MercadoLibre [MELI]
Spot $1,833.15 · 50.697m diluted shares · market cap $92,935m · net debt $5,748m (lease-INCLUSIVE) · EV $98,683m · TTM revenue $31,803m · EV/Sales 3.10x · EV/EBIT 32.9x
EV is stated lease-inclusive, per the DATA_DEFECTS requirement. Lease-exclusive net debt is $3,330m and EV $96,265m; the 2.5% difference is disclosed rather than buried.
Test applied (valuation.md). Not State A: operating margin is not low-variance — it has moved 14.1% → 12.7% → 11.1% → 9.6% (TTM) → 6.9% (Q1'26) across four years, and there is an active business-model transition (a lending balance sheet growing 90% a year inside a marketplace). Not State B: the driver is not an exogenous cycle but a deliberate, disclosed capital deployment. Not State D: revenue is $31.8bn, gross margin is 43.7%, the reverse DCF solves everywhere in range, and contribution economics are directly observable (NIMAL is published quarterly).
State C therefore, and the consequence is binding: the terminal margin must be BUILT through the expense bridge, not read off a trailing actual — and the mandatory output is the two-dimensional (g, m) expectations surface, not a point estimate.
Evidence grade: B. Volumes, take rates, credit quality and FX are all disclosed with unusual granularity. Two things hold it below A: the GMV redefinition of Q2 2025 is unquantified, and the free-cash-flow definition is contested even by the company's own non-GAAP measure.
The screen carried 11.1% labelled "own trailing operating margin". It is the FY2025 figure, not the
trailing figure — TTM is 9.6% — and it sits on a declining trajectory, which is the mirror image of
the ramp-extrapolation error steady_state_check.py exists to catch.
Opex bridge, FY2025 10-K lines, reconciled:
| FY2025 actual | Terminal | Basis for the change | |
|---|---|---|---|
Gross margin m_gross,T |
44.5% | 43.0% | continued 1P and principal-basis shipping mix, both disclosed and both structurally lower-margin |
| Product & technology | 7.9% | 7.5% | modest operating leverage on a $47bn+ revenue base |
| Sales & marketing | 11.1% | 10.5% | S&M headcount grew 55% in FY2025; that build does not repeat annually |
| Provision for doubtful accounts | 10.7% | 8.0% | the load-bearing assumption — see below |
| General & administrative | 3.5% | 3.5% | flat |
m_EBIT,T |
11.1% | 13.5% |
Hard constraint m_EBIT,T ≤ m_gross,T: 13.5% ≤ 43.0%. SATISFIED, with 29.5pp of headroom.
Why the provision line normalises to 8.0%. CECL requires the lifetime expected loss to be booked at origination while the interest accrues over the life of the loan. A book growing at 90% therefore carries a provision charge sized to next year's book against this year's revenue. FY2023 — originations growing far more slowly — ran 7.0%. As origination growth converges to revenue growth, the front-loading unwinds mechanically. 8.0% sits between the FY2023 7.0% and the FY2024 8.9% and is above neither. This is the whole reason the terminal margin exceeds the trailing actual, and it is a named causal bridge, not a wish.
Sanity band, not an override: MELI's own FY2023 operating margin was 14.1%. The terminal sits 0.6pp below the best year in the recent record and 2.4pp above FY2025. Screen peer median 12.3%, industry p75 21.1% — both used only as a band, per the rule that six names were previously destroyed by an industry cap.
steady_state_check.py — what it flaggedpython3 steady_state_check.py --ticker MELI --terminal-margin 0.096 --exit-multiple 24.4
| margin ramp 2020→2025 | 3.2% → 11.1%, +7.9pp; latest is window peak: False |
| ROIC measured | 18.0% |
| warranted multiple at 18% steady-state ROIC | 9.4x |
| exit multiple used (screen's) vs warranted | 2.59x |
| FINDING | EXIT_MULTIPLE_UNWARRANTED_AT_STEADY_STATE — 24.4x is 2.6x the 9.4x warranted at an 18% steady-state ROIC. Justifiable only if today's returns persist forever, which is the assumption under test. |
It did NOT fire MARGIN_STILL_RAMPING — correctly, because MELI's latest margin is not its window
peak; it is 3.5pp below it. The tool is calibrated for ramps upward and MELI's problem is the
opposite: the screen took a point on a descending line and held it flat forever. I treat that as the
same defect with the sign reversed, and it is why the terminal margin here is built rather than read.
The exit-multiple finding is the substantive one and I act on it: the base exit multiple is set at 18.0x, materially below MELI's own history and materially above the identity's 9.4x, with the whole surface reported rather than a point.
Terminal value is 84% of EV at the base case, so the reverse DCF is mandatory as the primary long-horizon instrument, not a cross-check.
Fixed: terminal margin 13.5%, WACC 11.0% (LatAm sovereign and FX risk; a 10% WACC would lower
required CAGR by roughly 2pp), 5 years, EBIT basis, net debt $5,748m, 50.697m shares.
Solved for: revenue CAGR.
