There is no evidence of rigging in the record: every number this site shows is written down the moment it is shown and scored against the settled result, and the market comes out honest. What people feel is the last minute. The contract settles on a 60-second average of the CF Benchmarks index, not on the last print, so a window can flip in the final seconds for a reason you cannot see on a spot chart.
Kalshi does not settle the 15-minute Bitcoin contract on the price at the bell. It takes the simple average of the CF Benchmarks real-time index over the 60 seconds before the window closes, and compares it with the same average taken a quarter-hour earlier. Up pays if the later average is higher.
Two things follow. With 30 seconds left, half of the settlement is already fixed: the first 30 seconds of the average have happened and cannot change. And the average lags a fast move: a print that crosses the strike with 10 seconds left moves the 60-second average by only a sixth of that distance. A spot chart shows the last print. The contract pays on the average. Those are different numbers, and the gap between them is where "rigged" comes from.
Sardine publishes a count, not a forecast: how often a contract this far from the strike with this much time left settled Up before. Every count shown is logged and scored. If the market were rigged against buyers, Kalshi's own price would be a worse guide to the result than the count. It is not. It is usually the better one.
Reading the record…
The full scorecard, bucket by bucket, is on the record.
A count, not a forecast. Kalshi's price is usually closer to the result than history is; the windows where the two split are the ones worth a look.