erosion vector
AdyenThe largest global merchants — the same Tier 0 accounts named in Mexico this quarterThe most direct threat to pricing. A single-platform incumbent with its own licences, able to price aggressively into LatAm to win a global merchant's whole wallet. This is the most credible explanation for Tier 0 merchants reaching "higher volume pricing tiers".
StripeGlobal merchants wanting one contract, plus developer-led mid-market Erodes the integration-convenience half of the network effect rather than the licence half. Deepening its own local acquiring reduces the number of countries where dLocal is the only practical route in.
EBANXHead-to-head in LatAm and increasingly Africa/Asia — the closest like-for-like competitorSame model, same regions, private and able to buy share. The clearest source of direct price competition in dLocal's core markets.
NuveiCross-border and alternative payment methodsUnder private ownership since 2024 and less price-disciplined as a result; overlaps in APM coverage.
Local acquirers (StoneCo, PagSeguro, Rede/Cielo in Brazil; Clip in Mexico)The local-to-local flows that are now a rising share of dLocal's mixStructurally the cheapest operators in their home market. Local-to-local carries a lower take rate precisely because these firms set the price there — this is the single largest mechanical driver of the mix shift.
Merchants' own rails (Uber, Netflix, Amazon, Booking)Insourcing at scale The tail risk. A merchant large enough to obtain its own local licences can bypass dLocal entirely. NRR 153% says none has yet; the Tier 0 pricing concession is what keeps it that way.

How this fed the sub-scores. The Tier 0 repricing and the local-to-local mix shift are the direct inputs to Pricing power 32. Cost of services outgrowing revenue by 17 percentage points, plus the named Nigeria and Mexico cost items, set Cost advantage 56. NRR 153% / TPV retention 188% — quantified, audited, five quarters running — set Switching costs 76, discounted from higher because the retention is partly purchased.

ROIC & Capital Allocation

ComponentReadingScore
ROIC (40%)Asset-light and net-cash: TTM operating profit US$235.2m, ROE 41.3%, ROA 8.6%, capex 2.5% of revenue. Invested capital is negligible, so ROIC sits far into the top quartile of any peer set.88
Capital allocation (30%)Of the US$300m buyback authorised in March 2026, US$86.1m was spent on ~6.9m Class A shares to 30 June — an average near US$12.5, below today's price. Annual dividend US$0.1967 (28.9% of TTM EPS). The new US$150m facility on a net-cash balance sheet is the one debatable call.75
Skin in the game (30%)Founder-linked ownership persists: director and co-founder Sebastián Kanovich retains 11.6m convertible Class B shares. Share-based compensation is US$6.5m in the quarter — 1.6% of revenue, exceptionally low for the sector. Dual-class structure is the governance debit.68

Composite: 78/100. Capital discipline is a genuine strength here and is doing real work holding the Quality pillar at 66 while the take-rate sub-signal drags at 22.

Earnings quality (step 7b) — checked, and clean. Non-operating items across the trailing twelve months (net finance result plus the IAS 29 Argentina hyperinflation adjustment) total US$11.6m against US$204.1m of net income — 5.7%, well inside the 15% threshold, so no normalisation is required and the clean multiple equals the reported one. Two notes for the record. First, FMP reports nonOperatingIncomeExcludingInterest of +US$22.5m for Q4 FY2025, which would have been material — it is a line-item mapping artefact for an IFRS filer; the company's own statement shows the gap between operating profit and pre-tax profit that quarter was only +US$1.4m. Second, the distortion in this window runs the other way: Q1 FY2026 net income was reduced by US$9.7m of prior-year (2023–2025) tax adjustments, so trailing EPS of US$0.68 understates the underlying run-rate.
4

Pillar Detail: Valuation Attractiveness

Whether the price is right — a computed intrinsic anchor first (a warranted multiple built from interest rates, disciplined growth and sector risk), then relative lenses that order the name within the band the anchor sets but can never override it.
Valuation Attractiveness — Pillar Score
Attractive on every basis tested — 13.15× against a warranted 27.7×
78
confidence 70% · band Attractive · ratio 0.475 · basis forward P/E

The anchor, in one line: at US$14.17 the shares trade on 13.15x forward earnings against a rate-and-growth-warranted 27.7x — a ratio of 0.475, deep inside the Attractive band. The valuation pillar is the reason this name still carries two BUY signals after a quarter that cut the Quality score.

Why forward P/E and not EV/Revenue — this choice matters. The obvious multiple for a high-growth payments name is EV/Revenue (2.85x here on the corporate-cash enterprise value used throughout this report, or 2.4x on the provider figure — superficially cheap either way). It is the wrong anchor for this business at this moment, and dangerously so: revenue grows faster than gross profit precisely because the take rate is compressing, so EV/Revenue mechanically improves as the underlying economics deteriorate. It would reward the exact thing the report is warning about. dLocal is solidly profitable with 125% cash conversion, so earnings resolve cleanly and forward P/E is both available and honest. It also keeps the sector-guardrail test live — a non-P/E basis silently disables it.

Which guardrail line — and the sector inconsistency it exposes. Scored against Information Technology (33x), matching the data providers' classification of dLocal as software infrastructure. The arguable alternative — capital-light financials, as one would treat a card network — is 30x. At an actual 13.15x the name clears both by a wide margin, so the choice changes nothing.

It is stated because a silent choice is an audit finding, and because this report is genuinely inconsistent about dLocal's sector: the valuation guardrail uses Information Technology, while §7's relative strength is measured against XLF and §6 reads the macro row for Financials. That ambiguity is real rather than careless — dLocal is a payments processor with a technology cost structure and a financial-services regulatory perimeter. The rule applied throughout: follow the macro report's mapping (Financials) for anything macro or benchmark-relative, and the data providers' mapping (Information Technology) for the valuation guardrail. Neither choice is binding — the guardrail clears at 13.15x on both lines — and the stored relative_strength_vs_sector now names XLF explicitly so the benchmark is never ambiguous.

The Warranted-Multiple Anchor

InputValueSource / discipline
Risk-free rate4.63%US 10-year Treasury, FRED DGS10 at 2026-08-13. Attribution matters here: this is the run's shared macro input, not a figure lifted from the macro report — that report is dated 2026-08-12, cites the 10-year at 4.70%, and carries no market-snapshot field. On 4.70% the discount rate would be 9.20%, the warranted multiple 27.4x and the ratio 0.480 — still Attractive, so nothing turns on it.
Equity risk premium4.50%Framework constant.
Risk add-on+0.00%Business Quality 66 clears the 65 threshold.
Discount rate r9.13%4.63 + 4.50 + 0.00.
g (years 1–5)15.0%Consensus forward growth near 27.5% × 0.75 = 20.6%, then capped at the 15% secular-growth ceiling. The haircut and the cap are both binding — this is deliberately not the headline growth rate.
g (terminal)3.0%Long-run nominal GDP ceiling. dLocal reports in US dollars, so the US- calibrated terminal rate is the internally consistent choice (see the currency note below).
Warranted forward P/E27.7xTwo-stage discounting; below the 33x Information Technology guardrail, so uncapped.
Actual forward P/E13.15xUS$14.17 ÷ FY2027 consensus diluted EPS of US$1.08 (n = 4).
Ratio → band0.475 → AttractiveThreshold for Attractive is 0.80.
Currency frame — one frame, stated plainly. dLocal earns across Brazil, Mexico, Argentina, Nigeria and a dozen other emerging markets, but it reports in US dollars under IFRS and trades in US dollars on NASDAQ. So the ordinary foreign-currency-issuer problem — pairing a US-calibrated terminal growth rate with a local discount rate, a units error — does not arise here. Every input above sits in one USD frame: US risk-free, US equity risk premium, US-anchored terminal growth, USD-reported earnings. No conversion is performed anywhere in this report and therefore no FX rate is stored. For dLocal, foreign exchange is an earnings exposure, not a units problem — and §5 treats it as exactly that.
The band does not depend on the basis — three ways of striking it. The forward estimate above comes from a panel that predates the 2026-08-13 print, which is a fair objection. It does not change the answer:
• FY2027 consensus EPS US$1.08 → 13.15x → ratio 0.475
• Blended next-twelve-months EPS ~US$0.97 → 14.6x → ratio 0.53
• FY2026 consensus EPS US$0.82 → 17.3x → ratio 0.62
All three land in the Attractive band. A further stress — cutting the FY2027 estimate 15% to US$0.92, which is roughly what a failed second-half operating-leverage step-up would do — gives 15.4x and a ratio of 0.56. Still Attractive. The valuation conclusion survives every version of the pessimism.

