Yes, and it happens every day. In 21,741 Kalshi Bitcoin 15-minute windows from March 24 to November 8, 2026, the price swung over to the other side at least once in the last 3 minutes in 20.48% of them, about 1 in 5. A swing is not always a loss: 41.1% of those swings swung back, and the side that was favored with 3 minutes left still won. Overall that favorite lost 12.07% of the time, about 1 in 8.
Expensive favorites swing less, but they still swing. Of favorites priced 90 cents or more with 3 minutes left, 4.87% saw the price cross to the other side and 2.86% lost. When a swing happened, it came back about 4 times in 10 at every price. Every loss starts as a swing, so a swing tells you the window is in play, not how it ends.
All figures are frozen on Kalshi KXBTC15M windows that closed from March 24 to November 8, 2026 (US Eastern time).
Three plain definitions, so the numbers mean something.
So the swings split cleanly in two: 1,830 that swung back and 2,623 that stuck. Nothing was left out for being a loss; 18 of the 21,759 windows in the period had no usable price with 3 minutes left and are not in the counts.
The cheaper the favorite, the more often the price crossed over. At 50 to 59 cents it crossed 81.1% of the time; at 99 cents and up, 0.60%.
| Favorite's price, 3 minutes left | Windows | Swung to the other side | Of swings, swung back | Favorite lost (95% range) |
|---|---|---|---|---|
| 50 to 59 cents | 1,362 | 81.06% | 42.9% | 46.26% (43.62 to 48.91) |
| 60 to 79 cents | 3,641 | 53.14% | 40.8% | 31.50% (30.01 to 33.03) |
| 80 to 89 cents | 2,654 | 27.43% | 39.0% | 16.73% (15.36 to 18.20) |
| 90 to 95 cents | 3,794 | 12.10% | 40.1% | 7.25% (6.47 to 8.12) |
| 96 to 98 cents | 5,083 | 3.88% | 43.1% | 2.20% (1.83 to 2.64) |
| 99 cents and up | 4,992 | 0.60% | 46.7% | 0.32% (0.20 to 0.52) |
Read down the "swung back" column and it barely moves: from 39.0% to 46.7% across the price rows. A swing against a 97-cent favorite came back about as often as a swing against a 65-cent one. What changes with price is how often a swing happens at all.
The "favorite lost" column is close to what each price implies: in every row the gap between how often the favorite won and its average price was inside normal luck (under 2.64 standard errors for six rows). That is the same thing our minutes-left study and 99-cent study found on their own samples. This page adds the ride: how often the price crossed over before the end.
It happens. Of 14,084 favorites priced 90 cents or more with 3 minutes left, 686 (4.87%) saw the price cross to the other side. 283 of those came back and won; 403 lost.
Counting only the end of each minute (2:00 and 1:00 left), 350 of these 90-cent-plus favorites were priced under 50 cents at a minute mark. Most swings are shorter than a minute and never show on a one-minute chart.
The arithmetic is the hard part. A 95-cent contract pays 5 cents when it wins and costs 95 cents when it loses, so one loss wipes out about 19 wins. A swing that comes back does not change that math.
Look only at the final 60 seconds, starting from the side favored with 1 minute left: the price crossed over in 6.89% of 21,741 windows, and that favorite lost 4.21%. So 38.9% of last-minute swings came back.
Per day: across 230 days, about 19.4 windows a day had a swing in the last 3 minutes, about 8.0 swung back and about 11.4 stuck. Kalshi runs 96 of these windows a day.
Our BTC logger reads the market twice in each window, about 5 minutes and about 3 minutes before the close. In 15,389 windows where the side we logged was priced 50 to 99 cents at the first read, that side was priced under 50 cents by the second read 13.17% of the time. The logger stores the ask for one side only, so for the other side we estimate 100 minus its ask, which ignores the spread. It is a rough check from a different source, and it points the same way: the lead changes hands late, and often.
