Artemis prediction-market metrics

What it is

The venue-level view of the sector, rather than the wallet-level one that dominates this category. Artemis runs event-contract venues through the same daily framework it uses for exchanges and protocols — volume, open interest, transactions, active users, fees, and total value locked where the venue is on-chain — and covers thirteen of them, from Polymarket and Kalshi down to ForecastEx, Myriad, Gemini Predictions and Hyperliquid's HIP-4 markets.

The reason to know about it is the methodology page, which is the most complete public account of where prediction-market aggregate numbers come from. It names the ingestion path per venue: on-chain venues indexed from archive nodes and raw block feeds rather than from a venue-reported dashboard figure; Kalshi from its official API market snapshots and trade feed; CME from the public daily volume and open-interest file; Nadex from the Daily Bulletin PDF that is its CFTC Part 16 disclosure; Gemini from per-contract market data published under CFTC Rule 16.01; ForecastEx from an end-of-day prices file.

It also names its own gaps, which is rarer still. CME's sports and culture contracts — the bulk of that venue's volume since the December 2025 FanDuel launch — are published only inside a Daily Bulletin PDF section and are flagged as a known coverage gap. Gemini reports no open interest, so the field is carried as an explicit null rather than a zero. Nadex and CME contracts are not dollar-face, so their contract counts are not dollars and must not be compared one-for-one with venues whose contracts settle at a dollar.

The wider product is an AI research platform for equities and crypto; prediction markets are one sector page among roughly sixty. That is the right way to read this card: a general research terminal that happens to carry the best-documented cross-venue aggregates.

Availability

Nothing to deposit and no venue account required. The terms set an age floor of 18, are governed by New York law with AAA arbitration in New York and a class-action waiver, and state that the services are hosted in the United States with data processed there. There is no geographic restriction, no KYC step and no sanctions clause in the terms.

One connection is worth flagging separately. Artemis takes brokerage credentials through Plaid to read your own holdings, and the terms say that by default portfolio information, including holdings and performance, may be displayed on the platform's social features for other users to see, with an opt-out. That is a read-only integration and nothing on the platform places an order, but the default is unusual.

Checked against the terms of service and privacy policy on 19 September 2026; both carry a last-updated date of 7 September 2026.

Pricing

Free is a real tier: the Intern plan is zero dollars, includes full historical data, exports and mobile access, and is capped at 20 analyst queries a month with daily rate limits, one dashboard and three charts. Note the vendor's own definition — a follow-up question in the same thread counts as another query.

Investor at 100 US dollars a month lifts that to about 100 queries, removes the daily limits, and adds the Excel and Google Sheets add-in at 50,000 calls a month. The Plugin is the add-in sold on its own at 500 US dollars a month per user with a million sheet calls and API access. Managing Director is quoted per engagement and the docs put a floor of 50,000 dollars a year on it; it is the tier that carries the REST API, the MCP server and the Snowflake data share.

Billing is in US dollars through Stripe, renewals are automatic, and the terms say no refunds or credits for partial billing periods. The pricing page has no monthly-versus-annual toggle at all.

Prices read from the vendor's pricing page and docs on 19 September 2026.

Markets & resolution

Nothing here resolves anything, and nothing here is a contract-level probability: what Artemis publishes are venue aggregates on a daily grain. The prediction-market metric set is volume, open interest, contracts traded, transactions, daily and monthly active users, fees, total value locked and active markets, with category splits for sports, politics and crypto and a volume figure with sports removed — which is the split most often wanted and most often unavailable.

Two definitions on that page are the ones to internalise before quoting a number anywhere. spot_volume is notional, not premium: the cash actually traded is roughly half of it at typical prices and lives in a separate field. And perpetual futures on Kalshi and Polymarket are dollar-native rather than dollar-face, sit in separate perp_ fields, and must never be summed with event-contract volume. Most published comparisons of "prediction market volume" get one of those two wrong.

Integrations

A deliberately small REST API — three endpoints, one of which fetches named metrics — plus a Python SDK, an MCP server, a Snowflake data share where daily models sit under named tables, and an Excel and Google Sheets add-in whose formulas refresh on pull. There is no WebSocket and no streaming product; the grain is daily. The API's own documentation warns that the current period is normally present with a null value because it has not closed, and that you should filter before taking a latest value.

One metric, contracts traded, is documented as not currently exposed through the REST API.

Limitations

The commercial-use restriction is the sharpest term on the card. The terms limit the service to personal, non-commercial use by individual retail investors, and name funds, banks, advisers and data vendors as excluded, along with redistribution, integration into a commercial product or workflow, and any use that indirectly generates revenue for a business. Systematic retrieval to build a database is separately prohibited, and third-party data from S&P Global Market Intelligence and Quartr carries its own internal-use-only terms. Anyone planning to cite these figures at work needs the written licence rather than the 100-dollar plan.

Programmatic access is priced out of reach of an individual. Reading the charts is free; getting the series into your own code starts at a five-figure annual engagement.

Daily grain, and no published lag. Nothing intraday, no streaming, and no stated time by which a day's figure lands.

Three named coverage gaps, and they are load-bearing. CME's sports and culture volume, Gemini's open interest, and the contract-count-versus-dollars problem on the two non-dollar-face venues are all documented — which is better than a competitor hiding them, but they are still holes in exactly the comparison most readers want.

The platform deprecates coverage. The data changelog records 122 chains and protocols turned off in mid-2026 and application-level activity metrics deprecated the same month. A series you build a workflow on can be withdrawn, with notice, on a page you have to go and read.

No affiliation statement exists. Neither the terms nor the privacy policy addresses the venues it tracks; the only third-party data sources named are S&P Global and Quartr.

Alternatives

Lychee is the opposite shape — per-trade rows for two venues, queryable from a browser, at a tenth of the price and without the documentation. Adjacent and Predexon sell cross-venue market data as an API for people who need contract-level rather than venue-level figures. Polymarket dashboards on Dune answers the on-chain half free, in SQL you can read and fork.