Relative Cross-Checks (they order within the band; they never override it)

LensWeightReadingScore
Warranted anchor40%Ratio 0.475 — deep in the Attractive band.88
Sector median20%13.2x sits above LatAm domestic processors (StoneCo, PagSeguro near 7–9x) and well below Adyen (~35x) and the global platforms. dLocal grows several times faster than the cheap comparables — but the cheap comparables are cheap for a reason that also applies here.62
Own-history decile15%Roughly decile 3 of its own post-IPO range (60x-plus in 2021, 9–10x at the 2025 lows). The 52-week position is 57% of the 10.64–16.78 range.72
PEG10%13.15x against consensus FY2026→FY2027 EPS growth of 31.3% = 0.42.88
Analyst consensus15%US$14.17 vs a mean target of US$18.35 (n = 10; high 21.00, low 14.50) = 22.8% below. Discounted from the 85–100 band because no target has been issued in the last month — the panel has not yet responded to the print.72

Valuation Attractiveness: 78/100 — unchanged from the 2026-07-31 report. Not because nothing happened, but because two things offset: the price fell 6.4% while the forward earnings basis was struck slightly lower.

Cash yield — and which cash the enterprise value nets. Two adjustments are needed before this number means anything, and both cut the same way.

On the numerator: statutory free cash flow looks like ~US$400m because operating cash flow sweeps in merchant float. The company's adjusted measure, which strips merchant working capital out, is the honest one and is used here: US$68.5m in the quarter, an annualised run-rate of ~US$274m.
On the denominator: the provider enterprise value of ~US$3.28bn nets total cash of US$794.9m — but most of that is merchant float held against US$1,555m of trade payables, and it is not dLocal's money to spend. §3 already separates the US$369.1m of corporate cash; consistency requires using it here too. On that basis EV is ~US$3.87bn.

The honest yield is therefore 7.1% on the Q2 run-rate, not 8.4%. The first half as reported was weaker (US$83.2m, depressed by working-capital timing the company flagged and has since reversed), annualising to 4.3%. The true figure sits between the two and comfortably in the attractive range — but it is a percentage point lower than the provider's EV would flatter it to be.

Embedded Optionality — what you get for nothing

Remaining buyback authorisation (~US$214m). Roughly 5% of the market capitalisation, live, and being executed below today's price. The market is paying nothing for the shrinkage.
dLocal for Platforms and pay-outs. Embedded-payments and disbursement products bundled inside one consolidated multiple; both scale on the same licence base with near-zero incremental capital.
Africa and Asia. Still small enough that this quarter's shortfall there was a rounding item, but built on the same licence moat that took years to assemble in LatAm.
Merchant float. US$795m of cash generating finance income, currently shrinking as rates and mix move — but a genuine call option should EM policy rates back up.
Effect on the pillar: a tilt, not a re-rating. None of this is why the name is cheap; it is why the downside is cushioned. No points were added to the score for it.

Implied-growth read — and the honest caveat that goes with it. Solving the same two-stage model backwards for the growth rate that justifies 13.15x, holding r at 9.13% and terminal growth at 3%, gives roughly −2.5%. At US$14.17 the market is not pricing slower growth than we assume; it is pricing shrinking earnings — against a company that just raised full-year guidance on two of three metrics. For scale: even plugging in zero five-year growth warrants 14.7x, still above where the shares trade.

The caveat, stated rather than buried: that gap is wide partly because the framework's 3% terminal growth rate puts a high floor under any warranted multiple — a large share of the 27.7x sits in the terminal value, so the model is structurally generous to a profitable business at almost any near-term growth rate. This is precisely why the relative cross-checks carry 60% of the pillar and the anchor only 40%, and why the fair-value estimate below is struck at US$17.45 from peer, history and consensus references rather than at the ~US$29.8 the anchor alone implies. The gap is the investment case — but §11's Bear is the discipline that says it could close from the wrong end.

5

Pillar Detail: Underlying Drivers

The external forces that sit above management's control — here, emerging-market payment volume and the US dollar. Each scored across history, the present and the forward view, then combined and mapped to its amplification role.
Primary Driver
EM payment volume (65%) + the US dollar against EM currencies (35%)
70
Tailwind — eligible to amplify, but not applied

dLocal's fortunes sit above its own execution in two places, and this quarter they pulled in opposite directions. Both are scored; neither is allowed to hide the other.

Driver 1 (65% weight) — Emerging-market cross-border payment volume

HorizonReadingScore
Historical (25%)Seven consecutive quarters of TPV growth above 50%, and growth has accelerated in each of the last five. Q2 FY2026 processed more volume than the whole of 2023.90
Current (50%)TPV US$17.7bn, +92% YoY (+80% constant currency), the fastest since Q1 2022. Full-year TPV guidance raised to 60–70% from 50–60%, and gross-profit guidance raised to 25–30% from 22.5–27.5%.88
Forward (25%)Management's own caution, quoted: growth "will naturally create more demanding comparisons as we move through the second half of the year and into 2027". Underlying EM digital-commerce growth is 15–20%, so the current rate is a share-gain rate, not a market rate — and share gains at these magnitudes do not annualise.70

Driver 1 score: 84 — Strong Tailwind.

Driver 2 (35% weight) — The US dollar against emerging-market currencies

For dLocal, foreign exchange is not incidental. It is the business — through two different channels that must not be conflated.

Channel A — translation. dLocal earns in Brazilian real, Argentine peso, Mexican peso, Nigerian naira and a dozen others, and reports in dollars. A soft dollar inflates reported growth. This quarter it did: reported TPV growth was +92% against +80% in constant currency, revenue +56% against +50%, gross profit +29% against +23%. Translation was worth roughly 6 percentage points of the reported gross-profit growth. It was a tailwind — which matters enormously for the next paragraph.

Channel B — FX spread revenue. Part of dLocal's take rate is the spread it earns converting currency. That spread is widest where currencies are least liquid and most volatile. This quarter the company attributed part of the miss to "lower contribution from higher FX spread markets (Mozambique and Vietnam)" — calmer conditions in those markets directly reduced revenue quality. This is a genuine economic headwind, not an accounting one.
The decomposition the take rate demands — and the finding that surprised us. Gross profit over TPV fell from 1.07% to 0.72%. The temptation is to write that off as currency noise. The numbers forbid it:

(a) Translation FX — a TAILWIND, not a headwind. Reported growth exceeded constant-currency growth on every line. Currency helped the reported figures. It therefore cannot explain a decline; strip it out and the compression is worse, not better.
(b) FX-spread mix — a modest headwind. Lower contribution from Mozambique and Vietnam, plus a one-off Nigerian cost increase. Real, named by the company, and worth part of the gap.
(c) Genuine pricing and mix compression — the dominant cause. A rising share of local-to-local flows, which structurally carry a lower take rate; and Mexican Tier 0 merchants "hitting higher volume pricing tier". The company's own FY2026 guidance had already flagged "structural volume-based discounting expected".

