Kalshi 15 minute Bitcoin: three checks before you buy, from our real trades

Price plus fee, the spread, and how much is on offer at that hour.

At 7:57 pm Eastern on 1 November our bot Ping bought YES on a Kalshi fifteen minute Bitcoin market at 80 cents, with three minutes left. It settled NO. That one trade cost 81 cents, the price plus a 1.12 cent fee. This is what we check before buying one of these contracts, using our own real fills.

Check 1: the price plus the fee is the win rate you need

A contract at 80 cents pays a dollar if it wins. Add Kalshi's fee and you need to win 81.12% of the time just to break even. The fee on our real fills is 7 × price × (1 − price) cents, rounded up to a hundredth of a cent:

PriceFeeWin rate needed
50¢1.75¢51.75%
80¢1.12¢81.12%
90¢0.63¢90.63%
97¢0.21¢97.21%

Ping's real trades from 28 October to 1 November: 27 trades at an average of 83 cents, with an average fee of 0.98 cents, so it needed to win 84%. It won 23, 85%, for plus 21 cents in total. One more loss and the week is negative. That is how thin the margin is at these prices.

Check 2: the spread

The spread is the gap between the best price to buy and the best price to sell. On the Bitcoin fifteen minute markets it is usually 1 cent: that was the median across 146,184 quotes from 21 July to 10 October. When it is wider than that, something unusual is going on. In our limit order test, signals with spreads wider than 4 cents lost more whichever way we bought.

Check 3: how much is on offer at that hour

We measured the size at the best price, the smaller of the two sides, across 950,483 order book snapshots over 76 days. By the hour, Eastern time:

  • Thinnest: 5 pm, a median of 688 contracts, and 5 to 7 am, about 722 to 728.
  • Thickest: 4 pm, 1,108, and 7 pm, 1,104.

One contract fills at any hour. Size starts to matter as you scale up, and it matters most in the hours when the book is thin.

Paper against real

On the same 46 decisions Ping made from 27 October to 1 November, paper counted 39 fills. Real money got 27. Paper assumes every order fills. In the last minute, the offers go fast, so the real record is the one to trust.

What to take from it

  • Work out the win rate you need before you buy, fee included.
  • A 1 cent spread is normal. A wide one is a reason to wait.
  • Judge any strategy on real fills, not paper ones.

The video version goes up on Tuesday, and the complete guide covers how these markets settle.

Want to run rules like this yourself? Start with a $5 first month, Ping included. Ping counts a paper fill only when a real trade proves it could have filled.

Where this comes from: Kalshi's fill records for Ping's live trades, 27 October to 1 November 2026, and our own recorded Kalshi quotes and order book snapshots, 21 July to 11 October 2026. Results are our own computation. Nothing here is advice. It is the record.