ZRR0 IX LLC — Financial Model & Roadmap
Pre-ops planning document. Prepared 2026-06-06. All dollar figures are marked [R] = researched-current (2026) or [E] = estimate. Re-verify every [R] marked "policy-volatile" with a licensed customs broker at the time of each shipment — the duty regime moved three times between Aug 2025 and Feb 2026.
0. The one-paragraph verdict
The cross-market chimera thesis is sound and the arbitrage is real. But as currently sequenced the plan rests on three load-bearing assumptions that are wrong or unstable, and on one funding dependency dressed up as independence. Fix these before any capital moves:
- Washington has NO wholesale-only license — it was abolished in 2019. There is one dealer license, and it forces a bonded + signed + staffed + zoned physical location from day one. The "lean wholesale start" does not exist. [R]
- The import duty is not a constant. It legally moved 2.5% → 15% → ~2.5%+10% between Aug 2025 and June 2026 (SCOTUS voided the IEEPA reciprocal tariffs; a Section 122 ~10% stopgap is live ~150 days from Feb 28 2026, sunsetting ~late July 2026, overlapping your likely first run). A stored constant WILL misprice and can halve a flip's margin. [R, policy-volatile]
- Engine-swapping a JDM car in Japan can make it non-importable. A non-original engine voids the EPA 21-yr/25-yr emissions exemption. The MUTT "swap in Japan then ship" play must move to US soil, post-entry, on cars already 25+. [R]
- The funding spine (the Lighthouse trading book) is in its FIRST live verdict month right now (June 2026). Do not launch the car business on "trading should fund it." Gate launch on a measured, withdrawn-cash milestone.
Everything below re-baselines the model around these realities.
1. Capital allocation — the $350k starting line
The binding constraint is inventory-float velocity (frequency × per-car edge), not capital size — exactly the chimera/effective-N logic from the trading book. Over-allocating to warehouse, retail, and racing early starves the engine that compounds.
1.1 Re-baselined allocation (corrects the "wholesale = cheap, no location" error)
| Bucket | Amount [E] | % | Notes |
|---|---|---|---|
| Inventory float (the compounding engine) | $175,000 | 50% | Reduced from naive $190k because the WA staffed location is now a day-one cost, not deferred. |
| WA dealer location (lease deposit + 6mo rent + signage + buildout) | $55,000 | 16% | ~5–10k SF flex unit, Fife/Tacoma/Kent corridor, ~$10–16/SF/yr NNN [R]. Day-one, not phase-2. |
| Licenses + bonds (WA dealer license, $30k bond premium, plates, AES/EIN, continuous customs bond) | $8,000 | 2% | Bond premium ~$300–900/yr (not the $30k face) [R]; continuous customs bond ~$500–1,000/yr [R]. |
| Opex buffer (6 months, re-baselined ≥ one full auction-to-sale cycle) | $48,000 | 14% | Stress-test fix: the buffer MUST exceed the 3–5 month first cycle. ~$8k/mo lean burn. |
| Valuation engine v0 (analytics contractor pre-ops + data feeds yr 1) | $22,000 | 6% | Pre-ops build — see §6. The centerpiece. |
| Brand skeleton (website, identity, trademark clearance, first merch run) | $18,000 | 5% | |
| Build-shop tooling reserve (2-post lifts, intake bay — HER0/MUTT) | $14,000 | 4% | Defer until run 1 closes. |
| Racing (self-sponsored, ring-fenced) | $10,000 | 3% | Ring-fenced. First bucket cut if trading wobbles. Funded by trading P&L, NOT car cashflow. |
| Total | $350,000 | 100% |
Stress-test fix folded in: "Float lock-up — 350k is really ~175k." The other $175k is location, buffer, engine, brand, tooling, racing. Plan accordingly: you are running runs of 3–4 cars, not 10.
1.2 What changed vs. the operator's original mental model
- "Wholesale first, retail later, ~$30k bond only later" → there is no wholesale tier in WA; the bond + staffed signed location is day one.
- "First run needs zero full-time hires" → mostly true on the Japan side (all vendors), but if you sell in WA you need a lot staffer for the mandatory 10am–4pm / 5-day staffed hours. Either eat that hire, or license the dealer entity in a no-staffed-hours state and keep WA warehouse-only (CPA + attorney signoff required).
