Historical Market Cap

Your agent pulls the market cap in force on each of your trade dates and re-applies your size cut against that number rather than against the current one.

Two numbers, dated the day they were true.

Market capitalization is share price times shares outstanding: what the whole company was worth at that moment. This dataset stores that number for each equity on each trading date, next to the share count it was computed from.

Storing it matters because both inputs move. Prices move every session, and share counts move whenever a company issues, repurchases, or converts stock. A market cap built from today's share count and an old price is a number nobody could have seen on that day, and a backtest that selects on it has quietly used information from the future.

A size filter is only honest if it is dated.

Any rule with a size threshold in it is a point-in-time question. A cut like 'only trade above two billion dollars' selects one universe when the cap is measured on the trade date and a very different one when it is measured today, because the winners grew into the threshold and the failures shrank or delisted out of the data. Applying current caps backward is one of the quietest ways to make a backtest look better than the strategy was.

Shares outstanding carries its own signal. A rising count is dilution in progress; a falling one is a buyback actually executing rather than being announced. Because the count is dated here, both become a series you can difference, align to filings or offerings, and study around.

• Apply size cuts using market_cap on each rebalance date instead of one screen run today.
• Difference shares_outstanding across dates to date issuance and buyback activity.
• Divide event-driven amounts, such as a filing's dollar size, by the market cap current on the event date.
• Build size deciles per date, so a size factor is ranked against the market as it stood.
• Check where a ticker's history begins before backfilling a long study.

Recorded per trading day, then left alone.

Each observation is written for its own trading date and carries the shares outstanding known at that time. That is what makes the series point-in-time: rows are dated on the day they describe rather than restated later, so a query for a past window returns the same values on every rerun.

Results come back ordered by date descending, then ticker descending, and large pulls page with a cursor. A full history for one ticker is a loop over next_cursor, sending its two parts back as cursor_date and cursor_ticker on each call; sending one without the other returns a 400. A companion lookup, GET /v1/historical-market-cap/tickers, lists every ticker the dataset covers along with the earliest date available for each, which is the cheapest way to check coverage before starting that loop.

Two values, and the cursor that pages them.

The headline fields below are a subset. Every field, with exact types and semantics, is documented in the API reference.

One call for a ticker, or one call for a date.

One key works across the REST API, the hosted MCP server, and this dashboard. Every response is JSON with a { count, data } envelope.

import requests

url = "https://api.alphanume.com/v1/historical-market-cap"
params = {
    "ticker": "AAPL",
    "date": "2026-02-06",
    "api_key": "alp_abc123"
}

r = requests.get(url, params=params)
print(r.json())

curl "https://api.alphanume.com/v1/historical-market-cap?ticker=AAPL&date=2026-02-06&api_key=alp_abc123"

{
  "count": 1,
  "has_more": false,
  "next_cursor": null,
  "data": [
    {
      "date": "2026-02-06",
      "ticker": "AAPL",
      "market_cap": 4109599296360,
      "shares_outstanding": 14776353000
    }
  ]
}

What this data does not claim.

• Rows are stored observations, not restated figures. If a source later revised a share count, the row still reflects what was known on its own date.
• shares_outstanding is a company-level count, not free float. Insider and restricted holdings sit inside the number, so it is not a measure of tradeable supply.
• There is no price column. Cap divided by share count implies a price, but the dataset is not meant to be used as a price source.
• Coverage begins at a different date for each ticker, so a long backtest will find some names simply absent at the front of the window rather than present at size zero.
• Size-factor studies require sufficient historical coverage. The limited public dashboard preview is for evaluating the data, not running a full size-factor study.