← The Evidence

Exhibit B Investigation · 2026-08-02 · 2 min read

Can You Beat Kalshi's 15-Minute Bitcoin Markets?

Verdict: Overturned
A coin flip, taxed every fifteen minutes. The house edge is structural.

The charge. Kalshi lets you bet on whether Bitcoin will be higher or lower fifteen minutes from now, around the clock. The pitch is intoxicating: fast, simple, binary, and it feels like the kind of short move a sharp eye could call. We tested whether anyone actually can.

The evidence — is it predictable?

We pulled every 15-minute Bitcoin move from the last week and a half — 717 of them — and asked whether the past predicts the next one:

At fifteen minutes, Bitcoin is a random walk. There is no trend to ride and no bounce to fade; the next bar doesn't know what the last one did.

The evidence — what it costs

Here is the part that turns "no edge" into "guaranteed loss." We pulled a live 15-minute market. At the moment a window opens, the contract sits near 50¢ — a true coin flip. To buy it you pay the ask (around 51¢) plus Kalshi's fee, which on crypto markets peaks right at 50¢. So you need to be right about 53% of the time just to break even — on an event that is genuinely 50/50. That ~3-point gap is a toll, and you pay it ninety-six times a day.

The window we looked at had also traded zero contracts. Even with an edge, you couldn't get filled.

The verdict

You can't beat the 15-minute market with a view, because there is no view to have — and the fees make a coin flip a slow, certain bleed. It's the same wall that kills naive crypto arbitrage: the opportunity is smaller than the cost of taking it.

Could anyone win here? Only by changing games entirely — providing liquidity instead of taking it, or trading order-flow microstructure at low latency. That's a market-making business with real capital at risk, not a "call the next candle" bet. For everyone clicking up or down every fifteen minutes, the house isn't cheating. The math simply isn't on your side.

← Back to The Evidence · See the graded record