Hourly Bitcoin markets: implied volatility and calibration

Crypto traders on Kalshi trade the same question many times a day: where will Bitcoin be at the top of the hour? Because a new event lists every hour, the market produces a dense record of forecasts that you can read and test.

Price	Markets	Avg price	Resolved YES
0¢–10¢	2120	1¢	0%
10¢–20¢	26	14¢	8%
20¢–30¢	16	25¢	19%
30¢–40¢	12	34¢	33%
40¢–50¢	12	43¢	42%
50¢–60¢	15	55¢	40%
60¢–70¢	12	66¢	58%
70¢–80¢	18	75¢	78%
80¢–90¢	23	86¢	87%
90¢–100¢	1398	99¢	99%

Price 30 minutes before settlement. Public Kalshi data, Oct 7, 2026 – Oct 8, 2026; buckets with fewer than 10 markets omitted. Points on the diagonal are perfectly calibrated. Past results, not a forecast.

#How the market is structured

Each hour has two Bitcoin events:

• Range (KXBTC): mutually exclusive brackets, roughly $100 wide, plus open tails. Exactly one settles YES.
• Above/below (KXBTCD): nested "above $X" strikes. Each price is the market's probability that Bitcoin finishes above that level.

Both settle on the simple average of the CF Benchmarks Bitcoin Real-Time Index (BRTI) over the 60 seconds before the event time, not on any single exchange's last trade. Ethereum, Solana, XRP and Dogecoin have similar series.

#Step 1: Read the next hour

From the above/below ladder, each strike's mid price is the probability of finishing above it. The probabilities should fall as the strike rises. Where a noisy book breaks that order, take the running minimum before you use it. The strike where the probability crosses 50% is the median. The strikes at 90% and 10% bound an 80% range.

A hypothetical ladder at 4pm for the 5pm settlement:

The median sits just above $77,000, and the 80% range is roughly $76,600 to $77,400.

#Step 2: Implied volatility

The width of that range is the market's expected one-hour move. Divide the distribution's standard deviation by the current price for an hourly volatility. Multiplying by the square root of the hours in a year gives an annualized figure for comparison with other volatility measures. That is an assumption about how hourly moves scale, so label it.

Compare it with realized volatility, the actual hour-to-hour changes between settlements. If implied stays above realized for many hours, the market is pricing more movement than it gets.

#Step 3: Test the tails

Collect a few days of settled above/below markets, each priced 30 minutes before its settlement. Each hourly event opens only an hour before it settles, so earlier prices do not exist. Bucket the contracts by that price, for example 0–10¢, 10–20¢ and so on, and compare each bucket's average price with how often it actually paid. Cheap contracts that pay less often than their price says are overpriced longshots. The Market calibration guide covers margins of error and the Brier score.

#Pitfalls

• Volume of data. About 190 range brackets list every hour, so a single day is thousands of rows. Work in days, not months, and keep the strikes whose price is between a few cents and a few cents short of a dollar.
• Linked contracts. The strikes in one hour move together. Count hours, not contracts, when you judge whether a gap is real.
• Regime changes. A quiet weekend and a volatile news day behave differently. Report the window you tested.
• Fees and spreads. A calibration gap at 5¢ can vanish after the spread and fee. Work the numbers in Expected value and edge.

#Run it in Jordan

"Is Kalshi's hourly Bitcoin market well calibrated?" is one of the starting questions in Jordan. Jordan collects a bounded window of settled hourly markets with each contract's price 30 minutes before settlement, builds the calibration table and chart with formulas, and keeps the full dataset on the sheet so you can change the bins or the window.

#Frequently asked questions

#What price does a Kalshi Bitcoin market settle on?

The simple average of the CF Benchmarks Bitcoin Real-Time Index over the minute before the event time. A spike on one exchange in the final seconds does not decide it.

#Why price contracts 30 minutes before settlement?

The last trade before settlement is close to 0 or 1, because the outcome is nearly known. A fixed earlier time tests what the market believed when the outcome was still uncertain.

#Can I compare hourly implied volatility with options markets?

Yes, as research, as long as the options figure comes from a dated source and you annualize both the same way. The two markets differ in fees, horizon and participants, so a gap is a question, not a signal.