fire-planner: ProjectionLab parity Wave 1 — tabbed shell, year stats, goals,

income streams, Sankey cashflow, progress overlay, settings sub-pages

Wave 1 (9 features across 4 streams):

Stream A — dashboard skeleton
  1.A.1 ScenarioShell with top tabs (Plan/Cash Flow/Tax Analytics/Compare/
        Reports/Estate/Settings) + left Sidebar with Plans switcher.
  1.A.2 GET /scenarios/{id}/year-stats?year=N returning per-year metrics
        (NW, Δ NW, taxable income, taxes, eff. rate, spending, contribs,
        investment growth). YearScrubber + YearStatsPanel render the
        right-hand sidebar; URL ?year= preserves selection.
  1.A.3 FanChart gains optional `milestones` prop (lib/milestone.ts maps
        life_event.kind → emoji) + selectedYear marker line.

Stream B — goals + progress
  1.B.1 New goals_eval module: target_nw_by_year / never_run_out /
        target_real_income probability evaluation. Wired into POST
        /simulate (exact, per-path) and GET /scenarios/{id}/projection
        (approximated from persisted fan via percentile interpolation).
        GoalsSection renders pass/fail badges.
  1.B.2 GET /scenarios/{id}/progress overlays AccountSnapshot totals on
        the projection fan; ProgressPage shows variance side-panel.

Stream C — income + cashflow
  1.C.1 New IncomeStream model + alembic 0003 + CRUD endpoints. Engine
        aggregates streams into per-year inflows + taxable arrays;
        income tax routes through the jurisdiction tax engine.
        IncomeStreamsSection on Plan tab.
  1.C.2 GET /scenarios/{id}/cashflow?year=N returns sources/sinks for
        an ECharts Sankey (sums conserve). CashflowTab body.

Stream D — settings
  1.D.1 SettingsTab + sub-nav (Milestones/Rates/Dividends/Bonds/Tax/
        Metrics/Other/Notes); placeholder cards for unbuilt sub-pages.
  1.D.2 LifeEventsSection relocated to /scenarios/:id/settings.
  1.D.3 RatesSettings (Fixed/Historical/Advanced segmented + per-asset
        cards). SimulateRequest gains rates_mode, inflation_pct,
        stocks/bonds growth + dividend, stocks_allocation. New
        build_fixed_paths() in simulator. Real-return arithmetic
        verified against (1+g+d)/(1+i)−1 ≈ 5.4%.
  1.D.4 NotesSettings — markdown textarea, save-on-blur, stored in
        scenario.config_json.notes.

Backend: 238 pytest pass (+19 new), mypy + ruff clean.
Frontend: typecheck + 7 unit tests + production build clean.

Roadmap for Wave 2-N is documented in the implementation plan.

tests/test_goals_eval.py

@@ -0,0 +1,126 @@
+"""Tests for fire_planner.goals_eval — parametrised over goal kinds."""
+from __future__ import annotations
+
+from dataclasses import dataclass
+from decimal import Decimal
+
+import numpy as np
+import pytest
+
+from fire_planner.goals_eval import evaluate_goals
+from fire_planner.simulator import SimulationResult
+
+
+@dataclass
+class _Goal:
+    kind: str
+    name: str
+    target_amount_gbp: Decimal | None = None
+    target_year: int | None = None
+    comparator: str = ">="
+    success_threshold: Decimal = Decimal("0.95")
+    enabled: bool = True
+
+
+def _make_result(
+    portfolio_paths: list[list[float]],
+    withdrawal_paths: list[list[float]] | None = None,
+) -> SimulationResult:
+    """Build a SimulationResult from explicit per-path arrays."""
+    portfolio = np.asarray(portfolio_paths, dtype=np.float64)
+    n_paths, ncols = portfolio.shape
+    n_years = ncols - 1
+    if withdrawal_paths is None:
+        wd = np.zeros((n_paths, n_years), dtype=np.float64)
+    else:
+        wd = np.asarray(withdrawal_paths, dtype=np.float64)
+    tax = np.zeros((n_paths, n_years), dtype=np.float64)
+    success_mask = portfolio[:, 1:-1].min(axis=1) > 0.0 if ncols >= 3 else np.ones(
+        n_paths, dtype=bool)
+    return SimulationResult(
+        portfolio_real=portfolio,
+        withdrawal_real=wd,
+        tax_real=tax,
+        success_mask=success_mask,
+    )
+
+
+def test_target_nw_by_year_exact_count() -> None:
+    # 4 paths, 3 years. At year 2: [200, 1500, 2500, 3000]. Target ≥ £2M
+    # → 2/4 hit → probability 0.5.
+    portfolio = [
+        [1000, 500, 200, 100],
+        [1000, 1200, 1500, 1700],
+        [1000, 2000, 2500, 2800],
+        [1000, 2400, 3000, 3500],
+    ]
+    result = _make_result(portfolio)
+    goal = _Goal(kind="target_nw_by_year",
+                 name="≥ £2M at y2",
+                 target_amount_gbp=Decimal("2000"),
+                 target_year=2,
+                 comparator=">=",
+                 success_threshold=Decimal("0.4"))
+    [eval_] = evaluate_goals(result, [goal])
+    assert eval_.probability == pytest.approx(0.5)
+    assert eval_.passed is True
+
+
+def test_never_run_out_full_horizon() -> None:
+    # 4 paths over 4 years. Path 0 hits 0 at year 2. Path 1 hits 0 at
+    # year 3. Path 2 + 3 stay positive throughout.
+    portfolio = [
+        [1000, 500, 0, 0, 0],
+        [1000, 800, 600, 0, 0],
+        [1000, 900, 800, 700, 600],
+        [1000, 1100, 1200, 1300, 1400],
+    ]
+    result = _make_result(portfolio)
+    goal = _Goal(kind="never_run_out",
+                 name="don't ruin",
+                 target_year=None,
+                 success_threshold=Decimal("0.5"))
+    [eval_] = evaluate_goals(result, [goal])
+    assert eval_.probability == pytest.approx(0.5)
+    assert eval_.passed is True
+
+
+def test_target_real_income_uses_path_median() -> None:
+    portfolio = [
+        [1000, 1000, 1000],
+        [1000, 1000, 1000],
+        [1000, 1000, 1000],
+    ]
+    withdrawals = [
+        [40_000, 40_000],
+        [60_000, 60_000],
+        [80_000, 80_000],
+    ]
+    result = _make_result(portfolio, withdrawals)
+    goal = _Goal(kind="target_real_income",
+                 name="≥ £50k income",
+                 target_amount_gbp=Decimal("50000"),
+                 target_year=0,
+                 comparator=">=",
+                 success_threshold=Decimal("0.5"))
+    [eval_] = evaluate_goals(result, [goal])
+    assert eval_.probability == pytest.approx(2 / 3)
+    assert eval_.passed is True
+
+
+def test_disabled_goals_skipped() -> None:
+    portfolio = [[1000, 500, 0]]
+    result = _make_result(portfolio)
+    enabled = _Goal(kind="never_run_out", name="active")
+    disabled = _Goal(kind="never_run_out", name="muted", enabled=False)
+    evals = evaluate_goals(result, [enabled, disabled])
+    assert [e.name for e in evals] == ["active"]
+
+
+def test_unknown_kind_returns_zero() -> None:
+    portfolio = [[1000, 1500, 2000]]
+    result = _make_result(portfolio)
+    goal = _Goal(kind="not_implemented", name="???")
+    [eval_] = evaluate_goals(result, [goal])
+    assert eval_.probability == 0.0
+    assert eval_.passed is False