finances — Personal Finance Visibility
Type: Personal Status: Planning — plumbing in flight Created: 2026-05-24
What is this?
Build repeatable transaction-level visibility into personal cashflow (TD chequing + TD Visa to start), so monthly/quarterly reviews run as a quick collaborative analysis instead of a fresh manual download-and-script cycle each time.
Motivation
Years of ad-hoc R scripts at /Users/davidtingle/Code/2105.finances/ proved the analytical value, but the pipeline (manual CSV downloads, regex categorization rewritten each month) never generalized into a regular cadence. Goal is to make the data flow automatic so the time is spent on insight, not ingestion.
What does success look like?
- Transaction-level data lands in a normalized local store on a regular schedule, no manual download steps
/finanswers questions like “what’s my current savings rate?” or “how has dining out trended?” in seconds, against a categorized dataset- Categorization rules live in version control and evolve incrementally, not rewritten monthly
- Historical CSV archive (2020–present) backfilled into the same store for trend continuity
Architecture (planned)
Wealthica (aggregator) → pull script (scheduled) → ~/topaz/finances/transactions.db (SQLite)
↓
categorization rules
↓
/fin queries
- Aggregator: Wealthica (Canadian-native, supports TD)
- Store: SQLite at
~/topaz/finances/transactions.db - Rules: versioned ruleset under this project (
categorization.md, to be created) - Analysis surface:
/finskill + ad-hoc collaboration
Known unknowns
- Wealthica API access path for personal/individual use vs. their formal add-on developer track
- Whether the free tier covers TD chequing + Visa, or whether a paid plan is needed
- How investment accounts (if any) get modeled — out of scope for v0
Assets
- Historical CSV archive:
/Users/davidtingle/Code/2105.finances/raw/(chequing 2020+, Visa 2021+) - Prior analysis scripts (R, regex categorization): same directory — useful as source material for the categorization rules