Bitcoin, by the hour
Kalshi lists Bitcoin contracts that settle every 15 minutes and every hour on the average price over the final minute. For each one, this page shows our model's probability next to Kalshi's price. The model knows something simple and measurable: Bitcoin moves far less on a Sunday morning than at the US market open. Whether that is worth anything after fees is a public test, recorded below with losses as visible as wins. Nothing here is a bet.
Now
Our replica of the settlement index is built from public exchange prices; we do not show the official index value.
Model vs Kalshi model, in shadow
Probability the contract settles YES. Gap = model minus Kalshi's mid, in percentage points. over fee marks a gap larger than Kalshi's fee plus a cent: what the shadow test would have recorded as a trade.
Top of the hour · settles 13:00 UTC (51 min)
| Contract | Kalshi bid / ask | Kalshi | Model | Gap |
|---|---|---|---|---|
| BTC at or above $77,500 | 97¢ / 98¢ | 97.5% | 97.3% | -0.2 |
| BTC at or above $77,600 | 93¢ / 94¢ | 93.5% | 94.0% | +0.5 |
| BTC at or above $77,700 | 84¢ / 85¢ | 84.5% | 86.5% | +2.0over fee |
| BTC at or above $77,800 | 68¢ / 69¢ | 68.5% | 70.7% | +2.2 |
| BTC at or above $77,900 | 43¢ / 44¢ | 43.5% | 46.6% | +3.1over fee |
| BTC at or above $78,000 | 22¢ / 23¢ | 22.5% | 24.2% | +1.7 |
| BTC at or above $78,100 | 9¢ / 10¢ | 9.5% | 10.9% | +1.4 |
| BTC at or above $78,200 | 4¢ / 5¢ | 4.5% | 4.9% | +0.4 |
| BTC at or above $78,300 | 1¢ / 2¢ | 1.5% | 2.3% | +0.8 |
This 15 minutes · settles 12:15 UTC (6 min)
| Contract | Kalshi bid / ask | Kalshi | Model | Gap |
|---|---|---|---|---|
| BTC at or above $77,769 | 97¢ / 97¢ | 96.8% | 96.7% | -0.1 |
When Bitcoin moves
A typical 15-minute move by weekday and hour (UTC; New York is UTC−4 in summer, UTC−5 in winter), measured over a year. Busiest: Wed 15:00 UTC. Quietest: Sun 06:00 UTC. The same $100 strike is a very different bet at those two times.
Desk record collecting
A pre-registered test, decided by rules written before the first trade: at least 500 shadow trades over 14+ days including a weekend, then the after-fee result must be positive with statistical confidence, and the model must beat Kalshi's own prices on accuracy. If it fails, it is retired and we say so here. Model fv1-0d7d9b1fd1.
Every trade, win or lose, is on the edge scoreboard. Raw numbers: /api/v1/btc/experiment.
Information, not advice. Prediction markets carry risk of total loss. DDN places no Bitcoin orders: this desk is measured in shadow until its test passes and the owner decides otherwise.