FCF margin: 14.2% base — the corrected figure from MELI_Research.md §6, not the 25.7% that
normalized_fcf.py returns.
| exit EV/EBIT → | 10x | 14x | 18x | 22x | 25x | 33x |
|---|---|---|---|---|---|---|
| FCF margin 0.0% (company's own adj. FCF basis) | 31.10 | 22.57 | 16.56 | 11.98 | 9.15 | 3.25 |
| FCF margin 14.2% (BASE) | 21.88 | 15.47 | 10.71 | 6.96 | 4.59 | −0.45 |
FCF margin 25.7% (normalized_fcf.py output) |
15.56 | 10.39 | 6.42 | 3.21 | 1.16 | −3.29 |
At the identity-warranted 9.4x with the base FCF margin: required CAGR 23.06%.
| Required CAGR (base: 18.0x, 13.5%, 14.2% FCF) | 10.71% |
| Demonstrated CAGR (3-year trailing revenue) | 38.9% |
| MARGIN (demonstrated − required) | +28.16pp — PASS |
| Screen's margin | +26.2pp |
| Implied compression from today's 32.9x to the 18.0x exit | −45.3% |
Robustness — this is the finding. The name passes at every cell of an 18-cell surface spanning
10x–33x exit multiples and a 25.7pp range of FCF margin. The worst cell (10x exit, 0% FCF) requires
31.10% against 38.9% demonstrated: still +7.8pp. Even at the steady-state-warranted 9.4x, required
is 23.06% versus 38.9% demonstrated: +15.8pp. MELI is the only one of the three names whose verdict
survives the steady_state_check correction.
The honest counterweight. The 38.9% demonstrated CAGR is a USD nominal figure inflated by LatAm inflation pass-through, by 1P gross-up, by a principal/agent flip in shipping, and by credit interest booked gross. The volume metrics are the check that does not share those defects — GMV +42.2% and TPV +49.5% in Q1'26 — and both exceed the required 10.71% by a factor of four. The PASS does not depend on believing the revenue line.
Multiple anchored on MELI's own history, percentile stated.
| current | 6-year (n=1,430–1,509 days) | 3-year window | |
|---|---|---|---|
| EV/Sales | 3.15x | 4th percentile (min 2.65, med 5.99, max 32.03) | 8th percentile (p25 4.73, med 5.47) |
| EV/EBIT | 32.86x | 7th percentile (min 26.45, med 49.46, max 1267) | 14th percentile (p25 36.82, med 41.11) |
Regime-overlap check. The six-year window opens 2020-07-27 and contains (a) the pandemic e-commerce bubble, where EV/Sales reached 32.03x, and (b) an era in which MELI earned a 14%+ operating margin. The current regime — credit-led growth at a sub-10% margin — begins in 2024. Overlap of the 6-year window with the current regime is roughly 42%; the 3-year window's is ~83%. The 3-year window is therefore the identified one and the 6-year figures are reported for disclosure only. Status: IDENTIFIED on the 3-year window.
Build. NTM revenue from consensus, time-weighted across the fiscal boundary (7/12 of FY2026 elapsed): 5/12 × $40,720m + 7/12 × $51,791m = $47,178m. NTM EPS on the same weighting = 5/12 × $39.98 + 7/12 × $57.36 = $50.12, implying NTM EBIT of roughly $3,308m (grossed at a 22% tax rate, plus modest net interest) — a 7.0% NTM margin, i.e. the Street models further compression, not recovery.
At the 3-year 25th percentile EV/EBIT of 36.82x: EV $121,734m − $5,748m net debt = $115,986m equity → $2,288 per share, +24.8% to spot. Holding the multiple flat at today's 32.86x (14th percentile) gives $2,031, +10.8% — the low case.
Named events inside 12 months that move the estimates: Q2'26 results (early August 2026) and Q3'26 (late October 2026), each carrying the NIMAL print and the origination-growth print, which together determine whether the provision line has peaked; and the Argentine macro path, where the inter-annual inflation rate has fallen to 32.7% from far higher — a disinflation that reduces the nominal revenue tailwind while improving the credit book.
Expect-targets-above-spot check: +24.8%, above spot. Consistent with the reference professional's +20% median and with calibration item B16, which found 16 of 16 house targets below spot.
Revision breadth is adverse and is not hidden: EARNINGS_ESTIMATES shows 0 upward and 7 downward
EPS revisions over the trailing 30 days on both fiscal years. That is the strongest single argument
against the target and it belongs in the same paragraph as the target.
Not shared with RBA or BKNG. Those are a $4.7bn auction house growing 7% and a $27bn travel agent growing 8–12%; MELI is a $31.8bn LatAm marketplace-plus-lender growing 39–49%. One multiple across the three would be the exact defect valuation.md documents (a peer set spanning 1.0–7.5% growth used to value a 38.8% grower).
| Comparator | Why in the set | Growth bracket at exit |
|---|---|---|
| Sea Limited (SE) | SE Asia marketplace + Monee/SeaMoney lending; the closest structural analogue anywhere | 20–30% |
| Coupang (CPNG) | 1P-heavy emerging-market e-commerce with its own logistics; brackets the 1P mix question | 15–20% |
| Nu Holdings (NU) | LatAm consumer credit at scale; brackets the credit-book half | 25–35% |
| PagSeguro / StoneCo | Brazilian acquiring; brackets the payments take rate | 8–15% |
| Amazon (AMZN) | the terminal-state marketplace + logistics + ads architecture MELI is building toward | 8–12% |
Bracketing requirement: MELI's exit-year (FY2031) growth under the base 10.71% required path sits inside the AMZN/Stone bracket and below the SE/NU bracket, so the set brackets the subject at exit in both directions. The set is admissible. Multiples were not applied mechanically from it — per valuation.md rule 4 the exit multiple is derived and the traded set is a sanity check only.
Named cause: the credit card book. If NIMAL continues at its current rate of decline (−4.9pp year over year in Q1'26) it reaches roughly 13% by FY2027 while the book keeps compounding. At that point credit revenue growth no longer covers its own provision, the provision line stays above 12% of revenue instead of falling to 8%, and the terminal margin is 8% rather than 13.5%. At an 18x exit and 8% terminal margin the required CAGR rises to roughly 18–19% — still below the 38.9% demonstrated, so the Criteria verdict survives, but the 12-month target does not: NTM EBIT falls by ~40% and the target moves to roughly $1,370, −25% to spot.
Type: MEASURED. Logged and scored; it does not reject the name.