Market inputs
Phase-one idea generation uses HyperLiquid reference-market data. Kenka calculates prices, recent returns, ranges, volume and liquidity measures, source freshness, and named cross-market comparisons from approved inputs before any idea is considered.
Selection cycle
The engine reviews the eligible universe every four hours. It publishes three to five current ideas only when enough differentiated setups pass the evidence threshold. If a refresh fails or too few setups qualify, the last healthy set remains visible with its status.
Evidence gate
A candidate needs measured evidence across trend or return, volatility or range, and volume or liquidity behavior. It also needs relative context against at least two eligible named assets or a declared benchmark over the same interval.
Generic asset commentary, flat data, stale inputs, incomplete candles, zero usable volume, and unsupported claims fail the gate.
Model role
A reasoning model ranks eligible evidence bundles and drafts the thesis, why-now explanation, risks, horizon, and invalidation proposal. It cannot introduce a new asset, price, statistic, comparison, or news claim that is absent from the validated input.
Prediction and previous performance
Prediction is the forward-looking, time-bounded view expressed by the idea. Previous performance is the selected asset's measured 24-hour move before publication.
Previous performance provides context; it is never treated as evidence that the prediction has already succeeded.
Publication and lifecycle
Every published set is validated and stored atomically with its facts, comparisons, source time, model version, cost, and policy result. Ideas move through current, active, invalidated, closed, or expired states, and prior ideas remain available for outcome tracking.
Standardized outcomes
Each published idea is measured with the same $10,000 notional from the first available server price after publication. It exits at the first terminal update or maximum horizon and is compared with its declared benchmark over the same interval.
This record measures idea quality independently from a user's selected amount, timing, or exit decision. Results appear only after a stored outcome is complete.
Account-level review
Before confirmation, Kenka rechecks the live reference price, idea status, account cash, concentration, existing exposure, reserve-gold eligibility, post-investment LTV, and stress cases. Server rules determine the available amount and always take priority over browser input.