← back to picks

Founded 2021 IPO · operating since 2006 HQ Vilnius (LT) · listed London Reporting EUR Credit Aaa/AAA (synth, net-cash) Valuation 2026-05-24 1000-iter Monte Carlo · 5 correlated axes FCFF · Dark v3

Deeply undervalued, against the JPM consensus · +34.0% margin of safety

Market below MC p5 — robustly cheap

Intrinsic value €3.28/share (after a 7% governance discount) vs market €2.16 (≈ 187p on LSE). The Monte Carlo p5 — the conservative tail — still lands at €2.24, about +3% above today's price, meaning 97% of the council's joint stress scenarios still show the stock cheap.

SectorOnline classifieds (Baltics)Country mixLT 71% · EE 27% · LV 2%MC σ±€0.46/sh (1000 runs)GovernanceNo controlling holder · 7% haircutQuality78% EBITDA · net-cash FY26 · ROIC 14%Drawdown−42% from Oct-25 peak

Intrinsic / share

€3.28

post 7% gov · ≈ 283p

Market / share

€2.16

22 May 2026 close · ≈ 187.1p LSE

Margin of safety

+34.0%

vs intrinsic

Enterprise value

€1.58B

67.4% terminal

Cost of equity / debt

8.11% / 2.38%

β 1.08 · CRP 1.06%

Terminal ROIC / g

30.00% / 2.00%

spread ~2198bp · ROC capped from auto-resolved 238%

What it sells, where it sells

Operating segments

Real EstateAruodas (LT) 48× site-leadership; KV.ee + City24 (EE). Highest-yield vertical, B2C ARPU +27% latest year.~35%

AutoAutoplius (LT) + Auto24 (EE 31× lead). Revenue flat in EE through −45% volume on +29% C2C yield — pricing-power proof.~28%

JobsCVBankas (LT) + CV.ee + CV.lv. Proprietary CV database — the segment LLM-search-based disintermediation can't scrape.~22%

GeneralistSkelbiu (LT, 30× lead) + Kuldne Börs (EE). The ~15% of revenue most exposed to Facebook Marketplace / AI search disruption.~15%

Real Estate and Auto together are 63% of revenue and earn the highest yields per listing — dealer and agent customers have no equally-trafficked alternative, which is why the operating margin debate is really a yield-elasticity debate, not a volume one.

Country mix (revenue-weighted CRP input)

🇱🇹Lithuania71.0%

🇪🇪Estonia27.0%

🇱🇻Latvia2.0%

Lithuania dominates because it has the most-monetised verticals (Aruodas in real estate, CVBankas in jobs) and the largest population of the three; Latvia at just 2% is upside optionality — a 10× runway if BCG can replicate the Lithuanian playbook on a 1.8M-person market.

Background — five things to know before the case

Context the share price doesn't carry on its face. Skim this once and the bull / bear bullets below stop reading as inside baseball.

