Kalshi 15-minute crypto markets: rules and data

Kalshi opens a new market every 15 minutes, around the clock, on whether bitcoin, ether and seven other coins end the window higher than they started. The bitcoin series, KXBTC15M, alone has 1.4 billion rows of order book changes and trades on file.

Checked against Kalshi's rules and the files on November 2, 2026.

How one settles

• The question is whether the coin is up over the next 15 minutes: “BTC price up in next 15 mins?”
• YES wins when the average of CF Benchmarks' Real Time Index (BRTI) over the 60 seconds before the close is at least its average over the 60 seconds before the open. Each coin has its own CF Benchmarks index (ETHUSDRTI for ether, SOLUSDRTI for Solana), not any one exchange's price.
• The opening average is the market's target price, shown on Kalshi as “Target Price” and in the files as floor_strike. Since one window opens as the last one closes, each market's target is the previous market's closing average.
• Settlement is expected about five minutes after the close.

Reading the tickers

The date and time in an event ticker are the close, in Eastern time: 26 is the year, NOV02 the day, 1815 the time.

The series

Kalshi also lists 15-minute markets on gold, oil, currencies and Treasury yields.

Prices and fees

These markets trade in 0.1¢ steps below 10¢ and above 90¢, and 1¢ steps in between. The taker fee per contract is 0.07 × P × (1 − P), rounded up to the cent per order: 1.75¢ a contract at 50¢. The backtesting guide covers fees and rounding in detail.

What the files hold

• settlements: each market's target (floor_strike), the closing average it settled on (expiration_value) and the result.
• markets: rules, open and close times, strikes and tick bands.
• book_l2, book_snap_1s, bbo, bbo_1m and trades: every order book change and trade, plus ready-made top of book and minute bars. A day of KXBTC15M book changes is about 260 MB.
• spot: best bid, ask and last for each coin on Coinbase, Bitstamp and Kraken, every update, to see the underlying move during the window.

A day of outcomes

SET TimeZone = 'UTC';

-- target: the opening 60-second average. settled_on: the closing one.
SELECT ticker, floor_strike AS target, expiration_value::DOUBLE AS settled_on, result, close_ts
FROM 'kalshieye-data/settlements/date=2026-11-01/settlements-2026-11-01.parquet'
WHERE series = 'KXBTC15M'
ORDER BY close_ts;

The price five minutes before the close

How often YES won, grouped by where the YES mid stood five minutes before each market closed:

SET TimeZone = 'UTC';

WITH settled AS (
  SELECT ticker, result, close_ts::TIMESTAMPTZ AS close_ts
  FROM 'kalshieye-data/settlements/date=2026-11-01/*.parquet'
  WHERE series = 'KXBTC15M' AND result IN ('yes', 'no')
),
five_out AS (
  -- Each market's YES mid at the end of the minute that ended 5 minutes before its close.
  SELECT b.ticker, s.result, b.mid_close AS mid
  FROM 'kalshieye-data/bbo_1m/date=2026-11-01/series=KXBTC15M/*.parquet' b
  JOIN settled s USING (ticker)
  WHERE b.minute::TIMESTAMPTZ + INTERVAL 1 MINUTE = s.close_ts - INTERVAL 5 MINUTE
)
SELECT floor(mid * 10) / 10 AS mid_bucket,
       count(*) AS markets,
       round(avg((result = 'yes')::INT), 2) AS yes_rate
FROM five_out
WHERE mid IS NOT NULL
GROUP BY mid_bucket
ORDER BY mid_bucket;

One day is 96 markets, too few to read anything into. Run it over many days, and see the backtesting guide for entry prices, fees and recorder gaps.

Questions

Next

• KXBTC15M data: recent results, datasets and code.
• Kalshi order book data: how the book works and how to rebuild it.
• How to get Kalshi historical data: what Kalshi's own API gives you.