Central Bank Policy Dashboard — Python/Flask Backend

A lightweight data API that fetches recent monetary policy decision statements from central banks websites and makes as summary using Claude Haiku. Chart data from public sources. https://bekindrewind.up.railway.app

Architecture

Browser (cb-policy-dashboard.html)
    │
    │  GET /api/data/pol
    ▼
Flask API  (app.py)
    │
    ├── assembler.py          merges 4 series into monthly rows
    │       │
    │       ├── fetchers/fred.py        FRED API  (rates + most yields)
    │       ├── fetchers/ecb.py         ECB Data Portal  (ECB rate + HICP)
    │       └── fetchers/eurostat.py    Eurostat  (CEE country CPI)
    │
    └── cache.py              file-based JSON cache (12h TTL)
                              stored in  .cache/

Data sources by series

Series	Fed	ECB	BoE	BoJ	Poland	Hungary	Romania	Czech
Policy rate	FRED	ECB Portal	FRED	FRED	FRED	FRED	FRED	FRED
CPI	FRED	ECB Portal	FRED	FRED	Eurostat	Eurostat	Eurostat	Eurostat
2Y yield	FRED	FRED	FRED	FRED	FRED	FRED	FRED	FRED
10Y yield	FRED	FRED	FRED	FRED	FRED	FRED	FRED	FRED

---

Setup

1. Prerequisites

• Python 3.11+
• A free FRED API key → https://fred.stlouisfed.org/docs/api/api_key.html
• A ANTHROPIC API key

2. Install

cd cb-api
python -m venv venv
source venv/bin/activate        # Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Configure

cp .env.example .env
# Edit .env and paste your FRED_API_KEY

4. Run (development)

python app.py
# → http://localhost:5000

5. Run (production)

gunicorn app:app --bind 0.0.0.0:5000 --workers 2

---

API reference

GET /api/banks

List all available central bank keys.

{
  "banks": [
    { "key": "fed", "name": "Federal Reserve (Fed)", "currency": "USD", "cpi_target": 2.0 },
    { "key": "pol", "name": "National Bank of Poland (NBP)", "currency": "PLN", "cpi_target": 2.5 },
    ...
  ]
}

---

GET /api/data/<cb_key>

Returns assembled monthly chart data for one central bank.

Path params

• cb_key — one of: fed, ecb, boe, boj, pol, hun, rom, cze

Query params

Param	Type	Default	Description
start	YYYY-MM-DD	2021-01-01	Earliest data point
refresh	0|1	0	Force cache refresh

Example

GET /api/data/pol?start=2022-01-01

Response

{
  "cb": "pol",
  "name": "National Bank of Poland (NBP)",
  "currency": "PLN",
  "cpi_target": 2.5,
  "start_date": "2022-01-01",
  "count": 36,
  "data": [
    { "date": "2022-01", "rate": 2.25, "cpi": 9.4, "y2": 4.10, "y10": 4.05 },
    { "date": "2022-02", "rate": 2.75, "cpi": 8.5, "y2": 4.50, "y10": 4.40 },
    ...
  ]
}

• data[].rate — policy rate (%)
• data[].cpi — CPI / HICP YoY (%)
• data[].y2 — 2Y government bond yield (%)
• data[].y10 — 10Y government bond yield (%)
• Any value may be null if the upstream source has not yet published data for that month.

---

GET /api/meta/<cb_key>

Static metadata: series labels, source APIs, series IDs.

---

GET /api/status

Health check + cache inventory.

{
  "status": "ok",
  "available_banks": ["fed", "ecb", "boe", "boj", "pol", "hun", "rom", "cze"],
  "cache": {
    "entries": 4,
    "ttl_seconds": 43200,
    "files": [
      { "key": "fred_FEDFUNDS_2021-01-01", "age_seconds": 3600, "expires_in_seconds": 39600 }
    ]
  }
}

---

DELETE /api/cache

Clears all cached files. Useful after correcting upstream data issues.

---

Connecting the frontend

In cb-policy-dashboard.html, replace the hardcoded CB_DB object with a live fetch. Find the switchCB function and change:

// BEFORE (hardcoded)
function switchCB(key) {
  const cb = CB_DB[key];
  buildChart(cb.data);
  ...
}

// AFTER (live API)
const API_BASE = "http://localhost:5000";

async function switchCB(key) {
  showLoadingState();
  const res  = await fetch(`${API_BASE}/api/data/${key}`);
  const json = await res.json();
  currentData = json.data;
  buildChart(currentData);
  buildKPI(json);
  ...
}

See cb-policy-dashboard-live.html for the fully wired version.

---

Adding a new central bank

1. Add an entry to CB_CONFIG in config.py with the four series codes.
2. No other code changes needed — the assembler and app pick it up automatically.

Caveats

• FRED's OECD-sourced CEE bond yield series (IRLTST01XXM156N, IRLTLT01XXM156N) may have a 1–2 month lag vs. market data. For real-time bond yields, a paid data provider (Refinitiv, Bloomberg, Nasdaq Data Link) is needed.
• Romania 2Y yields from FRED may have gaps; the assembler forward-fills these.
• The ECB deposit facility rate series starts in 1999; only post-2021 data is used by default.