Working name · KPI Edge
Prediction markets now price the operating metrics that move individual stocks — deliveries, subscribers, funded accounts, customer counts — as live, tradable probability ladders that resolve at the earnings print. We turn that stream into a fundamental-surprise signal on the underlying equities.
Kalshi has quietly built a fundamentals exchange. The single-name layer is bigger than it looks.
Each KPI market is not a single guess — it's a ladder of "above X" contracts that together encode a full implied probability distribution over the number a company will report. The market re-prices daily, ahead of the print, as money moves in.
| Company | Series | Metric the market prices | Equity |
|---|---|---|---|
| Robinhood | KXHOOD | Funded customers · Gold subscribers | HOOD |
| Palantir | KXPLTR | Total customers | PLTR |
| DoorDash | KXDASH | Total orders | DASH |
KXRDDT | Daily active uniques | RDDT | |
| Snap | KXSNAP | Daily active users | SNAP |
| Shopify | KXSHOP | GMV | SHOP |
| Tesla | KXTSLA | Deliveries / production | TSLA |
| Chipotle | KXCMG | Restaurant count · comps | CMG |
| Uber · Airbnb · Spotify · Coinbase · Roblox · … | +60 more | trips · nights · MAU · volume · hours | UBER · ABNB · SPOT · COIN · RBLX |
Acquisition odds (KXTAKEOVERUNP, KXTAKEOVERNEE) and S&P 500 add/remove (KXNEXTCOMPANYSP500) — clean jumps and mechanical index flows.
Robotaxi, new models, Vision Pro, Apple announcements — the market's live read on whether/when a catalyst lands.
CEO-change odds (TESLACEOCHANGE) and litigation outcomes (DOJ, antitrust) as a valuation-overhang gauge.
Universe counts from a live scan of the Kalshi public API on 2026-06-16. Ticker/metric mappings maintained in finmap/bloomberg_bridge.yaml.
A single stock gaps on the number it reports. That number now trades, in public, weeks before the print.
The dominant driver of a single-stock move into earnings is the gap between the reported KPI and what was expected. Today "expected" means a stale, sparse sell-side consensus — a handful of analyst point estimates, revised on their own schedule.
Kalshi replaces that with a continuously-updated, crowd-priced distribution over the exact metric. Where that implied distribution disagrees with the Street — or drifts before the print — is a directional read on the equity that almost no one is systematically harvesting yet.
One causal chain, fully observable end to end.
Every link in that chain is measurable from public data: the ladder prices come from Kalshi, the print is the reported fundamental, and the reaction is the equity return. That makes the thesis falsifiable — we can build the historical panel and test whether implied-vs-realized surprise actually predicts the move.
See the live, worked example — Tesla Q2 production — on the How it works page, including the real implied distribution and the first drift-vs-price test on actual data.
Four distinct, stackable signals — and a moat that compounds.
Implied consensus from the ladder minus the sell-side number. A live "whisper" gauge; the gap is the directional bet.
The full implied PDF flags asymmetric setups the point estimate hides — fat upside vs. capped downside into the print.
Daily Δ in the implied metric. Does the crowd revising up lead the stock before earnings? A tradable nowcast.
Post-print, a clean surprise variable per name, regressed on the event-window return — the systematic backbone.
Same pipeline, three go-to-market shapes — in increasing capital intensity.
A daily feed: per-name implied KPI, implied-vs-Street gap, drift, and post-print surprise scores. Buyers: fundamental L/S desks, quant PMs, IR/strategy teams. Lowest capital, fastest to revenue.
Event-driven book: position the underlying (or options into earnings) on surprise + drift signals, sized by liquidity and confidence. The signal feed is the alpha source. Real capacity lives here.
The ladders are thin today. A disciplined model of fair value across brackets can quote the KPI markets themselves — earning spread while the equity signal is the hedge/edge. Highest skill, thinnest current depth.
volume = 0 / open_interest = 0 this far from the print, and bracket prices go
non-monotone in the tails (a tell-tale illiquidity artifact). The entire thesis depends on depth building into the
report date. Step zero is a liquidity screen to find which names actually trade.
KPI markets settle on a third-party reported figure (e.g. fiscal.ai). Definition drift or restatements need monitoring per series.
Quarterly events → few resolved observations per name. Edge must be pooled cross-sectionally and validated out-of-sample.
Capacity lives in the underlying equity/options, which is deep — but the signal's uniqueness erodes if it gets crowded.
The Kalshi price may just track the stock. The drift signal is only alpha if it leads — that's the first thing to test.