BTC up or down, fifteen minutes

Straight answer first: rounds here are five minutes, not fifteen. Fifteen is the other common window, and rather than argue for ours, this page measures the difference — on real Binance candles, over the same ten days.

Where fifteen-minute BTC markets are

Bitcoin up-or-down windows are usually offered at 5 minutes, 15 minutes, an hour and four hours. Polymarket, Kalshi and Limitless each run the fifteen-minute version as a market you trade in, where the price is whatever the other side is willing to take. This site runs only the five-minute window, priced by a model against a published spread. If fifteen minutes is what you want to trade, those are the places that have it — nothing below is an argument that they should not.

What fifteen minutes changes: the size of the move

Price moves grow with the square root of time, so three times the window should be about 1.73 times the move. Measured over the same window — Binance BTCUSDT, 11 to 21 September 2026, BTC around $84,700 — it is about 1.6:

Two things follow. A fifteen-minute window is decided by about sixty-five dollars where a five-minute one is decided by forty — bigger, but nothing like three times bigger. And the tie, which refunds every bet here, effectively stops existing: it needs the close to land on the open to the cent, and over ten days not one fifteen-minute candle managed it.

What it does not change: the shape of the game

This is the part nobody publishes, and it is the reason the window length matters less than it looks. Take the side that is in front a given fraction of the way through the window, and ask how often it is still in front at the close. Same ten days, both window lengths, both read off the same one-minute grid:

The two columns are the same column. A fifteen-minute market is not a calmer market or a more predictable one — it is a five-minute market with the dollar axis stretched by about 1.6 and the clock by three. Anyone telling you a longer window is easier to read is describing a feeling, and the feeling is not in the candles.

Where the length does show up: in the price

Time left is half of what a side is worth. The same forty-dollar lead is worth much less with more of the window still to run, because σ·√t is larger:

d      = ln(price / price to beat) / (σ · √seconds left)
chance = Φ(d)

So the longer window pays better on the side in front and worse on the side behind, for the same move — which is arithmetic, not generosity. Nothing about it changes the margin: the spread is the same 3¢ either way, and what that costs is published.

Why five minutes here

Given the two are the same game, the case for the shorter one is practical rather than theoretical:

• You see the whole thing. Five minutes is a round you watch from open to close without leaving it running in a tab. Fifteen is long enough to walk away from, which is a different product.
• Three decisions instead of one. In the time one fifteen-minute window takes, there are three rounds here, and you can stop after any of them. There is no position to exit, because there is nothing to exit — a bet is settled by the candle, not sold back.
• The volatility estimate is refreshed constantly. σ is re-read every round from live trades, so pricing follows a quiet hour or a violent one within minutes rather than within quarters of an hour.
• The tie still exists. One five-minute round in about a hundred and ninety lands exactly on its open and hands everybody their stake back. Over fifteen minutes that refund is gone.

And the honest entry on the other side of the ledger: three rounds means three spreads. A player who bets every round of a fifteen-minute stretch pays the margin three times, not once. That is the real cost of the shorter window, it is not hidden anywhere, and it is the same arithmetic as the house-edge page — the edge is charged on turnover.