2. Unit economics
2.1 JDM mid-tier flip (the high-frequency core sleeve)
Worked example: $16,000 FOB auction car, 25+ years old, grade 4+, sold for $27,500 in the US.
| Line | Amount [E/R] | Source/notes |
|---|---|---|
| Auction hammer (FOB-ish basis) | $16,000 | [E] car-specific |
| Japan auction fee + inland + export pkg (~¥140k) | $950 | [R] ~¥140k typical under ¥2.5M |
| Agent service fee (flat) | $500 | [R] ~¥70k flat-fee model |
| Ocean freight Yokohama→Tacoma (RoRo) | $1,400 | [R] $900–2,000 RoRo, West Coast |
| Marine cargo insurance (1–2%) | $300 | [R] all-risk |
| US import duty | $2,400–4,500 | [R, policy-volatile] — model at the WORST plausible band (see §2.3) |
| MPF (0.3464%) + HMF (0.125%) | $90 | [R] |
| Customs broker + ISF | $750 | [R] ~$650 + $100 ISF |
| US port handling / drayage to warehouse | $450 | [R] |
| Recon / detail / safety / minor fix (often omitted — DON'T) | $1,200 | [E] stress-test fix: recon was missing from naive stacks |
| Total landed + ready-to-sell | ~$24,500–26,600 | |
| Sale price | $27,500 | |
| Contribution / car | ~$900–3,000 | after recon + worst-case duty |
The brutal honesty: at the worst duty band + with recon included, a thin flip can net under $1,000 — the stress-test "FX/duty eats a flip" scenario. The naive "~$6,200/car, 29% gross" only holds at ~2.5% duty + no recon. Bid only on spreads that survive ~15% duty + a buffer. At the favorable duty band (~2.5%), the same car nets ~$2.9–3.0k. The arb is real but margin-thin per unit — volume + decorrelation is what makes it work, not any single hero flip.
2.2 HER0 / MUTT build (the low-frequency, high-R niche sleeve)
| Line | Amount [E] |
|---|---|
| Base car (donor — salvage/cheap or already-25yr import) | $9,000 |
| Parts (engine/drivetrain/suspension/widebody/wheels) | $16,000 |
| Labor (sub-contracted — DO NOT own a shop early) | $9,000 |
| Recon/paint/finish | $3,000 |
| Cost-overrun reserve (15% — builds always overrun) | $1,200 |
| Total all-in | ~$38,200 |
| Sale price | ~$58,000 |
| Contribution / build | ~$16,300 (~28–34% gross) |
Risk flag: mod-value uplift runs on gut priors until you have tens of comparable sales logged in the engine. Until then, builds are the inventory that gets stuck (the max-consecutive-loss analogue). EPA hard rule: any emissions-relevant engine swap happens only on a car already 25+ AND already on US soil. Japan-side work = cosmetic/chassis/suspension only.
2.3 Duty: treat it as an FX-regime variable, not a constant
| Era | All-in duty, 25yr Japan-origin car | Status |
|---|---|---|
| Pre-2025 | ~2.5% | base MFN |
| Aug 2025 (US-Japan deal) | ~15% reciprocal (incl. 2.5% base) | struck down Feb 2026 |
| Feb 20 2026 | SCOTUS voids IEEPA reciprocal tariffs | — |
| Feb 28 2026 → ~late July 2026 | Section 122 ~10% layer; passenger cars currently 2.5% + ~10% | 150-day sunset; live during your likely first run [R, policy-volatile] |
Engine rule (hard-code it): max_bid = expected_US_sale − landed_overhead − target_margin − risk_haircut, and reject any bid whose spread is thinner than the current duty-volatility band (~15% worst case + buffer). Re-verify HTS 9903.94.04 + Section 122 + any Section 301 follow-on with the broker per entry. Keep a duty-contingency cash reserve.
3. Simple P&L, runway, break-even
3.1 Steady-state monthly fixed costs (lean, WA dealer location live)
| Item | Monthly [E/R] |
|---|---|
| Warehouse/dealer lease (NNN, ~7,500 SF @ ~$12/SF) | $7,500 [R] |
| Lot staffer (mandatory WA staffed hours) | $3,200 [E] |
| Data feeds (Marketcheck Basic + JP auction data + scrapers) | $700 [R] |
| Insurance (garage/dealer/driven inventory) | $1,200 [E] |
| Software/web/SEO tooling | $400 [E] |
| Misc (utilities, phone, supplies) | $600 [E] |
| Total fixed (pre owner-comp, pre racing) | ~$13,600/mo |
Racing (~$10k/yr ring-fenced) and the analytics contractor are funded separately and explicitly cut first if trading underperforms.
