How to replace yfinance in a script

The short way

Don't rewrite the script; give it the DataFrame it already expects. Most yfinance code reads yf.download(...) once and then never touches yfinance again, so the port is one function. Here is the shape yfinance 1.7.0 hands back for a single ticker on 4 October 2026, and an adapter that returns the same shape from Tiingo:

import pandas as pd
import requests
import yfinance as yf

before = yf.download("AAPL", start="2020-01-01", end="2026-09-01")
# columns: MultiIndex ('Close', 'AAPL'), ('High', 'AAPL'), ... — even for one ticker
# index:   'Date', timezone-naive; prices adjusted for splits AND dividends

YF_NAMES = {"adjOpen": "Open", "adjHigh": "High", "adjLow": "Low",
            "adjClose": "Close", "adjVolume": "Volume"}

def daily(ticker, start, end, token="YOUR_API_KEY"):
    url = f"https://api.tiingo.com/tiingo/daily/{ticker}/prices"
    rows = requests.get(url, params={"startDate": start, "endDate": end,
                                     "token": token}).json()
    df = pd.DataFrame(rows)
    df["date"] = pd.to_datetime(df["date"], utc=True).dt.tz_convert(None)
    df = df.set_index("date").rename_axis("Date").rename(columns=YF_NAMES)
    return df[["Open", "High", "Low", "Close", "Volume"]]

after = daily("AAPL", "2020-01-01", "2026-09-01")

Two decisions are hidden in there. Mapping Tiingo's adj* columns onto yfinance's names reproduces yfinance's default, auto_adjust=True; if your script passed auto_adjust=False, read the next section before mapping anything. And the adapter returns flat columns, which is what yf.download(..., multi_level_index=False) returns — code that indexes before["Close"]["AAPL"] needs that line changed too.

Choosing between providers is a separate question, answered in downloading historical prices in Python. This page is about the port.

What the options are

Closest drop-in for daily bars. Tiingo returns open, high, low, close and volume as traded, adjOpen through adjVolume adjusted by the CRSP method, and divCash and splitFactor on every row — which is yfinance's actions=True without a second call. History reaches 1962. Free is 50 requests an hour, 1,000 a day and 500 unique symbols a month; $30 a month is 10,000 an hour, personal and internal use.

Same coverage as Yahoo. EODHD spans 60-plus exchanges, which is the reason most yfinance users were on Yahoo in the first place. Every ticker takes an exchange suffix — AAPL.US, BP.LSE, SAP.XETRA — where Yahoo used BP.L and SAP.DE, so the symbol list needs mapping as well as the columns. The free key is 20 calls a day and the past year only; full history starts at $19.99 a month.

US only, with a different adjustment. Massive's aggregates endpoint returns o, h, l, c, v, vw and n with a t timestamp, adjusted for splits by default and for dividends never. If your script depended on yfinance's dividend adjustment, you rebuild it from Massive's dividends endpoint. The free Basic plan is end-of-day, two years back, five calls a minute.

Free, and narrower than it looks. Alpha Vantage's free key reaches TIME_SERIES_DAILY: raw as-traded prices, outputsize=compact, the latest 100 trading days. outputsize=full and the adjusted endpoint are premium, from $49.99 a month. Each day is a key in a JSON object, fields are named "1. open" to "5. volume", and every number is a string. 25 requests a day.

Fundamentals port separately and less cleanly. yfinance's income_stmt has periods as columns and line items such as Total Revenue as rows; Alpha Vantage's free INCOME_STATEMENT returns one object per period with totalRevenue as a string and missing values as the string "None". On the others it is a second purchase: EODHD's fundamentals plan has no prices in it, Tiingo's fundamentals are an add-on, and Massive's financials are a separate $29 a month below its Advanced tier.

Where this breaks

auto_adjust=False did not mean raw. This is the one that survives every test except a backtest. With auto_adjust=False, yfinance returns Close and Adj Close — and its Close is split-adjusted, only not dividend-adjusted. On 4 October 2026, Apple's close for 27 August 2020, two sessions before its four-for-one split, came back from yfinance as 125.01 with auto_adjust=False, and from EODHD's close as 500.04, which is what printed. Port a script that read yfinance's Close onto a vendor's raw close and every price before a split is several times too high. The adjusted figures did agree — 121.15 from both.

Every vendor names and adjusts differently. EODHD's JSON has raw OHLC, one adjusted_close for splits and dividends, and a volume that is split-adjusted though the prices beside it are not. Its CSV — the default when you leave out fmt=json — capitalises the headers to Adjusted_close, and a misspelled fmt is silently ignored rather than rejected. Massive's adjusted=true covers splits alone. Map each column deliberately, then check one split and one dividend by hand. Why adjusted close differs explains why two correct adjusted series still disagree.

The dates move. yf.download returns a timezone-naive index for daily bars; Ticker.history returns the same days as timezone-aware midnights in America/New_York. Massive's t is the Unix millisecond start of the bar in Eastern Time, so pd.to_datetime(t, unit="ms") gives 05:00 on the right day in winter and 04:00 in summer — and a join against a midnight-dated index then matches nothing, silently. Convert to New York time and normalise to the date before joining.

Errors that are not errors. Alpha Vantage answers a refused request with HTTP 200 and a JSON body whose only key is Information, so raise_for_status() passes it and the parser fails two lines later on a missing key. Check for the data key itself on every response.

One ticker per request. yf.download("AAPL MSFT ...") fetched a whole list in one call. Tiingo, EODHD's end-of-day endpoint and Alpha Vantage are one ticker per request, so a 500-name universe is 500 requests: ten hours on Tiingo's free tier, twenty days on Alpha Vantage's, and over the 500-symbol monthly ceiling at Tiingo the first time the list changes.

The licence is still personal. Leaving yfinance does not by itself make the data commercial. Tiingo's $30 plan is personal and internal use, and its commercial plan is $50 for one organisation; EODHD's cheapest licence that is not personal use is $399 a month; Massive's individual plans exclude business use; Alpha Vantage is personal and non-commercial by default. The difference from Yahoo is that there is now a contract to read, and a price for the use you actually have — see redistribution.

If you outgrow this

You need intraday history. yfinance stopped at 60 days of intraday bars; the paid tiers here go further, and backfilling minute bars covers the shape of that job.

You need delisted names. None of these adapters fixes a universe built from today's tickers. Backtesting on survivorship-free data is the next step.

You need events, not just prices. If the script used actions=True or Ticker.dividends, getting split and dividend history as data covers the endpoints that replace them. The rest of the field is in market data APIs.