## Outcome

**Schema 49 is already sufficient for Sprint 20B.** The minimal path is to activate and connect the existing canonical valuation, performance-scope, coverage, FX, and ingestion components. No schema migration is justified unless tracking mode must be user-configurable and durable rather than derived from account role/source.

## Existing reusable architecture

- **Performance engine:** `ledger/performance.py`
  - `ttwror_daily_v1()` at line 205
  - `xirr_v1()` at line 382
  - `twr_v1()`, `annualize_return()`, attribution and quality primitives
  - Explicit daily-boundary and missing-FX failure behavior; no live-data fallback.
- **Performance orchestration:** `services/portfolio_performance.py`
  - `build_performance_coverage()` at line 59
  - `_load_activities()` / `load_scope_activities()` at lines 296/374
  - `_load_valuations()` and account aggregation at lines 392/542
  - `build_portfolio_performance()` at line 667
- **Scope and cash-flow evidence:** `services/performance_scope.py`
  - `set_performance_scope_classification()` at line 23
  - `set_performance_cashflow_coverage()` at line 100
  - Portfolio transfers are neutralized while account-level transfers become external flows; covered by `test_scope_aware_raiffeisen_akb_transfer...`.
- **Canonical persistence, schema 49:** `storage/migrations.py`
  - `account_value_snapshots` line 1552
  - `portfolio_valuation_snapshots` line 1741
  - `performance_cashflow_coverage` line 1801
  - Immutable `portfolio_ingestion_batches/items` lines 2015–2080
  - Existing `fx_rates`, `transactions`, scope registry, market prices, audit log
  - Canonical valuation rows already carry scope/account, original/base currency, FX rate/direction, timestamp, capture cutoff, source/reference, version/supersession and quality reasons.
- **Generic preview/confirm/import:** `services/portfolio_data.py`
  - `preview_ingestion()` line 941
  - `_insert_valuation()` lines 1064–1080
  - `confirm_ingestion()` lines 1232–1367
  - Atomic `BEGIN IMMEDIATE`, preview expiry, source revision/fingerprint validation, confirmation idempotency, immutable batch/items and audit.
  - Existing source adapters cover PostFinance, legacy account totals, cash snapshots and canonical transactions (`_records()`, lines 913–922).
- **Provider paths:**
  - PostFinance: `preview_postfinance_import()` / `confirm_postfinance_import()` in `services/postfinance_service.py`; generic ingestion additionally writes a partial canonical baseline.
  - True Wealth: preview/confirm/manual-value workflows in `services/truewealth_service.py`; confirmed values land in `account_value_snapshots` and can already be promoted through `legacy_account_values`.
  - Daily securities valuation: `services/portfolio_analytics.py::run_daily_market_valuation()` line 395; values confirmed positions using dated prices/FX plus cash and writes account/instrument canonical snapshots.
- **APIs:** `api/routers/overview.py`
  - ingestion preview/confirm/history lines 80–116
  - reconciliation lines 119–139
  - coverage lines 142–145
  - stored-data-only performance API lines 148–175
- **Consumers:** `wealth_cockpit.py::build_wealth_cockpit()` lines 978–1130 already calls the canonical performance and period-aware coverage services.

## Gaps / contract fields

1. **No generic performance `tracking_mode` exists.**
   - `accounts.balance_mode` is cash-specific and should not be reused.
   - For schema-49 Sprint 20B, derive mode from role/source:
     - `ledger_daily_market` for holdings + cash accounts
     - `provider_total_value` for True Wealth/robo/pension/total-value accounts
     - `baseline_only` where transaction history is incomplete
   - Expose the derived mode in source/coverage responses. Add persistence only if users must override it.

2. **No external daily-valuation payload contract.**
   - `IngestionPreviewRequest` only supports source selection plus PostFinance file fields; it cannot accept generic daily rows.
   - If Sprint 20B requires CSV/API daily totals, add fields such as:
     - `file_name`, `content_base64` or structured `valuations`
     - per row: `account_id`, `valuation_at`, `value_original`, `currency`, `fx_rate_to_base`, `source_reference`, `quality_status`, `reason_codes`
   - These map directly to the existing snapshot table; no migration needed.

3. **Cash-flow coverage is not integrated into generic confirmation.**
   - Add optional preview-bound evidence fields:
     - `cashflow_coverage_from/to`
     - `cashflow_coverage_status`
     - `cashflow_coverage_source/note`
   - Update coverage only after successful atomic confirmation and only when the import actually proves completeness.

4. **Coverage route omits period/account parameters.**
   - The service supports `from_date`/`to_date`, but `/portfolio/performance/coverage` calls it without them. Expose the same period used by the performance request.

5. **Daily market job is not universal.**
   - It covers canonical securities and latest cash, not provider-total-only accounts such as True Wealth.
   - `INSERT OR REPLACE` with stable IDs may conflict with immutable valuation triggers when retrying a prior partial run. Replace this with the existing append-only version/supersession insert pattern.

## Minimal implementation plan

1. **`services/portfolio_data.py`**
   - Add one `daily_account_valuations` source adapter.
   - Normalize rows into existing `SourceRecord` valuation records.
   - Reuse `_insert_valuation()`, preview hash binding, versioning, batch/items, audit and idempotency.
   - Include tracking mode and cash-flow-coverage impact in preview.

2. **`api/schemas/portfolio_data.py`**
   - Add strict daily valuation row/file fields and optional coverage evidence.
   - Bind all new fields into preview/confirm identity; do not allow extras.

3. **`services/portfolio_analytics.py`**
   - Refactor daily snapshot writes to a shared append-only canonical valuation helper.
   - Keep this runner for ledger-revalued accounts only.

4. **`services/truewealth_service.py`**
   - Prefer promoting the confirmed total directly into a canonical valuation inside the same transaction, using shared versioning/idempotency logic.
   - Lower-risk alternative: retain the current two-stage `account_value_snapshots → legacy_account_values` promotion, but expose it explicitly as pending activation.

5. **`services/performance_scope.py` / `portfolio_performance.py`**
   - Add a deterministic role-to-tracking-mode helper.
   - Surface derived mode and period-specific blockers in coverage; do not modify TTWROR/XIRR calculations.

6. **`api/routers/overview.py`**
   - Add period/account query parameters to the coverage endpoint.

### Focused tests

- Extend `test_portfolio_data_ingestion_reconciliation.py`:
  - daily-row preview is storage-free
  - stale/tampered confirm rejected
  - duplicate confirm idempotent
  - corrections append versions and supersession
  - valuation + coverage + audit commit atomically
  - missing FX is blocked/partial, never estimated
- Extend `test_portfolio_market_analytics_v1.py`:
  - partial-run retry does not hit immutable-trigger failure
  - rerun produces no duplicate active value.
- Extend `test_true_wealth_valuation_pilot.py`:
  - confirmed provider total becomes exactly one canonical valuation.
- Preserve `test_sprint14_performance_contract.py` and `test_portfolio_performance_foundation.py`.
- Add an end-to-end Sprint 20B test: daily valuations + covered external flows yield TTWROR and XIRR; baseline-only and incomplete coverage remain explicitly unavailable.

## Verification / repository state

- Confirmed branch `sprint20b/performance-activation-daily-valuations-v1` at exact SHA `7f1d76bc62e19861e55929273859e27ca6d74970`.
- Ran the four focused suites with `uv run pytest`: **78 passed, 1 deprecation warning**.
- **No files created or modified.**
- Initial bare `pytest` command was unavailable; `uv run pytest` used the repository environment successfully.