What does the strategy actually do?
Most windows, nothing. The rule looks once an hour, at the window that closes on the hour, with five minutes left. It needs three things:
- A move. Bitcoin has to be up or down at least 0.036% over the prior three minutes.
- A price. The side it moved toward has to cost 65¢ or less at the ask, fee included. The cap arrived with version 3 on Nov 2.
- A clock. Windows closing between 5 p.m. and midnight ET are sat out.
The bot trading this rule on our own Kalshi account holds at most $3 a position and stops for the day at $25 of losses. Those are our limits, set to learn, not a sizing suggestion. If you want one, run the Kelly calculator on your own bankroll.
What has it returned?
Numbers are paper fills at the ask, net of the taker fee, read from our published record on Nov 6, 2026.
| Version | Period | Fills | Wins | Net per contract | Status |
|---|---|---|---|---|---|
| v1 | From Sep 23 | 178 | 126 | +4.05¢ | Retired |
| v2 | From Oct 17 | 58 | 32 | +11.18¢ | Retired |
| v3 | From Nov 2 | 12 | 5 | +6.46¢ | Running |
Read v2 with care. Its first half made +27.06¢ a contract and its second half lost 4.70¢. Twelve fills on v3 tell you nothing yet: its 90% lower bound is −20.93¢. The live record, updated all day, sits on Bitcoin Edge 15-Min.
What did not work?
These lanes are on the record too, because they are the honest part of any strategy guide.
- Resting late orders. Our 10¢ maker lane took 936 fills and netted +0.04¢ a contract, which missed its pass line. The 15¢ version lost 2.50¢ on 248 fills and was retired.
- The three common bot trades. On 2,710 windows, two lost after fees and one broke on settlement.
- The 12:30 rule. The popular advice to stop buying favorites after 12:30 did not hold up on 1,258 windows of tape. There was no cliff.
Two newer lanes are still running on small samples. Late velocity shows +14.37¢ on 42 fills, and the late-underdog lane shows −2.66¢ on 61. Neither is a result yet.
Why does the fee decide this?
Kalshi's taker fee is charged at the fill, win or lose, at 7% × price × (1 − price). That is about 1.75¢ a contract at 50¢ and about 0.63¢ at 90¢. On a contract with a one-cent edge, a coin-flip price eats the whole gain. Resting orders pay none on this series, but they only fill when someone takes them, which is why the maker lanes above look the way they do. The fee calculator prints the exact cost for any size.
What is scalping on Kalshi?
Scalping is buying and selling inside one window for a small gain instead of holding to settlement. It needs repetition, and the 15-minute series are one of the few places that offer it. The gain has to clear the fee on the way in and the way out. Our record holds positions to settlement, so we have no scalping number to show you.
Does a free Kalshi bot work?
A free bot pays the same fee as a paid one, and it usually runs one of the three trades we already tested. Before you trust one, ask what its rule is, how many fills back it, and whether the number is after fees. If it cannot answer, check the idea yourself. Bitcoin Edge 15-Min shows the live window, the bitcoin speed reading, the fee and the breakeven for free, and the 15-minute hub covers the other series.
What would break this strategy?
Fills that stop clearing the fee. Version 1 and version 2 both ended when something changed, and version 3 will too. When it does, the version number moves and the old record stays up. If you trade it, trade it knowing that, and trade responsibly.