BCG runs the dominant car-sales, real-estate, jobs, and general-listings websites in Lithuania, Estonia, and Latvia. Twelve to fourteen brand portals across three eurozone countries, 6.5M total residents, 156 employees. Charges car dealers, real-estate agents, recruiters, and individuals to list items. The moat is network-effect: more listings draw more buyers, which makes dealers pay more to be listed, which attracts more listings. After 20+ years, the Lithuanian real-estate site Aruodas now gets 48× more time-on-site than its nearest competitor — up from 27× one year ago.
78% EBITDA margin is the headline that keeps doubters up at night. For context: Apple is ~33%, Microsoft ~52%, Google ~35%. Such margins typically invite competition or regulation. In BCG's case they are sustained by a tiny cost base (€18M FY25) and structural monopoly in a closed linguistic market English-speaking platforms have never bothered entering. Margin has held in a 76-78% band for four consecutive full years and three consecutive half-years, including through the 2022-23 Baltic inflation peak.
Stock fell 42% from October 2025 peak after JPMorgan cut its price target from 770p to 178p. JPM's thesis: GenAI search (ChatGPT, Perplexity) will let people find cars/houses/jobs without going through portals, killing the portals' fees. JPM picked Scout24 (Germany) and SMG (Switzerland) as the only European winners — BCG, AutoTrader, Rightmove, Hemnet all downgraded together. A second blow came in December 2024 when Estonia introduced a vehicle tax that collapsed car transactions 45-66%; BCG's Estonian auto revenue held flat by raising listing prices 29%.
Apax (the private-equity sponsor) fully sold out in July 2024 at a 4.2× return — so there is no overhanging seller, but also no smart-money endorsement. Apax IX cleared at €2.94/share on an accelerated bookbuild. Sponsor exits at peak often precede 12-24 months of underperformance (the "PE-exit base rate"), and we are roughly 10 months into that window. Today the cap table is pure institutional float — Vanguard, BlackRock, Aberdeen, Baillie Gifford each holding 4-6%. CEO Šimkus retains 2.88% as modest skin-in-the-game.
Insiders are buying through the drawdown. Chairman Trevor Mather bought 250,000 shares for £465K on 4 December 2025 at 186p (close to today's 187p). NED Ed Williams added ~£200K worth in December 2025 at 180-189p, then bought another 90,515 shares at 192p on 27 April 2026 — i.e. after the JPMorgan price-target cut. The company will be debt-free in FY26 (announced ~July 2026), at which point management has flagged a capital-policy update — likely a step-up in dividends or buybacks. Cash conversion has been 99% of accounting earnings every year since IPO.

Where we diverge from the sector

Each row asks: if we'd used the sector median for this one assumption instead of our override, how much would the share price change? Negative = our override is more conservative (worth less); positive = more aggressive (worth more). Skim down the rightmost column to see which bets are doing the work.

Assumption

Our input

Sector median

Value impact

01Revenue growth10-year CAGR · €86M → €199M

8.7% blended; conservative vs 18% 4-y actual

~9.0% global Software (Internet) cross-section

±€0.05

02Operating marginYear-10 EBIT / Revenue

70.0% mid-70s EBITDA mgmt guide less ~5pp D&A

3.7% Software (Internet) median EBIT margin

+€2.10

03Sales-to-capital€ of new revenue per € of new capital

4.0× software-only capex; 99% cash conversion since IPO

1.35× Software (Internet) median reinvestment intensity

+€0.18

04Terminal growthYear 10+ steady-state growth forever

2.0% below EUR risk-free; Baltic demographic decline

2.4% EUR risk-free rate (Damodaran ceiling)

−€0.12

05Cost of capital10-year WACC; lower = higher value

8.02% β 1.08, ERP 4.5%, weighted CRP 1.06%, ~5% debt

~11.3% cross-sector median WACC under sector β 1.69

+€0.06

Net effect of overrides

Overrides add roughly +€2.27/share vs an all-sector-medians shadow DCF. Almost all of that lift comes from the operating-margin override — the rest cancels out. Translation: if you don't believe BCG can sustain mid-70s EBITDA margins, almost nothing else in the model matters.

+€2.27

Our override Sector median Override adds value Override subtracts value

The story & the five claims

The 10-year story this DCF is built on, plus the five anchor claims that translate the story into the model.

The 10-year story

Baltic Classifieds is the dominant online-listings company for cars, real estate, jobs, and general items across Lithuania, Estonia, and Latvia — three small eurozone countries with 6.5 million residents combined. The market hates it right now because JPMorgan published a thesis in November 2025 arguing that AI search (ChatGPT, Perplexity) will let buyers find cars and houses without going through portals, and because Estonia introduced a car tax in late 2024 that collapsed transaction volumes there by roughly half. The stock fell 42% from its October peak, JPM cut its target from 770p to 178p, and the share price now sits at 187p — almost exactly where company insiders, including the chairman, were buying with their own money throughout the drawdown. The base case here is not that AI fails to disrupt search. It is that BCG's revenue roughly doubles over the next decade (8.7% growth per year, against a 16% historical actual), that its operating margin holds in the 70% range (today 78% on EBITDA), and that the company keeps essentially all its earnings as cash because growth in this kind of software-only business needs almost no new capital. Add up ten years of discounted cash flows plus a perpetual tail, subtract a 7% governance discount for Baltic regulatory risk and the just-departed private-equity overhang, and you get €3.28 per share vs €2.16 in the market — a 34% margin of safety.

