Three risk tiers × two capital bands
The strategies
The same evidence supports five portfolios, not one. Three risk tiers — Defensive, Moderate, Aggressive — crossed with two capital bands gives six possible cells; five are populated and one is deliberately empty. % of capital is the unit throughout this study's €20–90k band; euro figures below only illustrate what a percentage implies, never the anchor. The universe is Romanian- and European-listed only — no US venues, and no gold, cash or crypto.
④ Balanced is the recommendation for this profile at €50–90k: a 40% European-equity engine, a 26% bond floor that clears inflation after tax, a 14% Romanian-equity bucket, and a 20% satellite spread across AI/semis, defense and financials — all-in AI ≈9.1%, about 60% of the ~15% ceiling. Below ~€50k, its five-line sibling ③ Balanced-Lite buys the identical bucket exposure. Tilt to ①/② Anchor for stability; tilt to ⑤ Growth only above €50k, and only if you can state its failed inflation test from memory.
Pick your cell: capital first, then risk
Your row is how much risk you want to carry; your column is how much capital is funding it. The one exception is the bottom-left: Aggressive at small capital is a finding about what the fee arithmetic can't support (below), not a sixth policy choice on offer.
Aggressive is the only tier the rule set allows past the 20% satellite cap — but only if the excess spreads across several domains, never concentrating into AI. Domain spread needs six to eight separate foreign tickets, and BT's own fee shape (see below) punishes exactly that combination hardest at small capital: the same eleven-line, seven-domain allocation blends to a 2.44% round-trip at €20k versus 1.36% at its actual €50k floor. There is no third option: pay roughly double every other cell's fee load to keep the spread, or concentrate to three or four lines — at which point the result is arithmetically ③ Balanced-Lite under a different name. Below ~€50k the answer is to buy ③ and add lines in a pre-declared order as capital grows, never a thinned-out ⑤.
① and ② share identical bucket totals (30/46/9/15); ③ and ④ share identical bucket totals (40/26/14/20) and land on the same all-in AI to the first decimal (9.10% vs 9.11%) despite different line counts — the concentration in the small-capital variants is fee-driven, not a different risk posture.
The five strategies
Each cell breaks down into the same four buckets — stable core, bond floor, RO equity, opportunistic satellite — re-weighted by tier and re-lined by capital.
The buy list
Twelve distinct instruments cover all five strategies. Two are BVB-native click-buys today — the Fidelis EUR 10y and the BET tracker; every other line routes through BT's foreign-market (Xetra) channel. Nothing here is graded NATIVE among the foreign ETFs — BT's curated-ETF-list membership could not be verified for any of them (the page 403s to automated fetch), so confirm each ISIN in the live app before ordering.
| Instrument | Ticker | ISIN | Bucket | BT | Horizon | The case |
|---|---|---|---|---|---|---|
| Amundi Core STOXX Europe 600 | LYP6 | LU0908500753 |
Stable core | BT ~ | Structural, 3–5y+ | For: cheapest (0.07% TER) and largest broad-Europe tracker verified, full physical replication — the compounding engine in every strategy. Against: ~19.6% UK, ~13.6% France — a "European" fund carrying real French fiscal risk (OAT-Bund spread 77bp). |
| iShares Core MSCI EMU | SXR7 | IE00B53QG562 |
Stable core | BT ~ | Structural, 3–5y+ | For: the only core line that genuinely changes economic exposure — eurozone-only, no UK/CH/Nordics. Against: doubles the France exposure already in LYP6 (~78% in its top-4 countries); ASML alone is ~8.3%, the highest single-line AI look-through in the core. |
| Fidelis EUR 10y (Jul-2026) | R3607AE | RO0OCX6C4XC5 |