3.2 Break-even
- Per-flip contribution (mid band, recon included): ~$2,000/car [E] — note: NOT the optimistic $6.2k; the stress-test forces the conservative number.
- Break-even ≈ $13,600 ÷ $2,000 ≈ ~7 flips/month at the conservative duty/recon assumption, or ~2.2 flips/month only if you hit the favorable ~2.5% duty band + $6.2k contribution. Plan to the conservative number.
- A blended month (4 flips + 1 build) at conservative contributions ≈ 4×$2,000 + 1×$16,300 = ~$24,300 contribution → covers fixed ~1.8×.
3.3 Runway / cash cycle (the real risk)
- A $16k car ties up ~$25k landed for a 6–10 week cycle before any sale; the first auction-to-sale cycle is 3–5 months — longer than a naive 3.3-month buffer. That is why §1.1 sets a 6-month / $48k buffer and caps run 1 at 3–4 cars.
- Cap simultaneous in-flight capital to a fraction (~50%) of the inventory float so a stuck title or slow first run can't trigger a cash crunch.
- Let trading cashflow fund the float top-ups — never force-sell inventory at a loss to make payroll.
4. Org design & hiring sequence
Principle: operator stays brand + capital + driver. Everyone is vendor/contract until a single car-flow line proves repeatable; then convert the bottleneck role to payroll.
4.1 Vendors first (engage in pre-ops)
| # | Vendor | Why | Cost [R/E] |
|---|---|---|---|
| V1 | JP auction agent (flat-fee) | Proxy-bids; foreigners can't log in directly | ~¥70k/car + ¥100k refundable deposit [R] |
| V2 | JP exporter agent | De-reg, export cert, books ocean | bundled [R] |
| V3 | Freight forwarder | RoRo/container, ISF | RoRo $900–2,000/car [R] |
| V4 | 25-yr-savvy US customs broker | Files HS-7 + EPA 3520-1; confirms duty per entry | ~$650 + ISF $100 [R] |
| V5 | Import/customs attorney | Zoning, EPA/DOT, transfer pricing | pre-ops retainer [E] |
| V6 | International CPA | B&O/sales-tax nexus, WA↔Japan transfer pricing | pre-ops retainer [E] |
| V7 | Analytics/data contractor | Builds the valuation engine v0 before run 1 | part of $22k engine budget |
4.2 Staff (convert as volume proves out)
| # | Role | Trigger to hire |
|---|---|---|
| H1 | WA lot staffer | Day one if selling retail in WA (staffed-hours rule) — not deferrable |
| H2 | Assistant | After run 1 closes |
| H3 | Analytics/data hire (FT) | Convert from contractor once feedback data accumulates |
| H4 | Marketer (system-owner, not do-everything) | After brand identity locked; contractors underneath |
| H5 | Logistics coordinator / Japan-resident GK rep | Only at GK formation (Stage 5) |
| H6 | MN retail/build staff | At MN expansion |
Stress-test fix (the hiring contradiction): you cannot make the data-driven max-bid engine the centerpiece that prices car #1 and defer the only person who builds it. Engage the analytics contractor in PRE-OPS so car #1's bid is data-set, not a gut number. Convert to FT later.
5. Readiness checklist (gate each before the next)
Pre-ops legal/financial spine:
- Decide WA-full-dealer-buildout vs dealer-entity-in-a-wholesale-tier-state + WA-warehouse-only (CPA + attorney signoff). This decision comes BEFORE the warehouse lease.
- Verify parcel-specific zoning allows vehicle sales + repair/build shop in the same premises (silent kill if not).
- WA dealer license: $30k surety bond, enclosed building, permanent signage, listed phone, 8-hr WSIADA course, criminal-history statement.
- Continuous customs bond ($50k face) + EIN registered for AES (USDM-export spine).
- Attorney designs WA-LLC ↔ future-Japan-GK transfer pricing before any intercompany sale.
- CPA confirms WA B&O + sales-tax treatment (note WA MV sales-tax rose 0.3%→0.5% Jan 1 2026 [R]).
Funding gate (the most important behavioral fix):
- Segregate 12 months of car fixed costs in cash from realized/withdrawn live trading P&L before spending on cars or racing. Do not launch on "trading should fund it." Racing = first bucket cut if trading underperforms.