Two figures from the earnings call sharpen this, and they point opposite ways — both belong here.
For cause (c): local-to-local flows reached 61% of TPV, up six percentage points from the first quarter. That is the mix shift, quantified, in a single quarter — and local-to-local structurally carries the lower take rate. It is the strongest single piece of evidence for the pricing-and-mix explanation.
Against it: management stated that backing out one very large ride-hailing merchant's mix of gains at a lower take rate, the take rate would have been relatively flat sequentially. If that is right, a large part of the quarter-on-quarter drop is one merchant ramping rather than broad price erosion — a materially more benign reading, and the core of the bull case in §11.

Verdict: still mostly (c), but held with less certainty than the headline invites. The single-merchant explanation is a company assertion we cannot audit, and it does not touch the multi-year series — 1.15% (FY2024) → 0.99% (FY2025) → 0.72% is not one merchant. What it does do is make the sequential step look worse than the trend, which is why the Quality mark-down was six points and not more. The compression is real pricing, not currency translation, and anyone attributing it to FX has the sign backwards.
HorizonReadingScore
Historical (25%)A soft dollar through the first half of 2026 was a consistent tailwind to reported results.70
Current (50%)The dollar firmed in August as Fed cuts were priced out — the macro report of 2026-08-12 names this as precisely the reason it trimmed DLO's short-horizon signal to Neutral. The translation tailwind that flattered the second quarter is fading as the third quarter runs.38
Forward (25%)With a September hike carrying roughly 44% market-implied odds and three hawkish July dissents, the risk skews to further dollar strength — a direct drag on reported growth.35

Driver 2 score: 45 — mild Headwind.

Combined Underlying Driver: 84 × 0.65 + 45 × 0.35 = 70/100 — Tailwind. At 70 the driver is eligible to amplify a base BUY to STRONG BUY — but amplification requires Economic Alignment's pressure to read Tailwind as well, and it reads Neutral (§6). No amplification is applied. The base signals stand as the Decision Matrix set them, and the driver score does not alter any pillar score.
6

Pillar Detail: Economic Alignment

How the current economic regime bears on this specific name, taken from the macro report of 2026-08-12 and using DLO's own watchlist signal in preference to its sector row.
Stance · Pressure
Trend-Following · Neutral
62
conviction

DLO carries its own entry in the macro report's watchlist forecast — short Neutral / medium Outperform / long Outperform — and that is what is used here, in preference to the sector row. Where it differs from the sector map, and why that matters: the macro report classifies DLO under Financials (XLF), whose sector signals are Neutral / Underperform / Neutral. The name-specific signal is better than its sector on both the medium and long horizons — the sector row is carrying US regional-bank credit and rate exposure that has nothing to do with an emerging-market payments processor. Using the XLF row would have imported a headwind DLO does not face. It also arguably sits closer to Information Technology (Neutral / Neutral / Outperform), which is how the data providers classify it; the name-specific read straddles the two sensibly.

The macro report's stated reason is worth quoting because it is the same mechanism §5 scores: "The soft-dollar tailwind is what powers this name, and the dollar just firmed as cuts came off the table — hence the short trim. The EM-payments volume story is structural and survives the middle and long horizons intact."

Pressure: Neutral — the near-term dollar move is a genuine headwind while the structural volume story is a tailwind, and honestly scored they offset. This is the reason the base BUY on medium and long is not amplified to STRONG BUY despite an Underlying Driver score of 70, which on its own would qualify. Amplification requires both conditions; only one is met. The dominant regime — energy-shock stagflation, with Fed cuts priced out and a live hike-versus-hold debate — reaches dLocal almost entirely through the dollar rather than through demand.

Source: watchlist-forecast (name-specific) · Macro report 2026-08-12

7

Pillar Detail: Entry/Exit Timing

The risk-reward framework, relative strength vs SPY and the sector ETF, the macro overlay, news-derived sentiment, and the catalyst cluster.
Entry/Exit Timing — Pillar Score
Structure intact, momentum broken — the print was sold, not bought
61
confidence 62% · MTF 66 · risk-reward 62 · RSI 46.3 · ATR 0.60

The tape did something specific on 2026-08-14 that the timing pillar has to take seriously, and it is worth stating before any indicator: the stock opened at US$15.28, traded down to US$13.995, and closed at US$14.17 — a full outside-day reversal on 5.07m shares, 2.58x the twenty-day average. The market gapped the print up, then sold every share of it. That is distribution, and it is the single most important input below.

ComponentWeightReadingScore
Multi-timeframe trend30%Monthly and weekly uptrend, daily technically a strong uptrend (14.17 > 50-day 13.85 > 200-day 13.37) — but hourly and 15-minute have rolled over hard. Weighted confluence 66.66
Risk-reward20%Favourable stop geometry: the 13.81–13.90 shelf (three swing lows, the 50-day SMA at 13.85 and the lower Bollinger band at 13.83 all cluster there) sits just 0.6 ATR below the price, so a tight stop is available. Offset by the setup being a fresh distribution day rather than a bounce.62
Macro15%Fed cuts priced out, a September hike near 44% implied, and a firming dollar working directly against dLocal's reported growth. The macro report's own DLO short-horizon signal is Neutral.45
Sentiment18%No rating change in thirty days — there were two actions, both maintains: Goldman Sachs held Buy on 2026-07-31 and Truist held Buy on 2026-07-24. The last actual change was UBS’s upgrade to Buy on 2026-07-01. The panel scored is FMP’s get_grades_consensus aggregate — 9 Buy / 3 Hold / 1 Sell, i.e. 69% bullish. That single Sell could not be corroborated: none of FMP’s own fifteen most recent individual grade actions is a Sell, and Yahoo’s panel (n = 10, recommendation mean 1.50) carries none either. It is kept because it is the framework’s specified endpoint and discarding it would flatter the score — but it is flagged as unattributed, and on either alternative panel this sub-signal would score higher. Against all of that, the print beat revenue by ~8% and missed EPS (US$0.18 vs US$0.19), and the tape’s answer was −4.0%.48
Catalysts17%Calm. The binary event just cleared; the next earnings date is 2026-11-11, eighty-seven days out. Nothing company-specific inside thirty days.78

Entry/Exit Timing: 61/100 — "Improving" by the framework's threshold (≥55), but only just, and the composition matters more than the number: it is carried by a calm calendar and a favourable stop, not by momentum.

Relative strength — a tale of two windows.
One month: DLO −2.1% vs SPY +4.5% and XLF +3.4% — underperforming both, by 6.6 and 5.5 percentage points.
Three months: DLO +28.7% vs SPY +5.0% and XLF +13.8% — outperforming both, decisively.
The three-month figure is real but flattered by its starting point: the stock bottomed at US$10.64 on 2026-05-18 after the Q1 print, the low point of a 20% single-week collapse (15 May's low was 10.975). The one-month window is the one that speaks to current timing, and it says the leadership has stalled.
Why the Technical entry group is unmet — the honest reading. A mechanical reading could claim the breakout branch fired: the close of 14.17 is above the 50-day SMA of 13.85, and volume was 2.58x average. That reading is wrong and this report will not take it. The condition exists to capture a trend turning; DLO has closed above its 50-day SMA every session this month, so nothing crossed. What actually happened is that heavy volume accompanied a 4% decline from a higher open — the same arithmetic, the opposite meaning. The MACD sub-condition then settles it independently: the histogram was last positive on 2026-07-17 and has been negative for 20 consecutive sessions since 2026-07-20 (−0.135 on 2026-08-14), with the MACD line below its signal throughout. The dates matter more than the count: the histogram was already negative on 2026-07-31, the day our previous report was written, and already negative on 2026-08-13, the day of the print. The earnings reaction confirmed the loss of momentum; it did not cause it. a higher low cannot be claimed one session into a reversal. Technical: unmet. All dates and levels in this section are taken from get_technical_indicators and Yahoo daily bars, never from get_stock_prices labels (see §15).
8

Economic Event Risk

High-impact macro releases in the next 14 days that could swing this stock, plus the last 7 days of surprises.

