# Budget 2026 Investor Action Matrix — Build Spec v2

**Status:** Handover draft for build agent (revised)
**Version:** 2 (supersedes v1)
**Owner site:** https://factual-au.setiyaputra.me
**Companion app:** https://australia-cgt-reform-calculator.setiyaputra.me
**Deliverable:** Markdown content files + a JSON data file conforming to `schema.json`. No UI in this scope.

**Changes from v1:**
- Archetypes reduced from 6 to 4 (passive, property, active, founder)
- FIRE / rentvester / trust_fo absorbed into `active`; trust-structure concerns moved to a per-cell flag
- Voter cohort tagging added to each cell via `salientFor.voterCohorts`
- Political actor axis dropped (party tactics deemed out of scope)
- Cell count: 180 → 120

---

## 1. Purpose

The Budget 2026 Claims Monitor scrutinises claims about negative gearing, CGT, and startup impacts against primary sources. This matrix extends the site from "is this claim supported?" to "given the claim is partly true, what should I actually do?"

For each combination of **investor archetype** and **life stage**, under each plausible **policy outcome**, the matrix gives the reader:

1. A recommended action
2. The expected payoff direction (qualitative)
3. The regret if the policy scenario flips to a different branch
4. The single key assumption the action depends on
5. A link to the CGT calculator scenario that pressure-tests it
6. Voter cohort salience tags — which slices of the electorate this cell is most relevant to (context only, not segmentation)

Tone: decision-support framed in primary-source-aware language, not financial advice.

---

## 2. Scope and non-goals

**In scope:**
- A 4 × 5 archetype-by-life-stage grid, replicated across 6 policy scenarios = 120 cells
- Markdown long-form for each archetype and each scenario
- A single `matrix.json` containing every cell's structured payload
- Voter cohort tagging on each cell (multi-dimensional taxonomy, §3.4)
- Cross-links into existing calculator scenarioIds

**Out of scope:**
- Visual UI, components, routing, deployment
- New CGT calculator scenarios
- Personalised financial recommendations or product picks
- Political-actor game theory (party / faction moves)
- Forecasting which scenario lands

---

## 3. Axes and tagging

### 3.1 Life stages (5)

| ID | Label | Definition |
|---|---|---|
| `early_career` | Early-career accumulator (20s) | Building first portfolio; little CGT history; may rent; pre-FHB. |
| `mid_career_fhb` | Mid-career builder (30s) | Often first-home buyer or recent buyer; growing portfolio; possibly young family. |
| `peak_earner` | Peak-earner (40s) | Top-bracket income common; portfolio compounding; super contributions material. |
| `pre_retiree_bridge` | Pre-retiree, bridge phase (50s) | Building non-super assets to bridge to preservation age; CGT realisation timing critical. |
| `retiree_decum` | Retiree, decumulation (60+) | Drawing down; pension-phase super possible; lower marginal rate windows. |

### 3.2 Archetypes (4)

| ID | Label | Definition |
|---|---|---|
| `passive` | Passive investor | Buy-and-hold index/ETF investor. No alpha-seeking. Long horizon. Returns track the market. |
| `property` | Property investor | Owns one or more investment properties. May be negatively geared, may use equity to grow portfolio. Rentvesters fall here. |
| `active` | Active investor | Returns depend materially on the investor's own decisions or effort. Includes frequent traders, options/concentrated bets, FIRE-style tax-optimised accumulators, and self-managed structure users (SMSF, trusts) where the structure is the strategy. |
| `founder` | Founder / business owner | Material equity in a private operating business. Exit gain dominates lifetime wealth. Treated separately because Subdivision 152 and any founder-relief carve-out apply only here. |

**Migration note from v1:**
- `fire` → `active`
- `rentvester` → `property` (note dual ETF + IP balance sheet in the archetype brief)
- `trust_fo` → `active` (with `usesStructure: true` per-cell flag where the recommendation depends on a trust/company structure)

### 3.3 Policy scenarios (6) — unchanged from v1

| ID | Label | What it means for the planner |
|---|---|---|
| `s_announced` | Passes as announced | 50% discount replaced by indexation + 30% minimum tax from 1 Jul 2027; NG restricted to new builds from same date; existing CGT assets get a 1 Jul 2027 valuation cost-base reset. |
| `s_delayed` | Delayed past 2027 election | Start date pushed back or paused; effectively reverts to status quo through the election cycle. |
| `s_repealed` | Repealed / not legislated | Package fails or is wound back. 50% discount and current NG persist. |
| `s_founder_relief` | Founder relief carve-out added | Package passes with QSBS/BADR-style relief for qualifying founders. |
| `s_floor_dropped` | 30% floor dropped, indexation kept | Minimum tax removed; inflation-indexed cost base remains. |
| `s_hybrid` | Indexation dropped, 50% discount + floor kept | No inflation indexation, but the 50% discount survives alongside a 30% minimum effective rate. |

### 3.4 Voter cohort taxonomy (tags, not axis)

Cells carry `salientFor.voterCohorts` — describing **which slices of the electorate the action is most relevant to**, not who the action is *for*. A FIRE realisation strategy might be most salient to inner-metro professional voters and pensioners, even though the action itself is archetype/life-stage driven.