Bond floor | BT ✓ | Structural, hold-to-maturity (matures 2036) | For: 6.20% coupon, tax-exempt under art. 93 — the first line in this study that clears the 5.5% BNR forecast after tax, without equity risk. Against: single issuer, single 10-year tenor, bought during a government vacuum weeks before Fitch (31 Jul) and Moody's (early Aug). |
| Xtrackers II Eurozone Govt Bond 1C | XGLE | LU0290355717 |
Bond floor | BT ~ | Flexible — diversifier / liquidity | For: issuer diversification and same-day liquidity once capital allows the bond floor to split into two lines. Against: no art. 93 exemption (3%/6% CGT), running yield below Fidelis; absent at small capital because a split floor would leave a line too small to execute economically. |
| ETF BET Patria-Tradeville | TVBETETF | ROETF0000019 |
RO equity | BT ✓ | Structural, 3–5y | For: Romania's largest listed BET tracker, one trade, zero FX leg, cheap at any order size. Against: BET is ~60% energy+utilities — a sector bet dressed as a country allocation, bought at BET's all-time high; distributing, so it leaks the 16% dividend tax. |
| VanEck Defense UCITS A | DFEN | IE000YYE6WK5 |
Satellite — defense | BT ~ | Structural, 3–5y (policy horizon) | For: the most policy-locked, AI-uncorrelated tailwind verified — RO's €16.68bn SAFE agreement, NATO Ankara's €70bn Ukraine aid. Against: the equity case de-rated hard (Rheinmetall −13–18% in one session on Germany's F126 cancellation); cap-weighted, so it's a bet on incumbents mid doctrine-rotation. |
| iShares MSCI Global Semiconductors | SEC0 | IE000I8KRLL9 |
Satellite — AI/semis | BT ~ | Tactical, months–3y | For: the only verified AI/semis expression in the universe — a deliberate, capped token in every strategy that holds it. Against: ~95% AI look-through in one line; the modelled AI pop is partially realised already (Nvidia ~17% off its high); the underlying is global, not European. |
| Amundi STOXX Eu 600 Banks | LBNK | LU1834983477 |
Satellite — financials | BT ~ | Tactical, months–3y | For: the cleanest "++" in the whole factor map — the ECB hiked to 2.25%, Stoxx 600 Banks +21% in Q2. Against: synthetic, unfunded swap (BNP Paribas counterparty); a +21% quarter is itself a concentration signal, and it double-counts the ECB-hiking thesis already priced elsewhere. |
| Amundi STOXX Eu 600 Energy Screened | LOGS | LU1834988278 |
Satellite — energy | BT ~ | Tactical, months–3y | For: the direct expression of the Iran-truce reversal (Brent ~$72→~$86); the only physical, non-swap Amundi sector line in the sleeve. Against: commodity-driven, not earnings-driven — a truce or OPEC+ surprise erases the premium overnight; overlaps TVBETETF's own energy weight. |
| VanEck Uranium & Nuclear Tech A | NUKL | IE000M7V94E1 |
Satellite — nuclear/uranium | BT ~ | Structural, 3–5y+ (SMR build-out back-loaded to ~2033) | For: concentrated access to uranium mining and the nuclear buildout — no BVB-listed uranium miners exist. Against: small, thinly traded underlying names; double-counts against the AI ceiling if pitched on the datacenter-power narrative. |
| Amundi STOXX Eu 600 Industrials | LIGS | LU1834987890 |
Satellite — industrials | BT ~ | Structural, 3–5y | For: broad, cheap exposure to European capital goods and the defense supply chain, picking up Germany's fiscal expansion from the supplier side. Against: "industrials" is the most consistently negative column in the whole factor map (Hormuz input costs, the EU-US metals dispute, rare-earths risk); synthetic swap. |
| Amundi STOXX Eu 600 Healthcare | LHTC | LU1834986900 |
Satellite — healthcare | BT ~ | Structural, 3–5y | For: the satellite's only genuinely low-correlation line — European pharma/medtech barely tracks the AI, defense or energy themes dominating the rest. Against: a hard-dated negative lands almost immediately, US Section 232's 100% tariff on patented pharma from 31 Jul 2026; synthetic swap. |
Gold, cash and crypto are excluded from this universe by mandate, not by evidence: the AI-bubble stress test found gold the only net-positive hedge line, so every strategy above runs a thinner hedge stack than a universe that permitted it would.
