Pond, a company building a neural network model that learns from all existing on-chain data and behaviors, will join the Allora Network as one of the first ML models on the network. This collaboration marks a significant step forward not just for Allora and Pond, but also for the development of decentralized AI as a whole.
Pond’s foundation model is based on a sophisticated neural network that extracts advanced insights from various graphs, such as the blockchain transaction graph. Whereas GPT aims to predict the next word, Pond’s GNNs are aimed at predicting complex relationships. Essentially, GNNs predict on-chain behaviors. Based on these insights, Pond’s model can fundamentally change how users, developers, and applications perceive on-chain information and unlock new business models in DeFi, social, market predictions, security, and more. By partnering with Allora, Pond’s model generates real-time predictions for a large range of assets, including newly-launched assets with limited market data. The model generates these predictions at a high frequency, with a time-frame that will eventually be reduced to minutes or even seconds. Pond's model serves as a powerful tool for traders, developers, and dapps, empowering them to create personalized trading strategies or build applications that directly leverage the model's capabilities.
Pond's team boasts an impressive track record, with more than 40 papers published in prestigious journals and conferences such as IEEE, Nature Communications, Nature Human Behavior, ICML, NeurIPS, and PNAS with extensive engineering and research experience at world-leading companies and organizations. In addition to their academic achievements, Pond’s team has launched its inaugural community fund operated by Weezu Capital @WeezuTrading. This fund's portfolio has yielded approximately a 3000% return across various strategies, demonstrating the team's ability to translate their expertise into real-world success.
What Will Pond’s ML Model Do on Allora?
As one of the early ML model creators on Allora, Pond’s model provides crypto AI infrastructure that intelligently interprets diverse data points to predict price movements for small cap tokens and memecoins. By analyzing on-chain graphical behaviors and patterns, Pond's model determines what they refer to as a potential Alpha Rate, which represents the potential for significant price appreciation. This Alpha Rate is defined as the on-chain behavioral and structural similarity with tokens that previously increased by 2-3x in the upcoming days or weeks.
Beyond price predictions, Pond's model has the potential to be used for a wide range of applications. The model's ability to analyze complex data points and identify patterns opens up exciting possibilities for developers, such as creating innovative tools for social recommendations, risk modeling, graph analysis, on-chain liquidity efficiency improvement, and more. As Pond continues to refine and expand its model on the Allora Network, we can expect to see a wave of innovative AI-driven solutions that push the boundaries of what's possible in the crypto space.
"We are thrilled to bring one of the first ML models to the Allora Network," says Dylan Zhang, CEO of Pond. "While our current model has been successful, it's conservative compared to what we could achieve in the future. Allora's self-improving design will enable us to showcase far more aggressive strategies for token price predictions and incorporate even more data points, pushing the boundaries of what's possible in this space.”
Why Pond and Other ML Model Creators Bring Their Models to Allora
ML models exist in abundance, with over half a million on Hugging Face alone, and many more to come. These models represent the most powerful form of compute to ever exist. However, despite this species-level breakthrough, the AI landscape has lacked the necessary infrastructure to intelligently combine all of these models into a single form of collective intelligence. Without a decentralized means to harness the power of these models collectively, AI’s true potential remains untapped. This limitation has left a handful of AI monoliths with abundant access to compute and data as the primary producers of leading machine intelligence.
Allora offers a unique venue for ML model creators like Pond to solve this problem. The network facilitates collaboration by organizing itself into focused sub-networks called topics. These topics allow ML model creators like Pond to collaborate on optimizing specific ML objectives, such as predicting the price of a memecoin, with clear evaluation criteria that guide the interactions between participants.
Through this self-improving design, Allora empowers Pond to capture immediate value from their model’s performance and bridges the gap between ML model creators and developers who actively seek cutting-edge AI solutions: the network creates the shortest path for model creators to develop innovative models and for developers to build upon their work. This approach fosters a community where model creators like Pond and developers connect, collaborate, and make a tangible impact on the development of decentralized AI together.
"We're thrilled to have Pond as one of the first ML model creators on the Allora Network," said Nick Emmons, Co-Founder of Allora Labs. "Their comprehensive approach to analyzing hard-to-price, long-tail assets showcases the potential of decentralized AI and paves the way for more groundbreaking models to join the network."
About Pond
Pond is a pioneering crypto-focused AI company that has developed a powerful foundation model of crypto for predicting on-chain behaviors. With a team of experienced machine learning experts and a proven track record of success, Pond is at the forefront of leveraging AI to navigate the complex world of digital assets.
To learn more about Pond, visit their website and follow them on Twitter.