Three dimensions. A cell can carry tags from any or all three.

**Demographic** (pick zero or more):
- `gen_z` (born ~1997–2012)
- `millennial` (~1981–1996)
- `gen_x` (~1965–1980)
- `boomer` (~1946–1964)
- `pensioner` (income-support recipient, age pension or DSP)

**Economic** (pick zero or more):
- `renter`
- `fhb_aspirant`
- `owner_occupier`
- `landlord`
- `asset_rich_retiree`

**Electoral** (pick zero or more):
- `inner_metro_progressive`
- `outer_suburb_mortgage_belt`
- `regional`
- `teal_seat`
- `safe_coalition`

**Editorial discipline:** tag sparingly. A cell salient to everyone is salient to no one. Aim for 2–5 tags total per cell across the three dimensions. If a cell genuinely cuts across all cohorts, keep the array short and say so in `cohortNote`.

---

## 4. Cell payload

Each cell: one (archetype × life_stage × scenario) tuple. 4 × 5 × 6 = 120.

| Field | Type | Notes |
|---|---|---|
| `cellId` | string | Slug: `{archetype}__{life_stage}__{scenario}` |
| `archetype` | enum | One of the 4 archetype IDs |
| `lifeStage` | enum | One of the 5 life stage IDs |
| `scenario` | enum | One of the 6 scenario IDs |
| `action` | string | Imperative phrasing. Max ~25 words. |
| `actionRationale` | string | 1–3 sentences. |
| `expectedPayoff` | enum | `strongly_positive` \| `positive` \| `neutral` \| `negative` \| `strongly_negative` |
| `payoffNarrative` | string | 1–2 sentences qualitative. |
| `regret` | object | What if the world flips. See 4.1. |
| `keyAssumption` | string | Single most load-bearing assumption. |
| `calculatorAnchor` | object \| null | CGT calculator scenarioId link. |
| `verdictTone` | enum | `confident` \| `hedged` \| `speculative` |
| `usesStructure` | boolean | True if action requires trust/company/SMSF. |
| `salientFor` | object | Voter cohort tagging. See §3.4 and §4.2. |
| `cohortNote` | string \| null | 1 sentence on political salience. Null if no tags. |
| `narrativesReferenced` | string[] | Optional. Dashboard narrative IDs. |

### 4.1 The `regret` object

```
regret: {
  ifScenario: "s_repealed",
  severity: "high",
  description: "If the package is repealed, locking in gains in FY26-27 crystallised tax that could have stayed deferred."
}
```

Populate the single most material regret, not all five counterfactuals.

### 4.2 The `salientFor` object

```
salientFor: {
  voterCohorts: {
    demographic: ["millennial", "gen_x"],
    economic: ["landlord", "owner_occupier"],
    electoral: ["outer_suburb_mortgage_belt"]
  }
}
```

Each sub-array may be empty. Use `cohortNote` to explain salience in one sentence.

### 4.3 Worked example

```
active__pre_retiree_bridge__s_announced
```

- **action:** "Bring forward partial CGT realisations on bridge-phase ETFs into FY26-27 to lock in the 50% discount on existing accrued gains."
- **actionRationale:** "Existing CGT assets get a market-value cost-base reset on 1 Jul 2027. Realising before applies the current 50% discount to gains accrued to date. Post-reset, only further gains face indexation plus the 30% floor."
- **expectedPayoff:** `positive`
- **payoffNarrative:** "Locks in the cheaper-of-two-regimes treatment for the pre-2027 portion of the gain."
- **regret:** `{ ifScenario: "s_repealed", severity: "medium", description: "If the package is repealed, the realisation crystallised tax that could have stayed deferred. Cost is the time-value of paid tax." }`
- **keyAssumption:** "The bridge-phase drawdown actually needs these specific parcels within 10–15 years; otherwise pre-2027 realisation is just early tax."
- **calculatorAnchor:** `{ scenarioId: "budget-2026-fire-bridge-phase-claim", label: "Open bridge-phase scenario" }`
- **verdictTone:** `hedged`
- **usesStructure:** `false`
- **salientFor:** `{ voterCohorts: { demographic: ["gen_x", "boomer"], economic: ["asset_rich_retiree"], electoral: ["inner_metro_progressive", "teal_seat"] } }`
- **cohortNote:** "Most relevant to professional-class pre-retirees in inner-metro and teal seats who hold material non-super ETF balances and read tax change as a personal-portfolio question."
- **narrativesReferenced:** `[]`

---

## 5. Calculator anchors — known scenarioIds

| scenarioId | Best fit for |
|---|---|
| `budget-2026-young-etf-home-deposit-claim` | ETF saving toward a deposit; first-home pathway |
| `budget-2026-zero-cost-base-business-claim` | Founder exits with no relief, top-rate, post-2027 |
| `budget-2026-fire-bridge-phase-claim` | Bridge-phase realisation timing |
| `budget-2026-young-founder-net-exit-claim` | Founder gross-to-net target with top-rate |
| `budget-2026-subdiv152-founder-relief-claim` | Founder exit with active-business relief |
| `budget-2026-pre-cgt-assets-claim` | Pre-1985 asset transition stress test |
| `budget-2026-family-home-distortion-claim` | Established-property post-cutoff cases |
| `budget-2026-cgt-founder-claim` | Generic top-rate founder exit |
| `budget-2026-capital-vs-labour-tax-claim` | Long-held discounted gain vs marginal labour tax |
| `housing-claim-negative-gearing` | Post-cutoff established-property with NG losses |

Set `calculatorAnchor: null` when no scenarioId cleanly fits. Do not stretch.

