Measured once per window, from the last 40 completed candles.
The room number is the mean height of the last 40 completed 15-minute candles times 0.69, which recovers the standard deviation of a full 15-minute move. It is re-measured once per window. Everything else — distance, clock, quote — updates on every tick.
Distance is measured from Kalshi's own strike, not from an exchange's candle open, because the strike is what the contract actually settles against.
A signal is suppressed whenever the gap between fair and the market sits inside two standard errors of the room number. Forty candles only pin the swing down to about ±9%, and without that filter one fairly-priced window in five looks like an edge. Expect most windows to say no trade. That is the tool working.
Kalshi settles on CF Benchmarks' BRTI, averaged over the 60 seconds before the close, against the same average before the open — not on the single Coinbase tick shown here. The two track each other closely but not exactly, which is basis risk this page does not model. The averaging itself, it does: remaining room is shrunk to account for it, since settling on a 60-second mean carries less variance than settling on one tick. Below 60 seconds left, that correction runs outside where it's been checked, and the tool refuses to produce a read at all rather than guess.
Fees use Kalshi's published shape: 0.07 × contracts × price × (1 − price), rounded up to the cent. Verify the live schedule before trading real money — it changes.
Typical candle height above, times two‑thirds. Don't measure — guess.
Each window has a hidden true volatility. The candles you see were drawn from it, so the chart genuinely contains the answer — your room number is a real measurement, not a guess at random noise.
The ticket price is sometimes fair and sometimes wrong, because the simulated market is using a stale estimate of how wild things are. Your job is to notice when it's wrong, not to guess direction. Direction is a coin flip and the tool will not reward you for calling it.
A decision counts as correct when you took a trade that really had an edge bigger than the fee, or passed on one that didn't. You can lose money on a correct decision and make money on a bad one. Over enough windows the decisions are what show up in the P&L, which is exactly why win rate is the least useful number on this page.
Fees use Kalshi's published shape: 0.07 × contracts × price × (1 − price), rounded up to the cent. Verify the live schedule before trading real money — it changes.