phanijapps/yf-data

YF Data

Overview

Use direct yfinance calls as the data foundation. Build reusable, validated datasets for downstream technical, fundamental, options, and risk workflows.

Workflow

1. Validate symbols and market type.
2. Pull data with yf.download first; fall back to Ticker.history when needed.
3. Normalize columns to open/high/low/close/adj_close/volume.
4. Align timezone to UTC and sort index ascending.
5. Detect and log gaps, duplicates, and stale end timestamps.
6. Persist outputs in runtime-defined formats and locations.

Minimal Pattern

import yfinance as yf
import pandas as pd


def fetch_ohlcv(symbols, start=None, end=None, period="1y", interval="1d"):
    raw = yf.download(
        tickers=symbols,
        start=start,
        end=end,
        period=None if start or end else period,
        interval=interval,
        auto_adjust=False,
        actions=True,
        progress=False,
        group_by="ticker",
        threads=True,
    )
    if isinstance(symbols, str):
        raw = raw.rename(columns=str.lower)
        raw.columns = [c.replace("adj close", "adj_close") for c in raw.columns]
        raw = raw.sort_index()
        raw.index = pd.to_datetime(raw.index, utc=True)
        return {symbols: raw}

    out = {}
    for symbol in symbols:
        df = raw[symbol].copy()
        df.columns = [c.lower().replace("adj close", "adj_close") for c in df.columns]
        df = df.sort_index()
        df.index = pd.to_datetime(df.index, utc=True)
        out[symbol] = df
    return out

Collaboration Contract

• Coordination role: define symbols, timeframe, interval, and required fields.
• Quant role: return normalized data plus a data-quality summary.
• Research role: add exchange-calendar context when gaps or anomalies appear.
• Reporting role: include data-quality caveats before higher-level conclusions.

Quality Gates

• Reject empty frames.
• Reject frames with unsorted or duplicate timestamps.
• Flag if the latest candle is older than expected for the interval.
• Record split/dividend rows when actions=True.

References

Read references/yfinance-recipes.md for ticker-format rules and fallback logic by asset class.