Two debates worth pressure-testing

Is the moat actually narrowing, or is JPM measuring the calm before AI disruption ships?

Our view: Two of the three biggest sites have widened their lead through the exact period of AI-disruption noise (Aruodas 27× → 48×, Skelbiu 21× → 30× in 12-18 months). The Contrarian on the council called this "the calm before the storm" — fair, but if 2024-26 is the calm, the bear case has to argue AI will reverse a four-year moat-widening trend in the next four. If JPM is right after all, margin compresses from 70% to ~55% and the fair value drops about 25%, landing near today's market price — still defensible, no longer a clear buy.

How much can BCG raise prices before the Estonian Competition Authority intervenes?

Our view: BCG raised Estonian car-listing prices 29% in one year while volumes fell 45%. That is the kind of move that triggers regulators in any market — especially a 1.3M-person country where the dealer body and the regulator know each other. The Authority already has open enquiries. If they cap annual price hikes at, say, CPI, the yield-expansion thesis breaks and operating margin glide-slopes toward 60% — costing roughly €0.40/share or 15% of intrinsic value.

CLAIM 01Revenue roughly doubles in 10 years.€86M TTM → €199M by 2036 (8.7% CAGR)Conservative vs the 18% actual 4-year CAGR. Growth comes from Lithuanian yield expansion, the Latvia white-space (still only 2% of revenue), and product-depth add-ons (AI-assisted listing, automated property valuation).

CLAIM 02Operating margin holds at ~70%.target_op_margin = 0.70Today's 78% EBITDA minus ~5pp normalised depreciation. Held in a 76-78% band for four years through Baltic inflation and the Estonian tax shock. Moat is widening, not compressing (Aruodas 20.7× → 48× site-leadership in four years).

CLAIM 03Growth needs almost no new capital.sales-to-capital = 4.0× (Y1-10)€1 of new invested capital produces €4 of new revenue. 99% cash conversion every year since IPO. 156 employees, no inventory, software-development capex offset by amortisation. The business is structurally close to its free-cash-flow ceiling.

CLAIM 04Margin and growth settle by year 10.year_of_convergence = 10Standard 10-year window. The moat-widening signal would argue for 12 years; the AI-disruption overhang argues for 10. We split the difference and let Monte Carlo explore the duration tail.

CLAIM 05Steady-state growth at 2% forever; zero failure risk.terminal_growth = 0.020 · p(failure) = 0Below EUR risk-free (2.4%) and below eurozone real GDP — because the Baltic population is shrinking ~0.5%/yr. Failure probability is zero: debt-free in FY26, no controlling shareholder, no related-party transactions, 99% cash conversion.

Where we diverge from sell-side

The market prices BCG as if AI disruption is imminent. We see the opposite signal in the data: Aruodas' lead has more than doubled to 48× during the exact two years of supposed disintermediation. The market is extrapolating a thesis the leadership-ratio numbers actively contradict.
The market is treating Apax's clean exit as a "smart money sell" — we read it as overhang removal. Apax sold in a marketed bookbuild at €2.94, well above today's €2.16. A 4.2× MOIC PE exit is normal; the resulting institutional-only cap table removes the forced-seller risk that previously discounted the equity.
The bear case has to ignore the insider buying. Chairman, NED, and the executive team have committed roughly £830K of personal capital at prices within 3% of today's. That is not token signalling — it is sized to matter to their personal net worth, including a £170K April 2026 purchase after the JPM price-target cut.
The debt-free pivot in FY26 is a near-term re-rating catalyst the model does not capture. Six months from now BCG will publish a capital-policy update; the company is already buying back ~€20M/year while paying a growing dividend. The bridge from "JPM-driven panic" to "Baltic Auto Trader compounding via capital return" runs through that single announcement.

