DEV Community

Johnny Nam
Johnny Nam

Posted on

Open-sourcing tcs-macro-pulse: a tiny Python toolkit for FRED + GDACS data

#data #opensource #python #showdev

TL;DR — I open-sourced the data-fetcher layer of my investment SaaS. ~1.4K LOC, MIT, pure Python. Pulls FRED macro indicators, GDACS disaster events, and runs lightweight news sentiment. Repo: github.com/TCS-PLATFORM-OFFICIAL/tcs-macro-pulse

Why I built this

I'm building TCS-PLATFORM — an investment-intelligence SaaS for Vietnamese retail investors. The system is layered (L1 macro → L9 signal generation), and the lower layers (L1-L3) are just data collection from public sources.

Locking those behind a paywall would be silly. The proprietary engine is the L4-L9 stack. So I carved out L1-L3 into a standalone package: tcs-macro-pulse.

What's inside

L1 — FRED macro fetcher

10 key US macro indicators in one call:

from tcs_macro_pulse.fetchers.fred import FREDFetcher

fred = FREDFetcher()
data = fred.fetch_key_indicators()
# → {'fed_funds_rate': 5.33, 'cpi_yoy': 3.1, 'unemployment': 3.9, ...}
Enter fullscreen mode Exit fullscreen mode

Built-in recession indicator:

spread = fred.yield_curve_spread()  # → -47.2 bps (10Y - 2Y, inverted)
Enter fullscreen mode Exit fullscreen mode

L2 — GDACS disaster events

Parses the GDACS RSS feed (no API key needed) and scores events:

from tcs_macro_pulse.fetchers.gdacs import GDACSSentinel

sentinel = GDACSSentinel()
events = sentinel.fetch_recent(days=7)
risk = sentinel.compute_risk_score(events)
# → {'risk_level': 'AMBER', 'risk_score': 25, 'n_red': 1, 'n_orange': 4, ...}
Enter fullscreen mode Exit fullscreen mode

L3 — Keyword sentiment

For when you don't want to spin up FinBERT just to skim 50 headlines:

from tcs_macro_pulse.analysis.keyword_sentiment import KeywordSentiment

ks = KeywordSentiment()
result = ks.analyze_headlines([
    "Fed signals rate cuts ahead",
    "Markets crash on inflation fears",
    "Tech stocks rally"
])
# → {'overall': 'NEUTRAL', 'bullish': 1, 'bearish': 1, 'neutral': 1, 'score': 0.0}
Enter fullscreen mode Exit fullscreen mode

The [nlp] extra installs transformers + torch if you want FinBERT instead.

Install 

pip install tcs-macro-pulse
# or
pip install "tcs-macro-pulse[nlp]"  # with FinBERT
Enter fullscreen mode Exit fullscreen mode

What's NOT in the package

  • L4-L9 signal generation (proprietary)
  • Anything Vietnam-specific (those live in the SaaS)
  • Trading execution / order management

This is just clean data fetchers. Bring your own analysis.

License & contributing

MIT. Contributions welcome — particularly fetchers for:

  • ECB (Eurostat macro)
  • BoJ (Japan macro)
  • IMF (cross-country)
  • World Bank (development indicators)

Open an issue or PR: github.com/TCS-PLATFORM-OFFICIAL/tcs-macro-pulse

If you find this useful, a ⭐ on GitHub helps a lot.

Cheers!

Top comments (0)