state-of-the-lab
The Renegade Rule. A stranger explained the physics and it changed the bot the same day.
We sorted 2,366 of our own trades by distance to strike. Two thirds of them earned almost nothing.
state-of-the-lab
We sorted 2,366 of our own trades by distance to strike. Two thirds of them earned almost nothing.
state-of-the-lab
state-of-the-lab
state-of-the-lab
No. 8 · 16 Aug 2026 · All ponds State of the Lab — week ending Aug 16, 2026 The Heron · 2 min read · b0fbe64e…60bf72 2026-08-16T04:14Z This week the lab ran 105 seats across 6 desks and settled 23,613 trades, nearly all simulated and one desk of them real. Best
Studies
One market making bot, run simultaneously on paper and with real dollars on Kalshi. The gap between the two columns is the true cost of simulation, printed weekly.
state-of-the-lab
A month of automated trading experiments, almost all of which failed, and the single boring strategy that earned real money.
state-of-the-lab
The reproduction kit: every trade behind this week's board, the series aware fee change, and what the verification flags mean.
state-of-the-lab
Every seat on the board for the week ending Aug 9, 2026: 123 seats, the three book experiment, fee drag by desk, and the full calibration table.
state-of-the-lab
Before fees the lab came out at +$3.16 this week, essentially flat. After fees it lost $405. The predictions stopped losing. The toll did not.
Studies
Makers trade free on 13,003 Kalshi series and pay a quarter of the taker rate on 130 of them. The full list, the rounding trap, and the prefix trap, verified against real fills.
Studies
Follow a big Kalshi trade or fade it? 14,000 simulated tests of both directions. Fading never works; following only clears the fee at the expensive end.
Studies
Kalshi prices favourites honestly and overcharges for longshots by about 1.5 points. Measured across 19,838 trades in one week.