Upcoming events (next 30 days)

DateEventImpactForecastPreviousRelevant?Why
2026-08-19FOMC MinutesHigh✅ YesThe dominant driver of the dollar, and the dollar is dLocal's translation lever. A hawkish read strengthens the USD and mechanically trims reported growth.
2026-08-25CB Consumer Confidence (Aug)High90.8⚠️ LowUS consumer sentiment barely touches dLocal — its consumers are Brazilian, Mexican and Nigerian.
2026-08-26Core PCE Price Index MoM (Jul)High0.3%0.1%✅ YesFeeds the September hike odds directly, and therefore the dollar.
2026-09-04Non-Farm Payrolls / Unemployment (Aug)High12K / 4.2%−23K / 4.1%✅ YesThe other half of the Fed's mandate and the other half of the rate path.
2026-09-11CPI (Aug)High3.4% YoY✅ YesSame transmission: inflation → rate path → dollar → dLocal's reported growth.

Recent surprises (last 7 days)

DateEventActualForecastSurpriseImpact
2026-08-12CPI YoY (Jul)3.4%3.4%In lineNeutral. Removed a downside catalyst for the dollar without adding an upside one.
2026-08-13Producer Price Index MoM (Jul)0.0%0.2%−0.2ppMildly dollar-negative, i.e. mildly positive for dLocal's translation.
2026-08-14Retail Sales MoM (Jul)−0.6%+0.1%−0.7ppLarge miss — but see the note below. Not a dLocal demand signal.
2026-08-14Michigan Consumer Sentiment (Aug)51.054.5−3.5Miss, from 55.2 in July. Again: a US consumer signal, and dLocal has no US consumer.

The 14 August US consumer break is not a dLocal event — and saying so is the discipline, not a dodge. July retail sales fell 0.6% against a consensus near +0.1%, and Michigan sentiment dropped to 51.0 from 55.2, both attributed to energy-driven cost-of-living fear. That pair cuts hard at anything needing a resilient US consumer. dLocal has none: it processes payments for global merchants from emerging-market consumers, and its named growth engines this quarter were Brazilian ride-hailing and travel and Argentine e-commerce and delivery. Marking the name down for a US retail-sales miss would be a category error.

What genuinely matters on this calendar is the rate path, and only because of what it does to the dollar. The FOMC minutes on 19 August and core PCE on 26 August both feed September hike odds, currently near 44% implied. A firmer dollar shrinks the translation tailwind that added roughly 6 percentage points to reported gross-profit growth this quarter. That is the transmission line to watch — and it is already why the macro report trimmed DLO's short-horizon signal to Neutral. As a medium-macro-sensitivity fintech with no high-impact event inside three days, no additional timing-confidence penalty is applied.

9

Multi-Timeframe Technical Analysis

Trend, RSI and breakout status across monthly / weekly / daily / hourly / 15-minute, with a confluence verdict.
TimeframeTrendDirectionRSIMACDKey S/RBreakoutVol
MonthlyUptrend ↑Bullish47.3+0.58 hist, risingS: 10.64 R: 16.78Resistance breakout0.6x
WeeklyUptrend ↑Bullish54.3+0.19 hist, risingS: 12.16 R: 15.51Resistance breakout1.2x
DailyStrong uptrend ↑Neutral46.3−0.13 hist, fallingS: 13.83 R: 14.95None2.6x
HourlyStrong downtrend ↓Bearish46.6−0.04 hist, fallingS: 13.99 R: 14.43None
15-minDowntrend ↓Bearish48.4−0.02 hist, turningS: 14.03 R: 14.43Support breakdown
Confluence: Mostly Bullish (65–79 band) · MTF Score 66

The higher timeframes and the lower timeframes are telling different stories, and the split is exactly where the print landed. Monthly and weekly remain in uptrends and the daily chart still satisfies the textbook definition of a strong uptrend — price 14.17 above the 50-day 13.85 above the 200-day 13.37. But the daily MACD histogram is negative and falling, price is below both the 20-day SMA (14.67) and the 20-day EMA (14.54), and the intraday frames have broken down outright. This is the classic shape of a healthy intermediate trend absorbing a sharp negative shock: the structure is intact, the momentum is not. That is why the confluence label reads Mostly Bullish on a score of 66 — the framework's 65–79 band — even though nothing about Friday's tape felt bullish: the score is carried almost entirely by the monthly and weekly frames, which together hold 55% of the weight and have not yet registered the print. The level that decides it is the 13.81–13.90 shelf — three swing lows, the 50-day SMA and the lower Bollinger band all converge there, and Friday's low of 13.995 probed toward it without closing through. Hold that shelf and the pullback is a buyable dip inside a live uptrend; close below it for two sessions and the whole daily structure changes character.

10

Price Chart (6-Month Daily)

A 6-month daily close line with SMA50 and key support/resistance — the visual companion to the MTF table.

DLO daily closes, last 80 sessions to 2026-08-14, with the 50-day SMA. Raw unadjusted closes from Yahoo daily bars. The two vertical facts of this chart are the $10.64 low on 18 May 2026 — the trough after the Q1 print — and the 14 August reversal — a $15.28 open, a $13.995 low and a $14.17 close on 2.58× average volume.

11

Scenario Summary

Three futures with explicit probabilities summing to one hundred, each priced from a stated multiple and a stated earnings figure so the arithmetic can be checked and disagreed with.

Bull $21.00 (25%)

The operating leverage arrives and the take rate finds its floor. Management said second-half operating leverage would improve, and the mechanism is credible: opex already fell 4% sequentially, the World Cup marketing spend was front-loaded into the first half, and the Mexican Tier 0 repricing annualises out of the comparison. Operating profit reaches the upper half of the maintained 27.5–32.5% guide, the gross-profit take rate stabilises around 0.70–0.72% as the local-to-local mix matures, and FY2027 EPS beats consensus at roughly US$1.20. On proof that near-doubling volume converts to profit, the multiple re-rates to ~17.5x — still barely half the warranted 27.7x. 17.5 × 1.20 = US$21.00, +48%. What confirms it: the Q3 print on 2026-11-11 showing gross profit over TPV flat or up sequentially.

Base $17.45 (50%)

Volume keeps compounding; the market keeps discounting the unit economics. FY2027 EPS lands near the US$1.08 consensus. The take rate keeps drifting — not collapsing — toward 0.68–0.70% as local-to-local grows. The market credits an exceptional volume franchise and refuses to pay a growth multiple for it while the take rate is falling, holding the shares around 16x. 16.2 × US$1.077 = US$17.45, +23%, which is also where the blended fair-value estimate lands. Note what this base case concedes: a 23% return over twelve months from a business growing volume at 92% is a de-rating in slow motion. That is the honest central case.

