quarq

French institutional portfolio analytics with RAG and local LLM.

pip install quarq

---

What it is

quarq is a CLI tool for institutional portfolio analysis. It combines a live data pipeline (French equities, ECB macro, FRED rates) with a RAG engine over financial documents (SFDR filings, ECB FSR, fund prospectuses) and a local LLM for plain-English narrative generation.

Your portfolio data never leaves your machine.

Status

Work in progress. v0.1.0 coming soon.

What works now

from datetime import date
from quarq.ingest.equity import EquityProvider
from quarq.ingest.fred import get_risk_free_rate
from quarq.ingest.ecb import get_ecb_rate
from quarq.ingest import get_provider
from quarq.config import load_config

# Live equity prices (.PA tickers, ^FCHI benchmark)
p = EquityProvider()
df = p.fetch("MC.PA", date(2025, 1, 1), date(2025, 12, 31))

# Risk-free rate (FRED OAT10Y, falls back to config if no key)
cfg = load_config()
rfr = get_risk_free_rate(cfg)  # e.g. 0.032

# ECB policy rate (no key required)
ecb = get_ecb_rate(cfg)  # e.g. 0.045

# Provider registry
equity = get_provider("equity")   # EquityProvider
fred   = get_provider("fred")     # FREDProvider
ecb_p  = get_provider("ecb")      # ECBProvider
oecd   = get_provider("oecd")     # OECDProvider

Set FRED_API_KEY in ~/.quarq/config.toml for live macro rates. All other providers work without a key.

Stack

• Data: yfinance, FRED, ECB SDW, OECD (direct REST, no framework dependency)
• RAG: sentence-transformers + ChromaDB (local, persistent)
• LLM: Qwen3 / Mistral via LM Studio (local inference)
• Report: Plotly 6 + Jinja2, PDF via playwright
• CLI: textual + rich

Usage (coming in v0.1.0)

quarq status quarq report --portfolio ./portfolio.toml --format pdf quarq query "What does the ECB FSR say about CAC 40 concentration?" quarq rag add ./docs/

Author

zkmarc — https://github.com/yodablocks