The dials — how the five compare
As you move Defensive → Aggressive, the bond floor narrows from 46% to 8%, the satellite grows from 15% to 54%, and lines rise from 4 to 11. All-in AI moves too — 3.5% up to 9.8% — but never gets close to the ~15% ceiling; even the tier built to spend the most leaves ~5 points of headroom unspent. At small capital, the platform's fee grid narrows the choice, not the AI rule — covered next.
Hard rails, all five: Romanian + European-listed only, no US venues · all-in AI ≤ ~15% look-through, netted · no strategy is ever an AI-only bet · opportunistic satellite ≤20% for Defensive and Moderate; Aggressive may exceed it only by spreading across domains, never concentrating into AI · accumulating UCITS, funded in EUR · every ISIN verified live on 2026-07-17.
Why five strategies, and why one cell is empty: the fee grid
BT Trade's own cost structure decides the line count, not risk appetite. BVB-native and Fidelis lines cost 0.8% with no minimum fee — a small RO order isn't meaningfully eaten. A foreign (Xetra) line costs roughly a fixed ~€20 plus 0.3% of order value — a fixed cost that a small order can't spread as cheaply as a large one: the same €400 line costs ~5.3% one-way at €20k and ~0.7% at €50k.
| Capital | 3 foreign lines | 6 foreign lines | 8 foreign lines | 10 foreign lines |
|---|---|---|---|---|
| €20,000 | 1.8% | 3.3% | 4.3% | 5.3% |
| €50,000 | 0.9% | 1.5% | 1.9% | 2.3% |
| €90,000 | 0.6% | 0.97% | 1.2% | 1.4% |
One-way cost of splitting a 20%-of-capital satellite across N foreign lines, at a
~€20-fixed + 0.3%-variable Xetra fee shape. Source: bt-instrument-shortlist.md §6.1.
| ① Anchor | ② Anchor+ | ③ Bal-Lite | ④ Balanced | ⑤ Growth | |
|---|---|---|---|---|---|
| Own capital floor | €20k | €50k | €20k | €50k | €50k |
| Blended round-trip fee cost | 1.11% | 1.00% | 1.28% | 1.15% | 1.36% |
Verified in Python against each strategy's own line count and capital floor. Four
of five cluster around 1.0–1.3%; ⑤ is structurally the most expensive to execute even at its
correct size, which is also why it has no cheaper sibling. Source: process/5-strategies/README.md.
What actually binds, in order: (1) the fee grid — it decides how many lines a cell can hold, which is why this page has five strategies instead of three, and why one cell is empty; (2) the ≤20% satellite cap on Defensive and Moderate — genuinely binding, ③ and ④ sit exactly at 20; (3) the ~15% all-in-AI ceiling — slack everywhere, retained as a guardrail against future drift, not a live constraint today. Our read of the AI-bubble evidence is holding AI down here, not the ceiling. If that read is wrong, the ceiling still permits roughly 5 points more, and the instrument to buy it (SEC0) is already in every strategy above.
Are there European government bonds reachable through BT Trade? Yes, but as a diversified EUR-government UCITS bond ETF (XGLE) — not as an individual Bund, OAT or BTP, which is broker-assisted rather than a retail self-service flow and whose coupon is taxable in Romania. For buy-and-hold, the tax-exempt Fidelis EUR usually out-yields the UCITS after tax anyway.
The four-voice debate behind them
Each decision was argued from four angles, with pros and cons, then reconciled. The voices agree on direction and differ on degree, which is exactly what the five cells span. See each voice react factor-by-factor in the Reaction Map.
Reconciled transcript: process/4-debate/debate.md. The reasoning is in Method & Why.