Bear $10.75 (25%)

The second-half step-up fails and November brings a guidance cut. This is the scenario the arithmetic makes uncomfortably live. FY2025 operating profit was US$220m; the maintained 27.5–32.5% guide implies US$280–292m for FY2026. The first half delivered US$116.9m, up only 15%. That leaves US$164–175m for the second half against US$118.4m a year earlier — +38% to +47% growth, from a first half that managed +15%. Basis: reported to reported, which is the right frame because the guidance is struck against reported FY2025. Adding back the US$4.4m prior-year tax item that sat in Q1 FY2026 operating expenses eases the bar to +34–44% — lower, and still a long way from +15%. The argument survives either basis, which is why it is worth stating which one is used. If that fails, guidance is cut on 2026-11-11, the take rate slides toward 0.60% as Adyen and EBANX press on price and local-to-local keeps rising, FY2027 EPS resets to ~US$0.86 and the multiple de-rates to 12.5x. 12.5 × 0.86 = US$10.75, −24%. This is not a theoretical level: the stock traded at US$10.64 three months ago, on 18 May 2026, in the wake of the Q1 print — a 13.0% single-session fall on 15 May inside a 19.6% week. A repeat is a precedent, not a projection.

On the market-wide tail: the macro report carries an armed 'S&P 500 concentration / AI earnings-quality unwind' risk, with the trigger currently receding as breadth broadens. dLocal does not inherit it. It is not levered to AI capex or monetisation, is not an index top-weight, and its earnings carry no material non-operating mark-to-market (5.7% of net income). Firing a cohort de-rating leg here would be over-reading the rule; the bear above is entirely idiosyncratic.

Probability-weighted value: US$16.66 (0.25 × 21.00 + 0.50 × 17.45 + 0.25 × 10.75), or +17.6% against US$14.17. Worth noting what the distribution actually looks like: the upside and downside are close to symmetric in magnitude (+48% versus −24%) but the bear has a live, dated trigger on 2026-11-11 while the bull needs two things to go right. That asymmetry in timing, not in size, is why the short horizon is a HOLD and the ladder reads Half-Size.

12

Entry / Exit Rules

Three independent entry paths (Fundamental · Technical · Catalyst) and three exit triggers (Stop-Loss · Thesis · Profit-Target). Any one entry path is a valid entry — the more that agree, the larger the position the conviction ladder suggests. Exits are graded by severity, not count.

How to read this — the Conviction Ladder

The three entry groups are alternative paths to a buy, not a checklist. A group counts only when all its sub-conditions hold. How many groups are satisfied sets the suggested size — it does not gate whether you may enter: 1 group = Half-Size (a valid starter/scale-in), 2 = Full-Size, 3 = Over-Size (highest conviction); 0 = Wait (no path open yet). A strong overall signal can still read Wait here when the stock is well above its entry zones — that flags "good business, no entry edge right now," not a contradiction. Exits are graded by severity of what is live, not by a count: a hard stop is an Exit on its own.
Entry conviction: Half-Size1 of 3 groups met — one path open — starter / scale-in

Fundamental — MET

Met — the valuation path is open.
✅ Price US$14.17 below the fair-value estimate of US$17.45
✅ No earnings within 7 calendar days (Q2 reported 2026-08-13; next 2026-11-11)
✅ Underlying-Driver score 70 ≥ 50

Technical — not MET

Unmet — momentum is against, and the volume was distribution.
⛔ A close above the 50-day SMA on >1.5x volume that constitutes an actual cross — price has been above the 50-day all month, and Friday's 2.58x volume accompanied a −4.0% reversal from a higher open, not a breakout
✅ RSI 14 between 35 and 65 — reading 46.3
⛔ MACD histogram positive two consecutive days, or turning up off support — last positive 2026-07-17, negative for 20 consecutive sessions since 2026-07-20 (−0.135 on 2026-08-14), MACD line below signal throughout

Catalyst — not MET

Unmet — decisively. The event happened and went the wrong way.
⛔ Post-earnings move within 24h above +5% — actual −4.0%
✅ Guidance raised or maintained — TPV raised to 60–70%, gross profit raised to 25–30%, operating profit maintained
✅ Volume above 2x the 20-day average — 5.07m, 2.58x

Forecast: Technical group — the binding constraint. Two sub-conditions must flip. The MACD histogram is at −0.135 and has flattened over the last three sessions (−0.156 → −0.122 → −0.135); at that rate a cross back above zero needs roughly 2–4 weeks and requires price to reclaim the 20-day SMA at 14.67, 3.5% above. Basis: the 20-day SMA is falling ~0.02/day while the 50-day rises ~0.05/day, so the two converge toward ~14.4 by early September — a reclaim becomes progressively easier without price doing much. Confidence: Moderate. The pullback branch is the faster path: a tested higher low off the 13.81–13.90 shelf could satisfy the group within 1–2 weeks if the shelf is retested and holds. Confidence: Moderate — the shelf has three touches and Friday's low of 13.995 stopped just above it.

Catalyst group — unlikely before 2026-11-11. Catalyst-dependent by construction and the catalyst has just passed. The next chance is the Q3 print on 2026-11-11 (consensus EPS US$0.22, revenue US$395m — note both figures predate this print and the revenue estimate now sits below the quarter just delivered, so expect them to be revised up). Confidence: Unlikely in the next four weeks; dLocal's last two prints produced single-session reactions of −13.0% (15 May, on the Q1 print — part of a −19.6% week) and −4.0% (14 August), so the base rate for a >+5% move is poor.

Fundamental group — already met and expected to stay met. It would only fail if price rose above US$17.45 (+23%) or the driver score fell below 50, neither plausible inside a month.

The practical read: the ladder is at Half-Size and the realistic path to Full-Size runs through the 13.81–13.90 shelf holding, not through a breakout.

Exit action: Holdno exit trigger is live — hold the position

Stop-Loss — not LIVE

⛔ Two consecutive closes below US$13.45 (0.6 ATR under the 13.81–13.90 shelf) — price US$14.17, 5.1% above

Thesis Invalidation — not LIVE

⛔ Full-year guidance cut — the opposite happened: TPV and gross-profit guidance were both raised
⛔ Gross-profit growth decelerates below the sector median — +29% YoY, far above
⛔ Gross profit over TPV falls below 0.65% — 0.72% now, down from 0.84%; this is the live one to watch
⛔ The primary volume driver turns to a headwind — TPV accelerating, not decelerating

Profit-Target — not LIVE

⛔ Price reaches the median analyst target of US$18.00 — US$14.17 now
⛔ RSI above 70 — 46.3
· Quality has not improved to justify the higher multiple

Forecast: Stop-loss — possible, not probable, inside 4–6 weeks. US$13.45 is 5.1% below the price and below a shelf that has held three times, but the ATR is 0.60 (4.2% of price) so the level is barely two average days away. A retest is likely; two consecutive closes through it is not, absent fresh news. Risk trigger: a hawkish FOMC minutes on 19 August driving the dollar higher.

Thesis invalidation — the one to diarise is 2026-11-11. Two of the four conditions could plausibly flip together on that date if the second-half operating-leverage step-up fails: a guidance cut, and gross profit over TPV breaking 0.65%. Two-of-N is the exit threshold. That single print is the highest-information event in this name's calendar, and the second half needs +38–47% operating-profit growth (reported basis; +34–44% adding back the Q1 prior-year tax item) against a first half that managed +15%.

Profit-target — unlikely in the next quarter. Requires +27% to US$18.00 and RSI above 70.

Imagine you act at the current price of 14.17 · as of 16 Aug 2026

What if you bought now?

If you are inclined to buy: the framework says a half-size starter is the honest size — one of three entry paths is open, and it is the valuation path, not the timing path. Scale the balance on either a tested higher low off the 13.81–13.90 shelf or a daily close back above the 20-day SMA at 14.67 with the MACD histogram positive two days running. The stop reference is US$13.45. Note the medium and long-term signals are BUY and the short-term is a HOLD purely on confirmation — this is a timing cap, not a verdict on the business.

What if you sold now?