Two-sided case — bear anchors

If the JPM AI thesis is right. Vertical scrapers and agentic search bypass the portal directly, sellers stop paying because buyers stop arriving, and the 78% EBITDA margin compresses to 50-55% over five years. Fair value drops to roughly €1.95/share (~169p) — and today's 187p is actually a fair price, not a buy.
If the Estonian Competition Authority caps pricing. The 29% yield hikes that held Estonian Auto revenue flat through the volume collapse get capped at CPI. Yield expansion — the core mechanic of the Lithuanian growth story too — becomes regulated. Fair value drops 12-15% (~€0.35/share).
If the PE-exit base rate plays out. Sponsor exits at peak typically precede 12-24 months of underperformance; we are roughly 10 months in. Another leg down of 15-25% is plausible before any operating-results-driven re-rating.
If Facebook Marketplace scales in the Baltics. The C2C generalist segment (~15% of revenue, Skelbiu) is the most exposed. Loss of half that segment costs ~7% of group revenue and ~5% of fair value, but more importantly removes the "compound forever" framing that justifies the terminal multiple.
For the bull thesis to be wrong overall, you need to believe the 2024-26 leadership-ratio data is a leading indicator of disruption (not its opposite), that insider buying is a signalling artifact rather than a conviction trade, and that the debt-free capital return announcement coming in July 2026 will not move the share price.

Risks to thesis (tail, not bear case)

EUR/GBp 10-year FX pathHigh

BCG reports in EUR but trades in GBp. A 10-year EUR/GBp path swings fair value materially — a 10% GBp strengthening reduces the per-share value in pence terms by ~10% even with operations unchanged. Not modeled in this DCF; needs a separate forward-curve overlay.

Strategic acquirer pre-emptsMed

At €974M market cap BCG is a strategic snack for Adevinta, Schibsted, or a re-entering PE consortium. Take-out premium is real expected value the DCF does not capture — partial offset to the bear scenarios. A typical European classifieds takeout has gone at 20-35% premium.

Apax exit pathway uncheckedMed

Did Apax sell to a strategic, a financial buyer, or distribute to LPs? Distribution patterns reveal whether smart money thinks terminal value remains. Reviewer 5 on the council flagged this as a checkable fact that was not checked.

Schulenberg-style succession eventMed

CEO Šimkus retains 2.88%; chairman Mather and other directors are recent. No equivalent founder-succession overhang to Eventim, but key-person risk on a 156-FTE business is non-trivial. A bench-strength check is the right follow-up.

Baltic geopolitical shockMed

All three Baltic states are EU/NATO/eurozone members, but Russian re-armament spillover or hybrid-warfare events would hit 100% of the book simultaneously. Damodaran's CRP captures part of this; the discrete tail is not fully priced.

Estonian tax extension to LithuaniaLow

Lithuania does not currently have a vehicle-transaction tax. If the Estonian model is copied across the border, the 71%-Lithuania pillar of the model gets a 3-5% volume haircut. Low probability — Lithuania has different fiscal priorities — but worth tracking.

10-year forecast

Revenue €94M → €199M over 10 years (blended 8.7% CAGR, a deliberate deceleration from the 18% historical 4-year actual). Operating margin lifts from 67.7% in Year 1 to 70.0% by Year 10 as the acquired-intangible amortisation drag rolls off. FCFF compounds with revenue almost 1-for-1 — the reinvestment line stays under €5M throughout because growth here essentially does not need capital.

Monte Carlo distribution

P(intrinsic < market €2.16) = 3.3% — in 96.7% of joint stress scenarios across the five council-widened axes (year-10 revenue, year-10 margin, terminal growth, governance discount, terminal ROC) the fair value still comes out above today's price. The p5 outcome — the 5% worst case across 1000 correlated draws — is €2.24/share, roughly +3% above today. After re-anchoring β to the 1.08 five-year regression (down from the 1.61 sector-relevered default), the undervaluation is robust across the entire MC distribution.

