{"id":30200,"date":"2026-06-15T15:03:03","date_gmt":"2026-06-15T14:03:03","guid":{"rendered":"https:\/\/investx.fr\/en\/2026\/06\/15\/wallet-v-public-benchmark-ai-trading-agents-hyperliquid-aster\/"},"modified":"2026-06-15T15:03:06","modified_gmt":"2026-06-15T14:03:06","slug":"wallet-v-public-benchmark-ai-trading-agents-hyperliquid-aster","status":"publish","type":"post","link":"https:\/\/investx.fr\/en\/crypto-news\/wallet-v-public-benchmark-ai-trading-agents-hyperliquid-aster\/","title":{"rendered":"Wallet V Unveils Public Benchmark for AI Trading Agents on Hyperliquid and Aster"},"content":{"rendered":"\n
A self-custody Web3 wallet<\/strong> has just reached an unprecedented milestone: making public the real performance data of hundreds of AI agents<\/strong> configured by its users on decentralized derivatives platforms<\/strong>.<\/p>\n\n\n\n Out of 688 agents<\/strong> analyzed, 42% finished in positive territory<\/strong> \u2014 and the top performer posted a ROI of +307%<\/strong>. A performance gap that raises as many questions as it answers.<\/p>\n\n\n\n Here is what this benchmark concretely reveals about the real state of AI algorithmic trading<\/strong> in 2026.<\/p>\n\n\n\n Wallet V<\/strong>, a Web3 wallet incubated by Virgo Group<\/strong>, has published an aggregated benchmark covering 688 AI trading agents<\/strong> deployed by its users over the past two months. These agents operated on Hyperliquid<\/a><\/strong> and Aster<\/a><\/strong> \u2014 two decentralized derivatives platforms<\/strong> \u2014 executing strategies on perpetual contracts<\/strong>.<\/p>\n\n\n\n Each agent was manually configured by the user, who also selected the large language model (LLM)<\/strong> responsible for generating trading decisions. The benchmark then aggregates performance by model family, covering seven distinct LLMs<\/strong>. Models represented by fewer than 10 agents are flagged as directional only, with no conclusive statistical significance.<\/p>\n\n\n\n The raw results: 42% of agents recorded a flat or positive P&L<\/strong> over the period. Peak ROI ranged from -30%<\/strong> for the worst-performing model to +307%<\/strong> for the best. A spread of 337 percentage points that illustrates just how decisive the choice of LLM<\/strong> \u2014 and configuration \u2014 can be.<\/p>\n\n\n\n The agents in the benchmark did not operate exclusively on crypto assets. They accessed four asset classes<\/strong> available on Hyperliquid<\/strong> and Aster<\/strong> via perpetual contracts<\/strong>:<\/p>\n\n\n\n This diversification reflects the broader ambition of Wallet V<\/strong>: to move beyond the crypto market alone and offer an agent infrastructure capable of operating across all tokenized financial markets<\/strong>. Adam Cai<\/strong>, founder and CEO of Virgo Group<\/strong>, sums up the approach: “Users are now choosing their AI model the way institutions evaluate fund managers \u2014 by examining observable performance over time.”<\/em><\/p>\n\n\n\n Virgo Group<\/strong> is backed by investors including Draper Dragon<\/strong>, OKX Ventures<\/strong>, and Cobo Ventures<\/strong>. The benchmark is hosted directly on the Wallet V<\/strong> website and updated continuously as new agents are deployed.<\/p>\n\n\n\n Wallet V<\/strong> has announced several developments for upcoming iterations of the benchmark. On the roadmap: the integration of new LLM families<\/strong>, support for prediction markets<\/a><\/strong>, advanced analytics features for copilot trading<\/strong>, and AI prompt generation personalized to each user’s trading style.<\/p>\n\n\n\n That last feature is particularly noteworthy: it points to a continuous improvement loop<\/strong> where the agent adapts to the user’s risk profile and preferences rather than applying a one-size-fits-all strategy. This is precisely the kind of personalization<\/strong> that sets an institutional-grade tool apart from a basic crypto trading<\/a><\/strong> bot.<\/p>\n\n\n\n Wallet V<\/strong>‘s public benchmark<\/strong> represents a rare initiative within the ecosystem: most AI trading agent<\/strong> solutions remain black boxes. Making this data accessible \u2014 even in aggregated form \u2014 introduces a standard of transparency and accountability<\/strong> that could set a precedent across the decentralized algorithmic trading<\/strong> sector.<\/p>\n\n\n\n688 Agents, 7 LLM Families: What the Numbers Really Say<\/h2>\n\n\n\n
<\/figure>\n\n\n\nBTC, ETH, Gold, Forex: The Asset Classes Covered by the Agents<\/h2>\n\n\n\n
\n
What the Next Versions of the Benchmark Will Change<\/h2>\n\n\n\n