Tennis is one of the most-traded sports on prediction markets — a match is a long sequence of near-binary moments, which makes the win-probability line jump around in ways that reward good microstructure work. So it's frustrating that historical tennis order-book data is almost impossible to buy.
Here's the honest landscape, and what the data looks like when you can get it.
Why tennis order books are so hard to source
There are two problems stacked on top of each other:
- The venues keep no history. Kalshi — the main US venue with deep tennis coverage — exposes a live order book through its API but retains no historical snapshots. Once a tick passes, the official record is gone. Same story as every other sport on the venue.
- The third-party vendors skip tennis. The capture shops that sell raw Polymarket L2 depth (Telonex, PMData, PolymarketData) are built around Polymarket's crypto-political markets. Tennis lives mostly on Kalshi, and the Kalshi-capture vendors focus on the big US ball sports. Tennis falls through the gap.
The net effect: match-results-plus-odds datasets (the Sackmann/Tennis-Data style) are everywhere and mostly free, but tick-level book depth — bid, ask, and size at each level as the match plays — is a genuine hole. And that's exactly the data you need for execution and liquidity research, because a price move only means something if you know how much size was defending it.
What ATP/WTA/ITF depth actually looks like
If you've read our guide on reading a prediction-market order book, this is that structure, recorded historically for tennis. One capture we've been running stores the ladder to 20 levels per side at a ~40-second cadence across ATP, WTA and ITF singles match markets, plus the ATP game-spread markets. A single match generates thousands of book snapshots; a busy tournament day is millions of rows.
The questions this data can answer, that a results-plus-odds file never could:
- How far does the book thin out on a break point versus a hold? Depth around a known information shock is the tennis version of an earnings-reaction study.
- Does displayed size predict the next move, or is it decoration that pulls the moment real money arrives?
- What would it actually have cost to trade size at the displayed price — was the top of book three lots deep or three hundred?
The honest caveats to demand from any tennis-data seller
We wrote a whole piece on vetting sports-data vendors, and it applies double here because the data is scarce enough that sellers get away with vague claims. For tennis order books specifically, insist on:
- Depth cap disclosed. Kalshi exposes more than 20 levels; most captures (including ours) store the top 20. "Full book" is almost always an overstatement — ask for the number.
- Cadence and gaps disclosed. A polled capture has a cadence (ours is ~40s) and inevitable outage windows. A seller who won't hand you a gap map is hiding something.
- Depth-only vs. trade tape. Kalshi's tennis markets don't produce the kind of public trade tape the ball sports do, so tennis order-book data is depth-only. Anyone claiming a tennis "trade tape" should explain exactly what they mean.
Where to actually get it: the capture described here is sold as a standalone Kalshi Tennis Order-Book slice ($39) — 46 million rows of ATP/WTA/ITF depth with the coverage map and gap disclosure baked in. There's a free sample and the full schema on the samples page if you want to inspect the columns before paying anything.
One honest word on edge
Order-book depth is for understanding execution and liquidity — not for finding a tradeable signal. These markets are efficient; the seller above publishes its own negative results saying so. If someone pitches tennis order-book data as a way to beat the line, reread the vetting guide. The real value is in the microstructure questions, which nobody has published good answers to yet because nobody had the data.
Want to see the columns first?
Free tennis order-book sample, full schema, no signup.