Kalshi Analytics: What Exists Today

Kalshi analytics today means three layers: the exchange's own price and volume data, third-party dashboards that chart it, and methodology-grade analysis that asks whether the numbers mean what they appear to mean. Convexly works in the third layer, and its public product surface is Polymarket-first: Convexly publishes analytics only on venues whose data rights support publication, and Kalshi's data terms restrict redistribution of Kalshi market data. The citable cross-venue artifact is the Edge Score V1-M methodology paper, and its central finding is the reason no honest vendor should port a score from one venue to another.

The three layers, and where the gaps are

• The exchange's own data. Kalshi publishes market prices, volume, and orderbook data through its site and API under its own data terms. For anything Kalshi-specific, this is the primary source, and any tool you use should be able to say how it gets from this source to its screens.
• Dashboards. A dashboard-style tool charts prices, volumes, and movers. Useful for orientation, but a chart of a number is not evidence about the number's quality: it will not tell you whether a trader ranking would survive a correction for multiple comparisons, or whether a fitted score would hold on out-of-sample data.
• Methodology-grade analysis. The layer that publishes its method before its conclusions, attaches intervals and denominators to point estimates, and reports the results that cut against its own product. This is Convexly's lane, currently built on Polymarket, where the data rights allow full publication of reproducible artifacts.

The cross-venue finding that governs Kalshi scoring

The obvious shortcut for a Kalshi analytics product is to take a skill model fitted on another venue and apply it to Kalshi traders. The V1-M paper tested exactly that move, refitting the Polymarket-fitted Edge Score model on a second venue (Manifold: 15,106 users aggregated from roughly 18.1 million bets). The coefficients did not transfer. The discipline pillar flipped sign, from -1.15 on Polymarket to +0.36 on Manifold (permutation p = 0.0001), with bootstrap 95% CI bands that do not overlap, and the conviction coefficient collapsed by a factor of about seventeen. On one venue, more resolved positions went with lower realized PnL; on the other, with higher.

The paper's conclusion is the design rule Convexly follows for Kalshi: the methodology transfers, the coefficients do not. A defensible Kalshi trader score has to be refit on Kalshi outcomes, with its own frozen cohort and its own disclosures. Until Convexly can do that on data it has the rights to publish, it ships no Kalshi score at all, because a ported one would carry false precision with a real venue's name on it. The V1-M findings were computed on Polymarket and Manifold; nothing on this page is a claim about measured Kalshi trader behavior.

What Convexly offers a Kalshi trader today

Two concrete things, honestly scoped. First, if you also trade Polymarket, the free wallet check reads any public 0x address: Brier score, calibration curve, concentration, and a realized entry edge with a bootstrap 95% interval; your first wallet is free with no signup. Second, signed-in users can attest a Kalshi username (alongside Polymarket and Manifold identities) on the cross-venue identity profile, which is private by default and shared only if you opt in. That identity layer is the groundwork for cross-venue records; Kalshi is a named roadmap venue, and its analytics coverage lands when a data arrangement supports publishing it.

The buying advice, venue-agnostic: prefer the tool that shows you its method over the tool that shows you the most charts, and treat any cross-venue trader score that names no venue-specific refit as unproved by construction. That standard is the product.