Kalshi 15-Minute Markets: How They Work And How To Read The Tape
Kalshi has taken its fastest product and made it faster. The exchange that started with elections and now runs the biggest regulated event-contract book in the country has rolled out 15-minute markets: one question per window, will this price finish the next quarter hour at or above a stamped target, settled and reset around the clock. They cover Bitcoin, Ethereum, XRP and a shelf of other coins, and the rollout now reaches gold, silver, WTI crude, the S&P 500, and the Nasdaq 100.
The Quick Answer
A 15-minute market is a yes/no contract on a single window. At the open, the exchange stamps a target from a reference price feed. For the next 15 minutes, traders buy Yes or No against an order book. At the close, the market compares averages, not ticks, and pays $1 to the right side. The full mechanics, a live board of every series we track, and the honest answer on whether these can be beaten are all below.
The Live Board
Live board — fetched August 4, 2026 at 11:30 AM ET. Prices are the exchange's own yes bid/ask in cents; settled columns cover each series' last 20 windows.
| Market | Current Window (target · close · yes bid/ask) | Last 20 Windows |
|---|---|---|
| Bitcoin | $63,892.42 by 11:45 AM ET · 51c / 52c | 13 above / 7 below · streak: 2 below |
| Ethereum | $1,864.46 by 11:45 AM ET · 62c / 64c | 14 above / 6 below · streak: 2 below |
| XRP | $1.0727 by 11:45 AM ET · 66c / 67c | 13 above / 7 below · streak: 2 below |
| Gold | $4,078.43 by 11:45 AM ET · 53c / 55c | 14 above / 6 below · streak: 1 below |
| Silver | $59.396 by 11:45 AM ET · 56c / 58c | 13 above / 7 below · streak: 2 below |
| WTI Crude | $76.34 by 11:45 AM ET · 36c / 38c | 7 above / 13 below · streak: 2 below |
| S&P 500 | — | — |
| Nasdaq 100 | — | — |
The board refreshes about every 15 minutes around the clock, one row per series. "Last 20 windows" counts how many of each series' recent windows settled at or above their target. When a book shows no quotes, that window's market has not opened yet; each window gets its own order book minutes before it starts.
One Morning On The Board, Read Out Loud
Here is a single snapshot, fetched from the exchange API on August 4 at 9:20 AM ET, and what each price actually says. The live board above supersedes these numbers the moment it refreshes; the point of the snapshot is the reading, which does not change.
| Market | Window Target | Yes Bid / Ask | What The Price Says |
|---|---|---|---|
| XRP | $1.0747 | 53c / 54c | Near the money. A one-cent spread on a coin flip; the book is tight and fair. |
| Bitcoin | $64,016.45 | 9c / 9c | Target well above the tape. Priced like the long shot it is. |
| Gold | $4,084.09 | 1c / 3c | Far target on a quiet metal. Cheap is not mispriced; cheap is unlikely. |
Exchange prices translate straight to probabilities, which is the fastest way to read any row. XRP's 53c bid is the market pricing "finishes above target" at 53%, with the 54c ask putting the other side of the coin at 46%. Bitcoin's 9c yes reads as a 9% chance of reaching a target stamped well above the tape. Gold's 1c bid and 3c ask bracket the market's read between 1% and 3%. None of those translations require judgment; a cent is a percentage point, which is most of what makes these books legible at a glance.
The XRP row is the one to internalize. Fifty-three cents bid against fifty-four asked is a market saying "we do not know" with one cent of disagreement. Crossing that spread costs you about 2% of your stake for the privilege of guessing a quarter hour of price drift. That arithmetic, repeated every window, is the house edge these markets do have, and it belongs to the spread, not to a side. In the same snapshot, XRP's previous 20 windows had settled 13 above and 7 below, a 65% run of above-target finishes that reflects five hours of drift, not a signal about the next window, which starts back at the 50% coin flip every time.
Get the fast lane graded before you trade it. Our AI model panel puts its calls on the record for slower markets and grades them in public. The Weekly PM Market Brief sends the panel's graded record, the week's biggest market-versus-model gaps, and what is spiking next. Free, one email, Sundays, via the signup at the end of this page.
How A Window Actually Settles
The settlement rule is the most misunderstood part of these markets, and it is worth reading once in full. Bitcoin's version: the market resolves Yes if the simple average of the sixty seconds of CF Benchmarks' Bitcoin Real Time Index before the window's close is at least the same sixty-second average taken before the window opened. Two averages, one comparison. The other series work the same way against their own reference feeds.