Mean €2.97 ± €0.46/sh over 1000 iterations (0 failed). Percentiles: p5 €2.24 · p25 €2.66 · p50 €2.97 · p75 €3.28 · p95 €3.74. P(intrinsic < market €2.16) = 3.3%.

Reference shelf

Where the numbers and the story come from, for the curious reader.

Qualitative read — why the market hates it, the counter-narrative, the capital-allocation and governance signals.
Five-claim build-up — each input number traced to a specific data point in the filings corpus.
Filings digest — 20 BCG filings since the July 2021 IPO: revenue progression, margin progression, capital allocation, site-leadership ratios, governance signals.
LLM-council pressure-test — mandated the terminal-ROC cap at 30% and raised the governance discount to 7%.
Country mix rationale — why Lithuania 71% / Estonia 27% / Latvia 2% (sourced from the FY25 AR geographic disclosure).
Independent buy-side note — Atomic Moat Research, March 2026: explicit base case 220p (+15%), bull case 280p (+45%), frames the de-rating as a Buffett-Washington-Post-style mispricing.
Original JPM downgrade — November 2025 European online classifieds note: PT cut 770p → 178p, Overweight → Underweight, the catalyst for the −42% drawdown that created the opportunity.

Cost of capital build

Risk-free rate	2.40%
Mature-market ERP	4.50%
Levered β	1.08
Weighted CRP	1.06%
Cost of equity	8.11%
Pre-tax cost of debt (synth Aaa/AAA)	2.38%
D / V	~5%
WACC	8.02%

Full year-by-year DCF

Year	Revenue	Op mgn	EBIT	EBIT(1−t)	Reinvest	FCFF	PV
1	€94M	67.70%	€63M	€55M	€2M	€53M	€49M
2	€102M	67.95%	€69M	€60M	€2M	€58M	€50M
3	€111M	68.21%	€75M	€66M	€2M	€64M	€51M
4	€120M	68.47%	€82M	€72M	€2M	€70M	€51M
5	€131M	68.72%	€90M	€79M	€3M	€76M	€52M
6	€142M	68.98%	€98M	€85M	€3M	€82M	€52M
7	€155M	69.23%	€107M	€93M	€3M	€89M	€52M
8	€168M	69.49%	€117M	€101M	€3M	€97M	€52M
9	€183M	69.74%	€128M	€109M	€4M	€105M	€53M
10	€199M	70.00%	€139M	€118M	€4M	€114M	€53M

Methodology & flags

Damodaran FCFF DCF, 10-year explicit forecast + perpetuity. Single-segment build (Software-Internet classifieds). R&D not capitalised (no qualifying spend under IFRS); IFRS-16 leases already inside book debt (€1.7M of €16M). Synthetic credit rating Aaa/AAA (net-cash trajectory, debt-free by FY26). Country-risk premium is the revenue-weighted average of Lithuania 71% × 1.10% + Estonia 27% × 0.91% + Latvia 2% × 1.55% = 1.06%. Monte Carlo: 1000 iterations across five correlated axes (year-10 revenue, year-10 margin, terminal growth, governance discount, terminal ROC). Council overrides applied: terminal ROC capped at 30% (from auto-resolved 238%), governance discount raised 5% → 7%. Engine v1.0.0 · result: valuations/bcg/output/2026-05-25-result.json

R&D cap: OFF · Lease cap: OFF · Failure: OFF · ESO: OFF
Governance discount: 7% applied post-DCF (€3.52 pre > €3.28 post-gov)
Sensitivity tornado: not run (Monte Carlo supersedes — 1000 iterations across 5 axes)

FCFF DCF · dark theme v3. Generated 2026-05-25T11:00:00 from valuations/bcg/output/2026-05-25-result.json.