---

## 6. Editorial rules

1. **No financial advice voice.** "Rational move under this assumption" not "you should."
2. **Hedge proportionate to confidence.** Match the claims monitor's idiom.
3. **No dollar figures in JSON.** Direction and severity only.
4. **Pre-2027 vs post-2027 mechanics must be correct.** Pre-1 Jul 2027 gains via valuation reset → 50% discount. Post-reset gains under `s_announced` → indexation + 30% floor.
5. **Super is untouched by the CGT changes.**
6. **Main residence exemption stays.**
7. **Subdivision 152 (small business CGT) is preserved.** Founder actions flag eligibility as an assumption.
8. **Pre-1985 assets:** gains accruing from 1 Jul 2027 enter the CGT net under `s_announced`.
9. **Pensioners and income-support recipients** exempt from the 30% floor.
10. **No predictions about which scenario lands.**
11. **Voter cohort tags describe salience, not segmentation.** A `millennial` tag doesn't mean "for millennials only"; it means "politically charged for millennials."
12. **Cohort tags must be defensible.** If you can't write a one-sentence `cohortNote` justifying a tag, drop it.

---

## 7. Content production sequence

1. Read https://factual-au.setiyaputra.me/dashboard for tone.
2. Read https://factual-au.setiyaputra.me/methodology for verdict taxonomy.
3. Write 4 archetype briefs. The `active` brief must enumerate the sub-types (FIRE, frequent trader, structure-driven) and how the matrix collapses them.
4. Write 6 scenario briefs.
5. Write `cohorts.md` documenting all 15 cohort tags.
6. Generate 120-cell `matrix.json` following `schema.json`. Order: by archetype, then by life stage, then by scenario.
7. **Differentiation check:** for each (archetype × life_stage), actions must differ across at least 4 of the 6 scenarios.
8. **Cohort tagging pass:** median cell should carry 2–4 tags total across all dimensions.
9. **Calculator anchor check:** verify every `scenarioId` exists in §5.

---

## 8. File layout

```
/matrix/
  SPEC.md
  schema.json
  matrix.json
  archetypes/
    passive.md
    property.md
    active.md
    founder.md
  scenarios/
    s_announced.md
    s_delayed.md
    s_repealed.md
    s_founder_relief.md
    s_floor_dropped.md
    s_hybrid.md
  life_stages/
    life_stages.md
  cohorts/
    cohorts.md
```

Each Markdown file: frontmatter with `id`, `label`, `summary` (≤ 240 chars), `lastReviewed` (ISO date).

---

## 9. Acceptance criteria

- [ ] `matrix.json` validates against `schema.json`
- [ ] Exactly 120 cells; every cellId unique
- [ ] Every `calculatorAnchor.scenarioId` referenced exists in §5
- [ ] All required fields populated (nullable: `calculatorAnchor`, `cohortNote`, `narrativesReferenced`)
- [ ] No cell contains a dollar figure
- [ ] No cell uses "you should" framing
- [ ] For each (archetype × life_stage), actions differ across ≥ 4 of 6 scenarios
- [ ] Median cell has 2–4 cohort tags total
- [ ] Every cell with cohort tags has `cohortNote`; every cell without has `cohortNote: null`
- [ ] All 4 archetype Markdown files exist
- [ ] All 6 scenario Markdown files exist
- [ ] `cohorts.md` exists documenting the 15 tags

---

## 10. Open questions for srvzt

1. The `active` archetype is heterogeneous. Brief can flag sub-types, but cells must collapse them. Is there a sub-type whose action would *differ materially* from the collapsed answer that we should split out? (My instinct: no.)
2. Cohort tags lean political. Add a `cohortStakes` field to capture *why* the cohort cares (economic interest vs identity vs ideology)? Adds editorial overhead but makes political salience more legible.
3. `cohortNote` is one sentence by design, but cells like `property__mid_career_fhb__s_announced` are politically loudest. Allow up to 3 sentences when `verdictTone: speculative`?
4. The separate 30% minimum tax on discretionary trust distributions from 1 Jul 2028 affects `active` cells with `usesStructure: true`. Second regret object, or fold into `keyAssumption`?

---

## 11. References

- Budget Paper 2, pp.21–22 — CGT and NG policy text
- Budget 2026 tax explainer — transition mechanics, pre-1985 asset treatment
- ATO: Small business CGT concessions (Subdiv 152)
- ATO: Main residence exemption
- ATO: Tax rates — Australian resident