If you already hold: nothing here says sell. No exit trigger is live, guidance was raised on two of three metrics, and the balance sheet is net cash. The exit discipline is the November print: if operating-profit guidance is cut and gross profit over TPV breaks 0.65%, that is two of the four thesis-invalidation conditions and the framework says exit rather than hold a half position hoping it mends.
13

Position Sizing Context

Illustrative portfolio math (not advice) translating conviction into an allocation given risk-per-share and volatility.

Position sizing not computed — specify your portfolio allocation and role for sizing guidance.

14

Calibration Snapshot

Machine-readable snapshot of every score, level and signal, saved alongside the HTML so the next run can compute deltas.
{
  "ticker": "DLO",
  "date": "2026-08-16",
  "version": "v6",
  "company": "DLocal Limited",
  "brand": "dLocal",
  "currency": "USD",
  "reporting_currency": "USD",
  "exchange": "NASDAQ",
  "exchange_ticker": "NASDAQ:DLO",
  "isin": "KYG290181018",
  "api_ticker": "DLO",
  "analysis_status": "donatien-pick",
  "finder_ticker": "DLO",
  "finder_exchange": "NASDAQ",
  "sector": "Information Technology",
  "gics_sector": "Information Technology",
  "industry": "Emerging-Markets Payments / Cross-Border Financial Infrastructure",
  "lifecycle_stage": "high_growth",
  "price_at_rating": 14.17,
  "price_asof_note": "2026-08-14 close (Friday). The run date 2026-08-16 is a Sunday; Friday's close is the latest print.",
  "prior_price": 15.14,
  "prior_report_date": "2026-07-31",
  "signal_short": "HOLD",
  "signal_medium": "BUY",
  "signal_long": "BUY",
  "primary_signal": "BUY",
  "short_entry_confirmed": false,
  "short_hold_reason": "technical_pending",
  "short_cap_reason": "Short-horizon technical-confirmation cap. The Fundamental entry group is met on valuation, but Technical and Catalyst are both unmet \u2014 the 2026-08-13 print produced a -4.0% reversal on 2.58x average volume, not the >+5% the Catalyst group requires, and the MACD histogram was last positive on 2026-07-17 and has been negative for 20 consecutive sessions since 2026-07-20 \u2014 it was already negative on 2026-07-31, the date of the previous report, and on 2026-08-13, the date of the print, so the earnings reaction confirmed the loss of momentum rather than causing it. Buy on confirmation: a daily close back above the 20-day SMA (14.67) with the MACD histogram positive two days running, or a tested higher low off the 13.81-13.90 support shelf.",
  "quality_score": 66,
  "valuation_score": 78,
  "timing_score": 61,
  "driver_score": 70,
  "quality_confidence": 72,
  "valuation_confidence": 70,
  "timing_confidence": 62,
  "overall_confidence": 62,
  "quality_detail": {
    "industry_benchmark_name": "EM Payments (TPV growth + take-rate stability)",
    "industry_benchmark_value": "TPV +92% YoY / gross-profit take rate 1.07% -> 0.72%",
    "industry_benchmark_score": 58,
    "moat_score": 60,
    "moat_pricing_power": 32,
    "moat_network_effects": 62,
    "moat_switching_costs": 76,
    "moat_cost_advantage": 56,
    "moat_intangible_assets": 72,
    "roic_percentile_vs_peers": 88,
    "capital_allocation": 75,
    "management_skin_in_game": 68,
    "gross_profit_over_tpv_q2_2026": 0.72,
    "gross_profit_over_tpv_q1_2026": 0.84,
    "gross_profit_over_tpv_q2_2025": 1.07,
    "gross_profit_over_tpv_fy2025": 0.99,
    "gross_profit_over_tpv_fy2024": 1.15,
    "net_revenue_retention_q2_2026": 153,
    "tpv_retention_q2_2026": 188
  },
  "economic_alignment_stance": "Trend-Following",
  "economic_alignment_conviction": 62,
  "economic_alignment_pressure": "Neutral",
  "economic_alignment_source": "watchlist-forecast (name-specific: DLO short N / medium O / long O)",
  "macro_report_date": "2026-08-12",
  "warranted_multiple": 27.7,
  "actual_multiple": 13.15,
  "val_multiple_basis": "forward P/E on FY2027 consensus diluted EPS of US$1.08 (FMP panel, n=4, dated before the 2026-08-13 print)",
  "warranted_ratio": 0.475,
  "val_band": "attractive",
  "discount_rate_r": 9.13,
  "risk_free_10y": 4.63,
  "g_near": 15.0,
  "g_term": 3.0,
  "eps_trailing": 0.68,
  "trailing_pe": 20.84,
  "eps_forward_fy2026": 0.82,
  "eps_forward_fy2027": 1.08,
  "forward_pe_fy2026": 17.27,
  "clean_pe": 13.15,
  "clean_peg": 0.42,
  "nonop_pct_of_net_income": 5.7,
  "valuation_detail": {
    "fcf_yield_q2_runrate": 7.1,
    "fcf_yield_h1_annualised": 4.3,
    "ev_usd_m": 3868,
    "ev_basis": "market cap US$4,169.8m less CORPORATE cash US$369.1m plus financial liabilities US$64.6m and lease liabilities US$2.7m. The provider EV of ~US$3,275m nets TOTAL cash of US$794.9m, most of which is merchant float held against US$1,555m of trade payables and is not the company's to spend.",
    "historical_valuation_decile": 3,
    "sector_median_fwd_pe_est": 15.0,
    "implied_growth_at_price": -2.5
  },
  "competitive_share_trajectory": "gaining volume, conceding price",
  "competitive_threat_level": "elevated",
  "timing_detail": {
    "mtf_confluence": 66,
    "risk_reward_score": 62,
    "relative_strength_vs_spy": -6.6,
    "relative_strength_vs_sector": -5.5,
    "catalyst_clustering_score": 78,
    "dynamic_macro_weight": 0.15,
    "rsi_14": 46.3,
    "atr_14": 0.6,
    "sma50": 13.85,
    "sma200": 13.37,
    "sma20": 14.67
  },
  "relative_strength_vs_spy": -6.6,
  "relative_strength_vs_sector": -5.5,
  "relative_strength_sector_etf": "XLF",
  "sector_mapping_note": "Deliberately split and disclosed in \u00a74: the valuation guardrail uses Information Technology (33x, the data providers' classification); relative strength and the macro row use Financials/XLF (the macro report's mapping). Neither is binding at an actual 13.15x.",
  "local_to_local_pct_of_tpv": 61,
  "local_to_local_pct_qoq_change": 6,
  "composite_short": 66,
  "composite_medium": 69,
  "composite_long": 69,
  "fair_value_est": 17.45,
  "stop_loss": 13.45,
  "target_price": 17.45,
  "scenario_bull_target": 21.0,
  "scenario_base_target": 17.45,
  "scenario_bear_target": 10.75,
  "scenario_bull_prob": 25,
  "scenario_base_prob": 50,
  "scenario_bear_prob": 25,
  "scenario_weighted_value": 16.66,
  "entry_groups_met": 1,
  "entry_conviction": "Half-Size",
  "exit_groups_live": 0,
  "exit_action": "Hold",
  "hard_gate_state": "caution",
  "gates_caution": [
    "Gate 5 - Regulatory / Binary Event: CAUTION, not triggered. No pending FDA-style decision, takeover or antitrust action, so the gate does not fire - but dLocal carries chronic elevated regulatory risk across its LatAm/Nigeria corridors, which is a standing elevated risk rather than a dated binary event, and is why the overall gate state is caution rather than clear."
  ],
  "gates_triggered": [],
  "do_not_buy_triggers": [],
  "systemic_tail_inherited": false,
  "systemic_tail_note": "DLO does not belong to the macro report's armed 'S&P 500 concentration / AI earnings-quality unwind' cohort: it is not levered to the AI capex or monetisation trade, is not an index top-weight, and its earnings carry no material non-operating mark-to-market (5.7% of TTM net income). The tail is not inherited into the Bear case.",
  "analyst_consensus_target": 18.35,
  "analyst_target_high": 21.0,
  "analyst_target_low": 14.5,
  "analyst_target_median": 18.0,
  "analyst_bullish_pct": 69,
  "analyst_coverage_count": 10,
  "analyst_grades_count": 13,
  "analyst_panel_staleness": "No price target issued in the last month (FMP lastMonthCount = 0; lastQuarterCount = 1). The panel predates the 2026-08-13 print \u2014 a recency discount is applied to the consensus sub-factor.",
  "q2_2026_tpv_usd_m": 17694,
  "q2_2026_revenue_usd_m": 399.664,
  "q2_2026_gross_profit_usd_m": 127.15,
  "q2_2026_operating_profit_usd_m": 64.161,
  "q2_2026_net_income_usd_m": 54.775,
  "q2_2026_diluted_eps": 0.18,
  "q2_2026_adj_fcf_usd_m": 68.548,
  "fy2025_operating_profit_usd_m": 220,
  "h1_2026_operating_profit_usd_m": 116.933,
  "fy2026_op_guidance_growth_pct": "27.5-32.5 (maintained)",
  "fy2026_tpv_guidance_growth_pct": "60-70 (raised from 50-60)",
  "fy2026_gp_guidance_growth_pct": "25-30 (raised from 22.5-27.5)",
  "debt_definition_used": "Company-reported IFRS balance sheet at 2026-06-30: financial liabilities US$64.6m + lease liabilities US$2.7m. FMP's lease-inclusive totalDebt was NOT used. The US$150m senior unsecured facility signed 2026-08-12 (SOFR+200bp, matures 2029-08-14) post-dates the balance sheet and is disclosed separately.",
  "next_update_date": "2026-08-31",
  "next_update_basis": "default +14d (no impactful dated event inside the window; next earnings 2026-11-11). 2026-08-30 is a Sunday, rolled forward to the next trading day.",
  "delta_vs_prior": "Price 15.14 -> 14.17 (-6.4%) across the 2026-08-13 Q2 print. The short-horizon signal is cut from BUY to HOLD on the technical-confirmation cap, so the Portfolio-Watchlist tile falls; medium and long stay BUY, so analysis_status keeps its donatien-pick value. THE CONVICTION LADDER WAS CUT: entry_conviction moved Full-Size -> Half-Size, entry_groups_met moved 2 -> 1, and short_entry_confirmed moved true -> false. The short entry path that the 31 July report recorded as confirmed is now closed; only the Fundamental path remains open. exit_action is Hold in both reports. Business Quality was marked down 6 points to 66, almost entirely on one sub-signal: the gross-profit take rate dropped again, to 0.72% of TPV from 0.84% last quarter, and pricing power inside the moat scorecard was cut with it. Valuation again scores 78 - the anchor still reads Attractive because the price fell faster than the forward estimate. Entry/Exit Timing eased 2 points to 61; the cause is NOT the print, since the MACD histogram went negative on 2026-07-20 and was already under water when the 31 July report was written. Separately the underlying-driver score rose 8 points to 70, because volume guidance was raised twice over and TPV growth accelerated to +92%.",
  "layer1_self_audit": "Clean on arithmetic and framework compliance after one internal fix round (see Data Sources). Two residual judgement calls flagged for the independent auditor: (1) the forward multiple is struck on FY2027 consensus EPS from a panel that predates the print - the band holds on FY2026 (17.3x, ratio 0.62) and on a blended NTM basis (14.6x, ratio 0.53), so the Attractive read is not basis-dependent; (2) geographic concentration is scored on qualitative disclosure only - the quarterly release names seven markets but publishes no revenue-by-country split."
}