Engine:
- Valuation engine v0 live (Manheim MMR pending license; Marketcheck + eBay sold + JP auction data + NHTSA vPIC) with duty modeled as a volatile band, and hard vetoes: sub-anniversary build-MONTH, grade R/RA for resale stock, exemption-voiding swaps.
Run-1 vendors:
- JP agent, exporter, forwarder, 25-yr customs broker contracted; 2–3 written quotes each.
6. The "Lighthouse for cars" — pricing/valuation engine (the centerpiece)
Direct 1:1 port of the trading architecture. Build it in pre-ops; it is the durable moat over gut-number flippers.
- Atlas → terrain: one normalized row per car-config (identity via free NHTSA vPIC, condition incl. JP auction grade + damage map, mods as structured
{category, part, brand, est_cost, reversible}+ mod_stack_hash, market/region/season/channel, outcome labels filling over time). - Query → one valuation: stored distribution of sold comps (median + IQR, last-90-day band), never a single book number.
- Utility function → max-bid:
max_bid = E[net_resale] − all_landed_costs − target_margin − risk_buffer;deal_score = (E[net_resale] − total_cost_at_bid) / E[net_resale]; bid only whendeal_score ≥ floor.risk_bufferscales with comp-variance (IQR), grade uncertainty (R/RA penalty), thin-comp penalty (n<10), season, and duty-volatility band. - Villain-map veto (free edge): R/RA grade, rust-belt title, value-destroying mods, sub-anniversary build month, exemption-voiding swaps → hard reject or risk-buffer subtraction.
- Feedback loop: every won/lost bid + every sale writes predicted-vs-actual + days-to-sell + realized margin; nightly re-fit of curves, mod-uplift coefficients, variance — the Atlas update step.
Data sources [R]: Manheim MMR API (license-gated — request quote at manheim.data@coxautoinc.com once licensed) + Black Book (wholesale truth); Marketcheck (US active+sold, Basic ~$299/mo); eBay Motors sold API; classic.com + BaT/Cars&Bids/PCarmarket (enthusiast/modded comps — official API where possible, throttled scrape otherwise, treat as supplemental); JapanStat / AuctionDataSearch / auctionsheetjp / your exporter's USS login (JP auction history). Cache aggressively — measure-once-query literally saves per-call fees ($0.07–0.13/call on price APIs; $8/GB residential proxies [R]).
Build path: Google Sheet (validate schema + max-bid formula on run 1) → Python + DuckDB scripts → Postgres + dbt + Metabase/Streamlit dashboard the analytics hire owns.
7. Sequenced roadmap
Stage 0 — Pre-ops (now → launch gate)
Lock the WA-vs-other-state license decision; verify zoning; retain attorney + CPA; build engine v0; segregate 12 mo of car fixed costs from withdrawn trading P&L; trademark-clear ZRR0 / HER0 / MUTT (MUTT especially may collide) before any logo/merch spend; sign run-1 vendors.
Stage 1 — First JDM run (4 cars, vendor-only Japan side)
Decorrelated basket: 2× liquid flippers (grade 4+ Supra/RX-7/Evo/STI) + 1× just-unlocked 25-yr premium (R34 '01 — verify build MONTH on the compliance plate, hard veto otherwise) + 1× drift/brand car (S15/JZX/ECR33 that doubles as content). Every bid spread must survive worst-case (~15%) duty. Front-load shipping into the USTR Section 301 RoRo-fee suspension window (through Nov 9 2026) [R] while it's cheaper. Prove one full auction-to-sale cycle before scaling anything.
Stage 2 — Retail go-live (WA)
Only after run 1 proves velocity. The retail cost is the staffed location + lot hire, not the bond. Stand up GBP/local-SEO/Map-Pack only once a real reviewable public location exists (note: Google killed the vehicle-listing rich result Sept 2025 [R] — win search via Google Business Profile + GEO/AEO + marketplace syndication, NOT schema carousels).
Stage 3 — USDM-export spine
Felony-grade compliance, not "the reverse of importing": AES/EEI mandatory for every used vehicle export; title to carrier ≥72h before departure; up to $10k/violation civil, criminal exposure for willful breaches [R]. EIN-for-AES + correct-filing forwarder before the first outbound car.
Stage 4 — Minnesota expansion (demand/seasonal node, NOT an import gateway)
Fed finished cleared cars from WA (~$1,000–1,200/car haul [R]); exploit spring/summer sports-car pricing power. $50k MN bond + Jan-1-2026 fingerprinting/background checks for ALL owners/officers [R] — a disqualifying record on any owner blocks the node. Quantify the seasonal premium in the engine before committing the bond.
