Wallet V launches AI trading benchmark on Hyperliquid, Aster

Wallet V launched a public performance benchmark that evaluates AI trading agents on the Hyperliquid and Aster platforms. The project runs agents under the same conditions on both venues and publishes the results on public leaderboards.

The benchmark requires participants to submit agent code or Docker containers. Wallet V executes the submissions inside a controlled environment that uses consistent input data, capital constraints and evaluation windows. The platform supports simulated historical replay and live test environments that mirror order execution and market interaction on each exchange.

Each evaluation produces a set of standardized metrics. Wallet V publishes measures covering liquidity handling, execution costs and market impact for runs on Hyperliquid and Aster. The underlying datasets and the evaluation code are available for inspection alongside the leaderboards.

Documentation outlines the benchmark protocol, submission guidelines and exact calculation methods for each reported metric. Wallet V designed the framework to reduce variation in reporting and to make it easier for teams to compare algorithmic performance across venues.

The first public leaderboard tracks short-term trading agents using data from an initial evaluation period. Wallet V plans to add other strategy classes and longer evaluation windows over time. Results are updated regularly and historical leaderboards remain accessible for trend tracking and verification.

Hyperliquid and Aster were chosen to provide distinct liquidity profiles and execution models for side-by-side comparison. The benchmark applies the same test conditions to each platform so differences in outcomes can be traced to model behavior or to exchange mechanics and fees.

Wallet V offers technical support including sandbox credentials and a reproducibility package to help teams verify published runs. The testing environment enforces resource limits, input validation and process isolation to prevent unintended interactions with platform infrastructure during evaluations.

Benchmarks for trading algorithms exist in traditional finance and within some cryptocurrency projects. Wallet V intends the initiative for model developers, asset managers and platform operators who require comparable, verifiable evidence of how AI agents perform across different trading environments.