Prediction markets

Every evaluation on the protocol is a real, uncertain contest with a public outcome, and that makes each one a natural betting market. The prediction-market surface lets anyone stake on whether a specific trader's account will pass or fail, with odds set by the crowd and settlement decided by the same engine that decides the account itself.

How a market works

Each market is a binary parimutuel pool on one account: a pass side and a fail side, each a pot of real stakes. There's no order book and no market maker; the odds at any moment are simply the ratio between the two pots, and winners split the losing pot in proportion to their stake. Parimutuel was chosen deliberately because it stays honest at any size: a thin market has honest thin-market odds instead of a fake depth chart, and no counterparty has to exist for you to take a position.

Settlement nobody adjudicates

A market settles when the account it tracks resolves, and the resolution is the evaluation's own real outcome: the pass/fail decision made by the fixed rulebook, enforced by the engine, anchored by the same settlement machinery as everything else on the protocol. There is no oracle committee, no dispute window, and no judgment call, because the bet is on an event the protocol itself decides deterministically. The market page shows the trader's real live equity, the same number their evaluation actually runs on, so what you're watching is what will settle.

What you can't do yet, stated honestly

There is no early cash-out: once staked, your position rides to settlement. The interface shows a sell tab so the limitation is visible rather than hidden, but it's disabled because a real close-position mechanism doesn't exist yet, and we'd rather show you a disabled button than a fake market. If cash-out ships, it ships here first.

Where it sits on the honesty scale

The prediction-market surface is one of the protocol's younger layers: live and real, with the odds, stakes, and settlement mechanics above all running today, but at the reviewed tier rather than devnet-proven (What's live today explains the difference). Size your participation accordingly while it earns the same proof depth as the core spine.