Lychee

What it is

A browser workspace for prediction-market data with no code in it. You pick a source, pull rows, and turn them into a chart, a data sheet, a backtest, a saved dashboard or a file. The vendor's own framing: "Every Kalshi + Polymarket trade, market, and order book since launch … Turn any question into a query, chart, backtest, export, or live dashboard in seconds."

Kalshi and Polymarket are the only event venues. Around them sit about fifteen other connectors — Chainlink, Binance, CoinGecko, GeckoTerminal, Twitter, Wall Street Bets, SEC EDGAR, US Treasuries, Census.gov, Crunchbase, Hacker News, Product Hunt — plus CSV and XLSX uploads, which is what makes it a workspace rather than a viewer: the interesting query is usually a venue's prices against something that is not the venue.

It is run by one named person, who signs the site's data-use policy as its creator and says he answers every email himself. The support address in the terms and privacy pages belongs to a different domain from the product, which is the usual sign of a solo operator with an older project behind them — and the oldest dated artefact on the site, a data-use addendum from 8 May 2024, says the same thing about the charting tool this was built on.

Availability

Nothing to deposit and no venue account to connect, so the jurisdiction questions that decide a venue card do not apply. Worth knowing that the terms are unusually thin for a paid product: roughly 350 words over eight clauses, with no governing-law clause, no arbitration clause, no age gate and no geographic restriction stated at all. The privacy policy and the data-use page say only an email address is collected at registration.

One caveat the vendor does state, and it is a good one: anything you run through an AI feature is shared with the AI provider, and the data-use page advises against putting sensitive data through it.

Checked against the terms of service and privacy policy on 19 September 2026; both carry a last-updated date of 16 September 2026.

Pricing

Three self-serve tiers, billed weekly, monthly or annually — the page opens on monthly, which already carries a 10% badge, with annual advertised at 25% off. The month-to-month rates are 19.99, 39.99 and 79.99 US dollars for Basic, Pro and Elite.

What the money actually buys is rows and concurrency. Basic caps a historical pull at 12,500 rows, allows one dashboard and one alert; Pro raises the pull to 500,000 rows and adds backtesting workflows, live dashboards, ten live streams, ten alerts and 4 GB of saved workspace; Elite is the same shape with "highest" unspecified pull limits, 100 live streams, 500 alerts and 10 GB. Every paid tier includes both venues, live and historical, and CSV, XLSX and JSON export.

There is no free plan. A free demo on the home page runs sample pulls without a card, and an Enterprise arrangement is mentioned on the data-use page with no price attached. An affiliate programme pays 20% on referrals.

Prices read from the vendor's plan section on 19 September 2026.

Markets & resolution

Nothing here resolves anything; the archive records how Kalshi and Polymarket resolved. The Kalshi side advertises markets, trades, prices, outcomes, categories and order-book history since July 2021; the Polymarket side is documented in the site's guides as ingestion from the Gamma API for metadata and events and the CLOB WebSocket for live market data. A cross-venue view compares the two venues' odds, charts and activity for the same question, which is the thing this category most often cannot do.

Two figures on the Kalshi page deserve a note, because a reader may meet them twice. Lychee advertises 7.68 million-plus unique markets, 72.1 million-plus historical trades since July 2021 and a 36 GB-plus compressed archive. Those three numbers are also, to the digit, the size of the MIT-licensed open Kalshi dataset published with a January 2026 research paper and mirrored in Prediction Market Analysis — 72.1 million trades across 7.68 million markets from June 2021, distributed as a roughly 36 GiB compressed archive. Lychee names no source for its own archive, and its order-book history has no counterpart in that dataset, so the overlap is not the whole story. If provenance matters for your use — attribution, redistribution, or citing it in research — ask before you subscribe.

Integrations

There is no Lychee API, no WebSocket, no Python client and no MCP server; "request new integrations, features, and APIs" on the Pro tier means asking the vendor to add a data source, not a key of your own. What leaves the product is files — CSV, XLSX and JSON — plus embeddable interactive charts and published dashboards. Removing Lychee branding from a public chart is a Pro feature.

Limitations

No published refresh interval anywhere. The closest the site comes is that Kalshi's live endpoints "return current exchange data when queried". Plans are metered in "live data streams" with no stated cadence per stream, so how fresh a live dashboard is is not a documented number.

No methodology page. How volume is computed, how a window is defined, what is done about gaps — none of it is written down, on a product whose entire output is derived figures. Order-book history is hedged on its own page as available "where available".

The social-proof row is not attributed. The home page carries the line "Trusted and actively beta tested by friends at" above logos of JPMorgan, Goldman Sachs, Meta, Google, Apple, MIT and Jump Trading, with no named person, no team and no permission statement. Read it as a claim the vendor has not backed up.

Nothing on the site addresses affiliation with Kalshi or Polymarket. Not a denial — an absence. The venues' names and marks are used throughout and no independence statement exists to quote, which is unusual in this category and worth knowing if you plan to cite its output.

The product surface has stubs. The CSV-exports and quant-analysis pages both say detailed content is coming soon, and the changelog has two entries, dated 15 April and 4 September 2026. Nothing on the site is dated before March 2026.

Two venues only. No Limitless, no ForecastEx, no Myriad, no CME.

Alternatives

Prediction Market Analysis is the free, MIT-licensed version of the same job if you write Python and have the disk. Adjacent and Predexon sell cross-venue data as an API rather than a workspace, which is the shape to choose if something has to run on a schedule. Polymarket dashboards on Dune is the SQL answer for the Polymarket half and costs nothing to read.