Jev Trading Bot Demo

Only the choice. The documented step is a price feed as state and buy or sell as the question; the post says nothing about position sizing, entry limits, risk rules, or what happens on a low-confidence answer. All of that would live in the surrounding code, and the post does not show it.

Mostly Monad. The 300ms cadence is the chain's block time rather than the model's latency, so the demo's tempo is set upstream of Jev. The post never reports how long an individual decision took, how many calls a block consumed, or what the per-call cost was.

No numbers at all: no P&L, win rate, drawdown, or comparison against a baseline appears anywhere. A reply in the thread dismisses the idea outright, and the author answered by open-sourcing the repository the next day rather than publishing results.

Yes. The author open-sourced jarrodwatts/jev-trader, described as one AI trade decision per Monad block on Kuru's MON-USDC pair. Since it acts on a real on-chain order book, read how order placement and sizing are handled before pointing it at funds you cannot lose.

Jarrod Watts, who lists himself as lead AI engineer at Monad, the chain the bot uses. So the demo comes from inside the platform hosting the orders rather than an unaffiliated builder, which is worth weighing when reading the pitch, though he did open-source the code the next day.

Kuru's MON-USDC pair on Monad. The repository describes it as one AI trade decision per Monad block on Jev, and the demo runs against that single pair. How the asset pair is chosen, or whether other pairs were tested, is not stated.

At jev-trader.vercel.app, linked from the post alongside the video. It is a hosted demo page rather than a self-hosted path, and the post gives no note on whether it still trades live funds or is a recorded interface, so check the repository before assuming what running it does.