That design has three consequences worth knowing before your first window.
First, the target is not a guess, it is the stamped open-average, so every window starts as a fresh coin flip on whether the next 15 minutes drift up or down from it. Second, no single print settles anything. A one-second wick through the target means nothing unless it moves the sixty-second average. Third, the averaging window is why these markets sometimes flip in the final seconds: the settlement average can catch a move the headline price has not fully reflected. That behavior looks alarming the first time you watch it, and it is the subject of its own page, are Kalshi's 15-minute Bitcoin markets rigged, where the short answer is no, it is arithmetic.
What Is Listed On The 15-Minute Shelf
The crypto side runs deepest: Bitcoin, Ethereum, XRP, BNB, Cardano, Solana-ecosystem names like HYPE, plus NEAR, TON, Zcash, Bitcoin Cash, and a Coin Race market on which major coin leads the window. The August expansion added the traditional-asset shelf: gold, silver, and WTI crude from the commodities floor, and the S&P 500 and Nasdaq 100 from equities. Most of these also run hourly versions, and the slower books, daily and beyond, are covered across our prediction-markets library.
Two practical notes on the shelf. Index and metals windows track their underlying market sessions, so expect quiet books on weekends there while crypto runs continuously. And every series lists exactly one open window at a time, with the next one queued behind it; if you want the close schedule in detail, our page on Kalshi market close times covers it.
Reading The Board Honestly
A few things the board will show you within an hour of watching it.
Near-the-money windows price like coin flips. When a target sits close to the live price, Yes trades in the 40s to 50s in cents. That is the market telling you the truth: nobody knows which way the next 15 minutes break.
Far targets price like lottery tickets. When the stamped target sits well away from where the asset is trading, Yes goes to single digits fast. A cheap contract is not a mispriced contract; it is a long shot priced like one.
Streaks are noise. A series showing 15 of its last 20 windows settling above target is not hot. It reflects a drifting tape during those five hours, and the drift owes you nothing for the next window. If a streak tempted you to bet continuation, you are paying the spread for a coin flip.
Why You Cannot Snipe These
We did the work so you can skip the tuition. Our team spent two weeks probing the hourly and 15-minute books for durable value, crypto first, with real orders and live feeds. The finding: the market makers in these books reprice off the same real-time indexes the contracts settle on, within seconds, around the clock. By the time a human sees a gap between the book and the spot feed, the book has moved. The hourly books offered no repeatable entry, and the 15-minute books are tighter still, because a shorter window leaves even less time for a stale quote to survive.
That is not a knock on the product. It means the prices are close to fair, which is exactly what you want from a venue if you are trading it for entertainment. But go in with the right frame: you are paying a small spread for a fair-odds game on a very short clock.
The takeaway in one line: the edge in a 15-minute market belongs to the spread, and the spread belongs to the market maker. Anyone selling a system for beating these windows is selling the thing our testing could not find.
Where a view can be worth something is on longer clocks, where information digests slowly and books are thin enough to lag the news. That is the lane our AI model panel works: eight models pricing slow markets blind, revising once in a Delphi round, with every call graded on our public scoreboard when the market settles. Fast tape for watching, slow tape for thinking.
Trading Mechanics, If You Do Step In
Everything that applies to Kalshi generally applies here, compressed. You can exit before settlement by selling your position, though in a 15-minute window there is rarely a reason. Limit orders matter more than usual because spreads on a short clock are the whole game; our Kalshi order types guide covers the difference in two minutes. And these are exchange contracts with a counterparty on the other side of every trade, not house lines, the structural difference explained in Kalshi versus a sportsbook.
FAQ
What are Kalshi's 15-minute markets? Yes/no contracts on whether an asset's price finishes a 15-minute window at or above a stamped target, one window at a time, per series, on crypto, metals, oil, and stock indexes.
How does settlement work? Two sixty-second averages of the reference index, one before the open and one before the close, compared once. At or above settles Yes.
Can they be beaten? Our testing says treat them as fairly priced entertainment. The books reprice off live feeds faster than a human can exploit, and the shorter the window, the truer that gets.
Why did a window flip at the last second? The settlement average caught a move the headline price had not reflected. Mechanics, not manipulation.
Nothing on this page is a pick, a play, or a recommendation. Kalshi event contracts are CFTC-regulated event derivatives traded on a designated contract market; a position can lose its full value. 18+; availability is state-specific. Confirm eligibility with the venue before funding an account.