Our paid feed sends its call about two minutes and fifty seconds before most windows close, inside the same late-window market this page measures, swings included. Every call is graded in public: 91.0% won across 13,784 graded alerts at a 91.5 cent average price. At that price you have to win that often just to break even, so the record sits below breakeven before Kalshi's fee and further below after it. Nothing on this page shows our alerts doing better than Kalshi's own price. Why a high win rate is not profit is explained in does a high win rate make money on Kalshi?
Yes. In 21,741 windows from March 24 to November 8, 2026, the price swung to the other side at least once in the last 3 minutes in 20.48% of them. 41.1% of those swings swung back and the 3-minute favorite still won; the favorite lost 12.07% of the time overall.
In our sample the side priced above 50 cents with 3 minutes left lost 12.07% of the time, about 1 in 8, close to what its average price of 88.47 cents implied.
Not always. When a side favored with 3 minutes left saw the price cross to the other side, it still won 41.1% of the time. That is a measurement, not a reason to hold or sell; every loss also starts as a swing.
Yes, rarely. Favorites priced 90 cents or more with 3 minutes left saw a swing 4.87% of the time and lost 2.86%. At 99 cents and up the swing rate was 0.60%. One loss at 95 cents costs about as much as 19 wins pay.
No. Our alerts go out about two minutes and fifty seconds before the close, so they face the same late swings. Every alert is graded in public, and the record sits below breakeven at its logged prices before Kalshi's fee.
We used Kalshi's official result for every KXBTC15M window that closed from March 24 to November 8, 2026 (21,759 windows) and its public one-minute price history (Yes bid and ask: open, high, low and close for each minute), pulled from Kalshi's public API on November 9, 2026. 21,741 windows had a two-sided quote at the end of the minute that left 3:00 on the clock with a midpoint not exactly 50 cents. The favorite and its price come from that minute. A swing counts if, in any of the last three one-minute bars, the low of the Yes ask fell below 50 cents (Yes favorite) or the high of the Yes bid rose above 50 cents (No favorite). An empty order book pushes the ask up and the bid down, so it cannot trigger this test. Every window is graded on Kalshi's official result.
The cross-check uses our own BTC logger (the files behind our public results log): one read about 310 seconds before the close and one about 170 seconds before. Those reads are regraded on Kalshi's official result. We do not use the logger's own outcome flags or its recorded opening price, which is not Kalshi's official strike on most windows.
This page publishes totals only. No exchange price is published. DegenHedge is independent and is not affiliated with or endorsed by Kalshi.
Related: will Bitcoin go up or down in the next 15 minutes? · can a 99-cent contract lose? · the 12:30 rule, tested · exactly at the strike · when Bitcoin moves big
Free to read, free to check. Want to watch our calls get graded before paying anything? Follow the free @DegenHedgeProof Telegram channel; the paid feed is optional.
Every signal in our BTC 15-minute feed goes into a public results log and a downloadable CSV, graded against Kalshi's official result, losses included, with nothing pulled. The same record is available as JSON from our read-only API.
This is a free trial that becomes an automatically renewing subscription. Eligible new customers receive a 5-day trial with a payment card and $0 due today. Returning trial users see $4.99 due today before checkout. At the end of it your card is charged $4.99 and then $4.99 every month until you cancel. Cancel any time before the trial ends and you pay nothing: use your account page, send /manage in the bot, or email [email protected]. Cancelling later stops future renewals at the end of the current paid month and does not refund it. No refunds except as required by law. By starting a trial you authorize these recurring charges and agree to the Terms and Privacy Policy. Signals are informational and for entertainment only, are not financial or betting advice, and any single signal can lose 100% of your stake. Must be 18+ (some platforms require 21+). If gambling is a problem for you, call 1-800-GAMBLER.
These are 15-minute binary prediction-market contracts. Any single window can lose 100% of your stake. They settle on a 60-second index average, not on the exchange chart you may be watching. At high prices a loss costs many times what a win pays, and Kalshi's fee raises the win rate you need above the price. Nothing on this page is financial, investment, or betting advice.
Past performance does not guarantee future results. Trade only with money you can afford to lose. Must be 18+ or of legal age where you live. If gambling is a problem for you, call 1-800-GAMBLER.