Home › Guides

Point-in-time US stock market data for 2026: prices, VIX and news

A frozen copy of the 2026 US stock market ships with the open-source Risk-Aware Trading Agents package: daily prices, the VIX term structure and the year's news, each value as it was known at that day's close. Every backtest and every LLM trading agent run on it gives the same numbers on any machine, and the files carry SHA-256 hashes that a paper can cite.

What is inside

2026 is the test year. 2024 and 2025 are there so that slow indicators such as the 200-day EMA and the 52-week high are fully warmed up on the first trading day of 2026.

Files and columns

src/riskaware/research/data/market_2026.csv.gz (1.4 MB, 69,738 rows):

date,symbol,open,high,low,close,volume,adj,dividend,split
2026-04-08,NVDA,...

• open to close: prices as traded, adjusted for splits only;
• adj: the factor that turns them into total-return prices (adjusted close divided by close);
• dividend, split: the cash dividend and the split ratio of that day.

src/riskaware/news/data/news_2026.jsonl.gz: one JSON object per line with ts (publication time, UTC), symbol, headline, source and feed; SEC filings also carry the form items (for example items:2.02,9.01 for an earnings release) and the sec.gov link. manifest.json lists the period, the symbols and the SHA-256 of both files.

Load it

from riskaware.research.data import bundled_market

market = bundled_market()
aapl = market.frame("AAPL")        # total-return OHLCV, 2024-01-02 to 2026-11-11
traded = market.raw("AAPL")        # prices as traded, dividends, splits, adj factor
vix = market.raw("^VIX")["close"]

import pandas as pd                # or read the file directly
df = pd.read_csv("src/riskaware/research/data/market_2026.csv.gz", parse_dates=["date"])

One day at a time

The point of the data is what was known on each day. riskaware research day 2026-04-08 (or market_day over MCP) shows everything known at that close: the market regime, VIX and its term structure, fear and greed, sector breadth, the macro events of the next two weeks, and for each stock its returns, indicators, technical setups and the news published before 16:00 New York time.

2026-04-08 (Wednesday)  decided after the close (16:00 New York); orders fill at the next open, 2026-04-09
Regime: UNCERTAIN (Uncertain: hold, no selling, buy only at strong support)
VIX 21.04  VIX9D 19.16  VIX3M 22.68  VVIX 111.07  term structure contango  fear & greed 54.47
Scheduled: CPI inflation 2026-04-10, PPI producer prices 2026-04-14

What "point in time" means here

• Indicators are computed only from earlier bars; the tests change everything after a day and check that the day's values do not move.
• News counts from the first close after its publication time (13:00 on early-close days).
• VIX indices close at 16:15, a quarter of an hour after stocks; decisions are taken after the close, before the next open, when they are known.
• Futures, the dollar index and bitcoin itself are left out because they settle after the stock close.

Cite it

Cite the data version (2026-11-11) and the SHA-256 of the two files from manifest.json. The repository's CITATION.cff gives the software citation, and every experiment's own manifest records the package version and commit.

Sources and limits

• Prices come from Yahoo Finance daily bars; scripts/build_market_bundle.py rebuilds the file. Adjustments are as of the last day of the data.
• The stock lists are today's large caps, chosen with hindsight (survivorship bias): compare strategies with each other and with the equal-weight universe, not only with SPY.
• The news is one headline feed plus SEC filings, so most stories have a single source.

More guides