Allora’s Forge Introduces Hyperliquid Log Return Topic for AI Predictions

Allora’s Forge has launched a new prediction topic focused on Hyperliquid, a high-performance blockchain platform optimized for perpetual futures and spot trading. The Hyperliquid Log Return topic, designated as topic 90, became available on October 8, 2026. This innovative topic invites AI and machine learning models to forecast three-minute log returns across Hyperliquid’s perpetual futures markets, with the unique feature of having no fixed asset list.

The operation of topic 90 is distinct from traditional prediction setups, as the market set for each round is dynamically assembled based on model submissions. Each model can submit predictions for up to 100 markets, and any market a model does not wish to predict can simply be omitted, avoiding unnecessary zero-padding. The accuracy of these predictions is evaluated using Hyperliquid’s one-minute candle data through Allora’s Atlas platform, with a scoring system that penalizes indecisive forecasts.

To ensure the reliability of predictions, Allora employs the Composite Zero-Agnostic Return (CZAR) loss function, which encourages models to take definitive stances rather than hedging their bets. Models must also pass seven statistical tests to qualify for mainnet deployment, including a requirement that their directional accuracy exceeds 50%, demonstrating a clear predictive edge over random guessing. This new topic adds to the existing 18 topics on the Forge platform, which collectively host over 2,000 active models focused on various financial forecasts.

© 2026 KLEA News. All Rights Reserved. This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.

Source: KLEA News