The Fee-Gated Scalper Still Paid the Toll
This week I gave a small online model one hour and a fresh $1,000 PredArena strategy account. The rule was simple: use live public order boo
PredArena Blog
Technical essays for builders testing Kalshi-style bots, AI agents, execution, slippage, fee modeling, and portfolio telemetry before live capital is involved.
This week I gave a small online model one hour and a fresh $1,000 PredArena strategy account. The rule was simple: use live public order boo
This week I gave a small model one hour and a fresh $1,000 PredArena strategy account. The rule was simple: use live public order books, kee
I built a small fourth-quarter model, checked Kalshi and Polymarket, placed one PredArena paper trade, and tracked whether it made money.
How fast is the prediction market paper trading loop, and where do the milliseconds actually go? We measured both to find out.
A prediction market bot can be right and still lose. The missing object is usually not the model. It is the fill.
The spread on the screen is not the trade. It is the rumor of a trade.
A bot without a ledger eventually starts believing its own story.
Latency is not one number. For an agent, the useful question is where the time goes and whether p99 changes the trading loop.
In late May 2026, MicroStrategy quietly sold 32 Bitcoin. It was the company’s first Bitcoin sale since 2022, breaking a long-held.
The easiest way to fool yourself with a prediction market bot is to give it fills it could never have received.
The risky part of an AI prediction market agent is often not the model. It is everything around the model.
Before a Kalshi-style bot trades live, make sure the boring parts work. Most expensive bugs are boring.