The calibration below is the machine-readable record this report is scored from, and the input to the next refresh's delta. The fields that matter most for the audit trail are the anchor set (warranted 27.7×, actual 13.15×, ratio 0.475, band attractive), the rate build (4.63 + 4.50 + 0.00 = 9.13, all in percent), the short-horizon cap fields (short_entry_confirmed false, short_hold_reason technical_pending) and the Donatien-Pick status, which is retained because medium and long remain BUY.

15

Data Sources & Methodology

Every source used, what it was trusted for, what it was not trusted for, and the author's own falsifiable self-audit of this report before it reached an independent reviewer.
Data Source Status
dLocal Q2 2026 earnings release (SEC 6-K, accession 0001846832-26-000031, filed 2026-08-13 16:06 ET, period 2026-06-30) PRIMARY SOURCE for every company figure in this report: TPV, revenue, gross profit, gross profit over TPV, operating profit, net income, diluted EPS, adjusted FCF, the balance sheet, the guidance update and the US$150m credit facility. Quoted directly rather than via any data provider.
dLocal Q4/FY2025 earnings release (SEC 6-K, filed 2026-03-18) Source for the FY2025 baseline the guidance is struck against — operating profit US$220m, TPV US$40.8bn, gross profit US$402.8m — and for the original FY2026 guidance ranges. This is what makes the §11 Bear arithmetic checkable.
dLocal Q1 2026 earnings release (SEC 6-K, filed 2026-05-14) Source for the Q1 FY2026 comparatives and the US$9.7m prior-year (2023–2025) tax adjustment used in the step-7b earnings-quality note.
get_technical_indicators (Polygon) + get_yahoo_prices Every dated price claim, level and indicator. get_stock_prices was deliberately NOT used for any dated statement: its bar labels run one session behind (it emits Sunday-labelled bars carrying full volume). The 2026-08-14 close of US$14.17, the 15.28 open / 13.995 low, and all moving averages come from the two verified sources.
get_financial_ratios / get_income_statement (FMP) USED WITH CARE, AND NOT FOR HEADLINE FIGURES. FMP's income statement stops at Q1 2026, so every TTM ratio it returns (P/E 21.47, net margin 15.84%, P/B 7.68) predates the quarter this report is about. Its quarterly operating-profit figures also diverge from the company's IFRS statements (Q1 FY2026: FMP 53.552 vs company 52.772; Q2 FY2025: FMP 57.180 vs company 55.765) — company figures were used throughout. Its dividendPayoutRatioTTM of 0.78 is wrong; the correct figure is 28.9% (US$0.1967 ÷ US$0.68). Its nonOperatingIncomeExcludingInterest of +US$22.5m for Q4 FY2025 is a mapping artefact for an IFRS filer, disproved against the company's own statement.
Debt definition (trap 3c) Net cash and coverage are computed from the company's own 30 June 2026 balance sheet — financial liabilities US$64.6m plus lease liabilities US$2.7m — NOT from FMP's lease-inclusive totalDebt. The US$150m senior unsecured facility (signed 2026-08-12, SOFR+200bp, matures 2029-08-14) post-dates that balance sheet and is disclosed separately rather than folded in.
get_price_target_consensus / get_price_target_summary / get_yahoo_analyst_targets FMP's panel is too thin to use (lastQuarterCount = 1, a single US$20 target; lastMonthCount = 0). The Yahoo panel was used instead: mean US$18.35, median US$18.00, high US$21.00, low US$14.50, n = 10. Not degenerate (high ≠ low), so the fallback condition is satisfied. Crucially the panel PREDATES the 2026-08-13 print — a recency discount was applied to the consensus sub-factor rather than scoring it at face value.
get_analyst_estimates (FMP) FY2026 consensus EPS US$0.8206 (n = 4) and FY2027 US$1.07746 (n = 4). The FY2027 figure is the forward-P/E denominator. Yahoo's eps_forward of 1.1308 comes from a different panel and was NOT used — the FMP series is fiscal-year-dated with analyst counts, which is what an audit can check. The more conservative of the two was preferred.
get_stock_dividends Verified rather than taken from a summary. dLocal pays an ANNUAL dividend, not a quarterly one, so there is one payment in the trailing twelve months, not four: US$0.196666, declared 2026-03-16, ex 2026-05-27, paid 2026-06-10. It sums exactly to dividendPerShareTTM of 0.19667. The prior US$0.5107 (2025) was a special cash distribution, so the year-on-year comparison is NOT a dividend cut and is not presented as one.
MacroDriver-state-20260812.json DLO carries its OWN entry in watchlist_forecast — short N / medium O / long O — which was used in preference to the sector row. The macro report maps DLO to Financials (XLF, signals N/U/N); the name-specific signal is better than that row on both medium and long, and the difference is stated in §6.
Net revenue retention (153%) and TPV retention (188%) Not in the SEC-filed release text; sourced from the Q2 2026 earnings call, corroborated across two independent transcripts. Treated as company-stated but one tier below the filed figures, and the switching-cost sub-score was held at 76 rather than higher partly for that reason.
Geographic revenue concentration CONFIRMED ABSENCE, stated rather than assumed. The quarterly release names seven markets (Brazil, Mexico, Argentina, Egypt, Nigeria, South Africa, Other Africa & Asia) and discusses each qualitatively, but publishes NO revenue-by-country split; the annual 20-F does. The ~50% single-country concentration test could therefore not be computed. It is scored 70 on qualitative evidence — including this quarter's offsetting regional swings — and the limitation is carried into the pillar confidence rather than papered over.
Impact on scores: Net impact on the signal: none of the gaps changes a horizon. The two partial sources that could matter are the analyst panel and the geographic split. The panel feeds 15% of the Valuation pillar and was already discounted for staleness; the geographic split feeds one of seven Quality sub-signals. Neither moves a pillar across a Decision-Matrix threshold.

