Kalshi Upgrades Institutional Infrastructure with Direct Level 2 Data Feeds

The prediction market venue now ships a direct Level 2 multicast feed for systematic firms, covering sports contracts and crypto perpetual futures — together roughly 60% of its weekly notional volume. That is not a side experiment. That is where the size lives.

The Latency Tax Is Gone

Quants stitching Kalshi data into their stacks had two options until now: poll REST endpoints and rebuild the book themselves, or ignore the venue entirely. Every poll added milliseconds. Every reconstruction introduced the probability that your snapshot was stale by the time the model priced against it. That tax is gone.

The new feed, built with low-latency provider DoubleZero Edge, pushes full depth directly into firm infrastructure via Kalshi Research. Level 1 top-of-book and Level 2 depth arrive at machine speed — multicast, no polling, no manual rebuild. Andy Roth, Kalshi's head of institutional business, framed it as "better connectivity makes better markets." That is the correct framing if your definition of "better" is thinner spreads and tighter book integrity. Firms currently active on Kalshi were demanding institutional-grade plumbing. They just got it.

Data Is Not Edge. Execution Is.

I want to be blunt here. A clean Level 2 feed does not hand you an alpha source. It removes one inefficiency from your stack. That is necessary, not sufficient. If your pricing model is wrong, faster data just lets you lose capital at institutional speed. Sample size still matters. Confluence still matters. Signal-to-noise ratio still matters.

The real edge for systematic traders sits downstream of the wire. Slippage modeling gets sharper. Cross-contract pricing on correlated Kalshi markets becomes realistic instead of aspirational. Order routing logic can finally size fills against visible liquidity rather than guessing at depth. For firms that already had a thesis on prediction markets, this is where the compounding happens. The startup scaling real trading infrastructure here is genuine, but the plumbing is not the strategy. Confusing the two is how retail traders blow up.

Kalshi is waiving its share of data revenue for the first year. Read that as a seat-filling campaign. They want firms consuming the feed so the book thickens enough to attract more flow. Whether the resulting liquidity premium accrues to early adopters or gets competed away depends entirely on how fast the next ten firms plug in.

What I Am Watching

Three things over the next two quarters. Whether order book depth on Kalshi's sports and crypto perps actually compresses — I want bid-ask spread tightening visible on the screen, not buried in a press release. Whether market makers routed through Talos start posting meaningfully larger size during thin sessions. And whether third-party data vendors pick up the feed and redistribute it downstream.

If spreads widen, the feed is decoration. If they compress, the venue finally trades like the others.