Clarion Portfolio Monitor

Clarion Portfolio Monitor

Fetches live portfolio data from TastyTrade (balances, positions, P/L, daily transactions) Stores every snapshot in DuckDB for auditable performance history (cost basis, mark price, NLV per position per day) Integration points: Portfolio Manager (live data + cost basis), LP Voice (performance claims for investor letter)

Runs as a on-demand script — no network exposure, no persistent MCP server. Reads credentials from Zo secrets, authenticates with a refresh token (never expires), and writes a timestamped snapshot.

What it fetches

• Account balances: net liquidating value (NLV), cash balance, buying power, margin equity
• Open positions: symbol, quantity, direction, mark price, cost basis, unrealized P/L, realized day P/L
• Today's transactions: fills (buys + sells), dividends, cash events
• NLV history: trailing 30-day NLV for trend context

Output

Prerequisites

1. A TastyTrade OAuth application created at https://my.tastytrade.com/app.html#/manage/api-access/oauth-applications
2. A refresh token generated via OAuth Applications > Manage > Create Grant
3. Both stored as Zo secrets in Settings > Advanced:
   • TASTYTRADE_CLIENT_SECRET — the OAuth app client secret
   • TASTYTRADE_REFRESH_TOKEN — the generated refresh token

Refresh tokens never expire. One-time setup.

Usage

python /home/workspace/Skills/clarion-portfolio-monitor/scripts/fetch.py

Options:

• --json-only — skip markdown summary, produce only JSON
• --account N — specify account number (auto-detects if omitted)
• --no-history — skip NLV history fetch (faster)

Trade ledger (cost-basis verification)

scripts/trade-ledger.py maintains a transactions table in ~/clarion/portfolio/portfolio.duckdb, backfilled from TastyTrade full account history and upserted idempotently by transaction ID. It derives per-symbol cost basis and realized P/L from fills via FIFO lot replay, and reconciles against the broker snapshot and thesis YAML.

python trade-ledger.py sync [--days N] [--full]   # fetch + upsert transactions
python trade-ledger.py positions                  # derived positions from ledger
python trade-ledger.py lots [--markdown] [--json] # open lots per position
python trade-ledger.py verify                     # ledger vs broker vs thesis YAML
python trade-ledger.py realized                   # realized P/L by symbol

• First sync (or --full) backfills from CLARION_LEDGER_INCEPTION (env var, ISO date, default 2000-01-01 — set it to your account's inception to keep backfills fast). Subsequent syncs default to the last 7 days.
• fetch.py automatically appends open-lot detail to the markdown summary when the ledger is available; it degrades gracefully when it isn't.
• Uses the same TastyTrade secrets as fetch.py. verify and positions are offline (no API call).

Scheduling

For daily snapshots, create a Zo Automation:

• Rrule: FREQ=DAILY;BYHOUR=17;BYMINUTE=30 (after market close Eastern)
• Instruction: Run python /home/workspace/Skills/clarion-portfolio-monitor/scripts/fetch.py and report any positions with >5% daily move.

Integrating with other Clarion skills

• Clarion LP Voice (clarion-living-letter-update): reads ~/clarion/portfolio/latest.json for the "What We Did" section and NLV context
• Clarion Portfolio Manager (clarion-thesis-monitor): compares live positions against active theses to detect thesis drift
• Clarion Macro Sentinel (clarion-regime-check): not directly integrated — portfolio data is downstream of regime decisions