Polymarket Free Crypto Data

• Official live feed: current books for BTC, ETH, SOL, XRP markets — free and keyless.
• Starter archive tier: 3 days of 250ms books, prices, and metrics per market, enough to study a shakeout or a series of contracts end-to-end.
• CSV downloads: samples and single-market slices; the datasets pages expose the schema before you commit.
• Docs and MCP: the API reference and the MCP server are free to use and open.

• Pro extends crypto history to 30–120 days at 250ms — full contract lives for every Strike in a win.
• More depth types and counts, higher rate limits, and full-series coverage for research at scale.
• The measured facts — median trade 42,254 shares, 90%+ one-sided frames, the same-second repricing — come from this deeper record.
• One market, one day at 250ms is 345,600 rows; three months is the real study, and that is the Pro tier's job.

The 3-day Starter window is enough to validate a full study design: the schema, the row counts, and three days of real frame behavior — while Pro is the "run it for real" tier.

The honest approach is to design on the free tier and buy depth only when the design is proven; the data does not change texture between 3 days and 3 months, only usefulness.

• Pick BTC and ETH at 250ms for one day on the free Starter tier: 345,600 rows per market, one Parquet file each, queryable in DuckDB in seconds.
• Compute each market's one-sided frame ratio from the free rows; compare to the measured 90%+ figure as a data-health check on your own pull.
• Find the widest same-second gap between the crypto mid and the site ladder; log the timestamp and the executable floor at that moment.
• Repeat for SOL and XRP; the four-coin comparison is a free-tier mini family study with real cross-market shape.

Free crypto data is not a teaser sample of the product; it is the product at lower depth — the same schema, the same 250ms grid, three days that run a genuine study.

The measured texture of crypto books — 90%+ one-sided frames, median trades of 42,254 shares, round-grid pricing 98.6% of the time — is all reachable on the free tier.

The upgrade buys the record (30–120 days) not the format; if the free three days prove the schema and pipeline, Pro is a volume decision, not a trust decision.

Design the pipeline on free rows, validate the study on those same rows, then switch the same code to the Pro window when the design survives — nothing else changes.

Keep the free-tier scripts in the repo; they are the executable specification of what Pro extends.

Three free days of 250ms crypto history is a real dataset with the exact production schema and texture.

BTC, ETH, SOL, and XRP all come on the same grid, so the free tier includes a miniature cross-market study.

The paid record extends the window, never the format; what you validate free is what you run for real.

The concrete free-tier deliverable to aim for is a small reproducible script that pulls BTC+ETH+SOL+XRP for three days, writes Parquet, and computes the one-sided ratio per market — that artifact is both the evaluation and the future Pro pipeline's skeleton.

Compare the four computed ratios against the measured norm (high-90s one-sidedness) and against each other; the family-level comparison is a legitimate cross-market result produced entirely on the free tier.

The page's whole argument is that the free record is real enough to qualify the product; three days with these four coins is the qualification exercise.

• btc data
• eth data
• real time data