Stage 5 — Japan GK + direct USS access
Recruit the Japan-resident manager BEFORE incorporating. Form GK + file kobutsusho the SAME DAY to start the mandatory 1-year USS seasoning clock; keep trading through the agent for the full 12–15 month bridge. Real physical premises (virtual office fails the police check). Do NOT pursue the Business-Manager visa early — the Oct-2025 rule demands ¥30M ($200k) paid-in capital + a JP employee + an N2 speaker [R]. Then: Japan warehouse, cheaper JDM-shop builds (cosmetic/chassis only pre-ship), USDM-inbound.
Stage 6 — Expansion
Korean value/volume + parts sleeve (decorrelates from JDM collector pull); EURO 80s–2000s niche; trucks later (model the 25% "chicken tax" first — kills naive truck arb [R]); other countries/vehicle types.
8. Honest risk register (folded from the stress-tests)
| Risk | Severity | Mitigation |
|---|---|---|
| Funding spine unproven — Lighthouse is in its first live verdict month | HIGH | Gate launch on withdrawn-cash milestone; racing cut first |
| Duty roulette — rate moved 3× in 6 mo, Section 122 sunsets ~late July 2026 | HIGH | Verify per shipment; reject thin spreads; duty-contingency reserve |
| WA wholesale fiction — abolished 2019; full buildout is day-one | HIGH | Decide license state before lease; budget staffed location now |
| EPA swap trap — JP engine swap voids exemption, can strand a build | HIGH | Swaps only on US soil, post-entry, on 25+ cars |
| Per-build-MONTH seizure (not a fine) | HIGH | Hard veto in engine; verify plate; never let exporter self-clear |
| Founder-bandwidth collapse — 5 deep workstreams at once | HIGH | Sequence, don't parallelize; build founder-independent content lines |
| Cash cycle > buffer — 3–5 mo first cycle | MED-HIGH | 6-mo buffer; cap in-flight capital; small run 1 |
| GK single-point-of-failure (resident rep) | MED | Backup rep; agent for runs 1–3 |
| USDM-export felony exposure ($10k+/violation) | MED | Export spine before first outbound car |
| Build cost-overrun / stuck inventory | MED | Sub-contract labor; 15% overrun reserve; villain-map veto |
| FX (JPY/USD) on 3–5 day payment window | MED | Pre-position JPY; model FX as a regime variable |
| Trademark collision on MUTT | MED | Clear before any logo/merch spend |
| Zoning silent kill | MED | Parcel-specific verification before lease |
9. Final recommendation
Conditional GO — narrow. With $350k, the responsible scope is: JDM flips first, runs of 3–4 cars, worst-case-duty-survivable bids, builds deferred until the engine has feedback data, racing ring-fenced and trading-funded, and the WA license/location decision + valuation engine + funding gate all settled in pre-ops. The end-state the operator wants — over-qualified, delegated, brand-focused, engine-runs-itself — is the right target, but it must be sequenced into existence, not assumed alongside a simultaneous racing + four-sub-brand + live-trading launch. The single highest-leverage discipline: stop treating duty as a stored constant, and gate the whole launch on withdrawn trading cash. Those two behaviors protect more capital than anything else in this document.
Key sources
WA dealer license/bond: dol.wa.gov, suretybonds.com (wholesale-eliminated-2019), acvauctions.com · Duty/tariff: cbp.gov §232 FAQs, federalregister.gov 2025-17908, providecars.co.jp Section-122, wcshipping.com classic-car-exemptions · EPA/DOT: nhtsa.gov importation-FAQs, epa.gov importing-vehicles · JDM import: projectjdm.org 2026 guide, 33gears.com 25-year-rule · Japan ops: ussnet.co.jp, small-business-japan.com kobutsusho, mailmate.jp GK costs, housingjapan.com Business-Manager-visa · Shipping: wcshipping.com RoRo + §301 suspension, ustr.gov · Valuation data: developer.manheim.com, marketcheck.com/apis/pricing, classic.com, auctiondatasearch.jp, vpic.nhtsa.dot.gov · USDM export: cbp.gov motor-vehicle export, linearshipping.com AES · MN: dps.mn.gov dealers · Racing: news.formulad.com 2026 Pro-Am reorg, driftschoolevergreen.com