Layer-1 author self-audit. Recomputed independently and reconciled: the warranted multiple from its stated inputs (27.70x), the ratio (13.15 ÷ 27.70 = 0.475), all three horizon composites against the framework weights (Short 66.25, Medium 68.70, Long 68.85), the discount-rate build (4.63 + 4.50 + 0.00 = 9.13), trailing P/E against price ÷ EPS (14.17 ÷ 0.68 = 20.84), the scenario probabilities (25 + 50 + 25 = 100), the conviction ladder against the entry-group states (1 of 3 → Half-Size), and the §11 Bear's second-half arithmetic against the FY2025 release (US$220m × 1.275–1.325, less US$116.9m delivered, against US$118.4m a year earlier → +38–47% on a reported basis, +34–44% adding back the Q1 prior-year tax item).

Four corrections were made before publication, and are listed because a report that hides its own fix round is worse than one that had no defects. (1) The forward-P/E denominator was initially taken from Yahoo's eps_forward of 1.1308; it was replaced with the fiscal-year-dated FMP FY2027 consensus of 1.07746, which is both more conservative and auditable, and every downstream figure — actual multiple, ratio, PEG, fair value and all three scenario targets — was rebuilt from the replacement rather than patched. (2) The implied-growth read was first stated as +6.5%; reversing the two-stage model properly gives −2.5%, and the accompanying caveat about the terminal rate's floor was added at the same time. (3) The cash-yield denominator used the provider's enterprise value, which nets merchant float that is not the company's money — it was restruck on corporate cash, cutting the headline yield from 8.4% to 7.1%; §3 had already drawn that distinction, and §4 now matches it. (4) The multi-timeframe confluence was labelled 'Mixed / Transitioning' against a score of 66, which belongs in the framework's 65–79 'Mostly Bullish' band; the label was corrected and the prose now explains why a bullish label sits on an unbullish tape. None of the four changed a pillar score, a horizon signal or the conviction ladder.

Independent-audit round (post-publication corrections). An independent auditor returned FAIL with 2 MAJOR and 8 MINOR findings. It re-derived the signal set, every band, both gates and the tile-fall independently and confirmed all of them, so nothing below changes a horizon signal, a pillar score, the conviction ladder or the Donatien-Pick status. The defects were in the justification prose and the machine-readable record, which is arguably worse, and they are listed in full.
MAJOR-1 — a false 'held' claim that concealed a downgrade. The change log said the conviction ladder 'held at Half-Size'. The 31 July calibration records entry_groups_met: 2, entry_conviction: Full-Size and short_entry_confirmed: true. The ladder was cut Full-Size → Half-Size and the short entry path went from confirmed to closed. Corrected in the change log and in delta_vs_prior. The prior calibration was in hand when the claim was written, which makes this a process failure, not a data gap: no 'held' or 'unchanged' claim should be written without reading the prior file.
MAJOR-2 — the MACD run length was wrong in four places, and so was the causation. Stated as eight sessions; the histogram was last positive on 2026-07-17 and has been negative from 2026-07-20 through 14 August, 20 consecutive sessions. Worse, the delta narrative blamed the print for putting it under water — it was already negative on 2026-07-31 (the previous report's date) and on 2026-08-13 (the print date). Corrected in the change log, §7, §12 and short_cap_reason. The true figure makes the Technical group more clearly unmet, so the cap and the tile-fall are unaffected.
Eight MINOR corrections: (1) the 4.63% risk-free was attributed to the macro report, which actually cites 4.70% and carries no market-snapshot field — re-attributed to FRED DGS10 at 2026-08-13 with the 4.70% sensitivity shown (ratio 0.480, still Attractive); (2) the $10.64 low was 18 May 2026, not 15 May — corrected in §7, the §10 caption and §11; (3) the Q1 single-session reaction was −13.0%, not −20% (that was the week's move, −19.6%); (4) Gate 3's EV/Revenue was quoted at 2.4x on the provider enterprise value this report rejects at length in §4 — restruck at 2.85x on the corporate-cash basis; (5) 'no rating action in thirty days' contradicted itself in the same sentence and should have read no rating change; (6) the 1 Sell in the grades panel could not be corroborated from FMP's own individual grade records or Yahoo's panel — retained with the discrepancy disclosed rather than quietly dropped, since dropping it would flatter the score; (7) two quantifications from the earnings call were missing and are now in §5, including local-to-local at 61% of TPV, up six points on the quarter, which supports our own conclusion, and management's claim that the take rate would have been roughly flat sequentially excluding one large ride-hailing merchant, which cuts against it; (8) the report used Information Technology for the valuation guardrail while measuring relative strength against XLF and reading the Financials macro row, with nothing reconciling them — the split is now stated as a deliberate rule in §4 and the stored relative-strength field names XLF.

Two judgement calls flagged for the independent auditor. (1) The forward multiple rests on a consensus panel that predates the print; the Attractive band was therefore re-tested on FY2026 (17.3x, ratio 0.62), on a blended next-twelve-months basis (14.6x, 0.53) and on a 15% estimate cut (15.4x, 0.56), and holds on all four. (2) Business Quality at 66 sits two points above the 65 threshold that sets the discount-rate add-on at zero; had it scored 64, r would be 10.13%, the warranted multiple 23.6x and the ratio 0.556 — still Attractive. The valuation band is not sensitive to that judgement, and this is stated so the auditor does not have to derive it.
DISCLAIMER: This is a quantitative framework for educational purposes only. It is not financial advice. Always do your own research and consult a licensed financial advisor before making investment decisions.