nihonmacro 0.1.1

Institutional-grade Python library for Japanese macroeconomic and financial data

NihonMacro 🇯🇵

Institutional-grade Python library for Japanese macroeconomic and financial data.

Built for global macro hedge funds, quantitative researchers, and financial content creators. Think "OpenBB for Japan."

Quick Start

pip install nihonmacro

import asyncio
from nihonmacro import BOJClient, fit_nelson_siegel

async def main():
    # Fetch BoJ interest rate data
    boj = BOJClient()
    rates = await boj.interest_rates(start="2024-01-01")
    print(rates)

    # Fit Nelson-Siegel yield curve
    maturities = [0.5, 1, 2, 5, 10, 20, 30, 40]
    yields = [0.01, 0.05, 0.15, 0.45, 0.75, 1.40, 1.85, 2.10]
    params = fit_nelson_siegel(maturities, yields)
    print(f"β₀={params['beta0']:.3f}, β₁={params['beta1']:.3f}, τ={params['tau']:.2f}")

    await boj.close()

asyncio.run(main())

Architecture

nihonmacro/
├── core/
│   ├── config.py          # Pydantic BaseSettings (.env management)
│   ├── rate_limiter.py    # Token bucket + Tenacity retry (429 handling)
│   ├── cache.py           # SQLite + Redis dual-layer caching
│   └── exceptions.py      # Typed exception hierarchy
├── providers/
│   ├── boj.py             # Bank of Japan Statistics API
│   ├── jquants.py         # JPX J-Quants (auth lifecycle, daily bars)
│   ├── edinet.py          # EDINET v2 (metadata only, no XBRL)
│   └── buffettcode.py     # Buffett Code normalized financials
├── scrapers/
│   ├── mof.py             # MoF JGB constant-maturity yields (CSV)
│   ├── mof_auction.py     # MoF historical auction results (XLS)
│   ├── jsda.py            # JSDA OTC bond reference stats (CSV)
│   ├── jsda_repo.py       # JSDA Tokyo Repo Rate (XLS)
│   ├── estat.py           # e-Stat govt statistics API (JSON)
│   └── jpx.py             # JPX margin trading statistics (XLS)
├── extensions/
│   ├── macro.py           # Nelson-Siegel, Tankan DI
│   └── equities.py        # Activist screener, sector divergence
├── utils/
│   └── calendars.py       # Japanese holidays & TSE trading days
└── showcase_app/          # Reflex web application
    ├── showcase_app.py
    └── components/

Providers

Provider	Data	Auth
BoJ	Interest rates, JGB yields, Tankan survey	Public (optional API key)
J-Quants	Daily OHLCV, listed info, sector indices	Refresh token → ID token
EDINET	Filing metadata, 5% ownership reports	Public (optional key)
Buffett Code	Normalized financials, screening	API key required

Scrapers (Public Data)

No API key required — scrapes directly from Japanese government and exchange websites.

Source	Client	Data
MoF	MOFClient	JGB constant-maturity yields (1974–present, 15 tenors)
MoF Auctions	MOFAuctionClient	Historical JGB auction results (14 tenor sheets, coupon/price/yield)
JSDA	JSDAClient	OTC bond reference statistics (daily, all listed bonds)
JSDA Repo	JSDARepoClient	Tokyo Repo Rate — ON through 1Y (daily from 2012)
e-Stat	EStatClient	Government statistics (CPI, GDP, labour, industrial production)
JPX	JPXClient	Margin trading statistics (weekly snapshots + historical)

from nihonmacro.scrapers import MOFClient, JSDARepoClient

# JGB constant-maturity yields
mof = MOFClient()
yields = await mof.jgb_yields_full_history()  # Polars DataFrame, 1974–present

# Tokyo Repo Rate time series
repo = JSDARepoClient()
rates = await repo.rate_history()  # daily ON, 1W, 1M, 3M, 6M, 1Y

Key Features

Nelson-Siegel Yield Curve Fitting

$$y(t) = \beta_0 + \beta_1 \frac{1 - e^{-t/\tau}}{t/\tau} + \beta_2 \left( \frac{1 - e^{-t/\tau}}{t/\tau} - e^{-t/\tau} \right)$$

from nihonmacro.extensions.macro import fit_nelson_siegel, nelson_siegel_curve
import numpy as np

params = fit_nelson_siegel(
    maturities=[0.5, 1, 2, 5, 10, 20, 30],
    observed_yields=[0.01, 0.05, 0.15, 0.45, 0.75, 1.40, 1.85],
)

# Interpolate at any maturity
fine_grid = np.linspace(0.25, 40, 200)
fitted = nelson_siegel_curve(params, fine_grid)

Tankan Diffusion Index

$$DI = %\text{Favorable} - %\text{Unfavorable}$$

from nihonmacro.extensions.macro import tankan_diffusion_index

di = tankan_diffusion_index(favorable_pct=45.2, unfavorable_pct=12.8)
# → 32.4

Activist Target Screener

from nihonmacro.extensions.equities import screen_activist_targets

targets = await screen_activist_targets(
    tickers=["8601", "8604", "6178"],
    pb_threshold=1.0,
    filing_lookback_days=180,
)

Macro-Sector Divergence

from nihonmacro.extensions.equities import sector_rate_divergence

merged = await sector_rate_divergence(start="2023-01-01", end="2025-01-01")

Configuration

Set API credentials via environment variables or a .env file:

# .env
JQUANTS_REFRESH_TOKEN=your_jquants_token
BUFFETTCODE_API_KEY=your_buffettcode_key
EDINET_API_KEY=your_edinet_key
ESTAT_APP_ID=your_estat_app_id

Showcase Website

Run the interactive Reflex showcase:

pip install nihonmacro[showcase]
reflex run

Development

pip install -e ".[dev]"
pytest                # 429 tests
ruff check nihonmacro/ tests/

Design Principles

• Polars-first: All DataFrames are pl.DataFrame. Use .to_pandas() only at pipeline end.
• Fault-tolerant: Stale cache fallback on API failure. Custom exceptions, not generic tracebacks.
• Rate-limit aware: Token-bucket + Tenacity with Retry-After header parsing.
• Type-safe: Full type hints throughout. Pydantic validation at boundaries.

License

MIT