What is Allora?
The Allora Network is designed to crowdsource financial alpha produced by machine learning models, powered by our Proof of Alpha consensus mechanism.
The Allora Network is a machine learning (ML) model network that is incentivized to contribute to the collective optimization of some underlying objective function. Machine intelligence networks are composed of multiple subnetworks, each defined by their own objective function, anything from predicting the price of an asset at some time in the future to constructing portfolios to optimize for some risk/return profile. This creates a network with vastly more applicability than the capabilities of its individual subnetworks.
Machine learning and artificial intelligence (AI) are powerful in many settings. They harness vast computational power and data-driven algorithms to uncover insights, make predictions, and drive optimizations that are beyond the scope of human cognition or what can currently be expressed on-chain. Machine intelligence networks allow us to utilize this power in a decentralized form factor, enabling us to build significantly more advanced on-chain primitives.
In the case of Allora, AI-powered Decentralized Finance (DeFi) represents its most immediately useful primitive. The network is optimized for financial applications and incentivizes network participants to contribute their asymmetric insights around financial markets (i.e. their “alpha”). Network participants submit their alpha in the form of data, predictive model features, and predictions, ultimately for the betterment of Allora.
The Proof of Alpha mechanism then aggregates and rewards network participants based on the usefulness of their contributions to optimize some objective function (such as minimizing mean absolute directional loss or maximizing Sharpe ratio). Additional security, in the form of verifiable computation, is built in for any applications that build on the network. This is achieved through a new zkSNARK proof system that is designed to verify the output of tree-based ML models. Leveraging this proof system is a new tool for predicting asset prices using ML models, what we’re calling zkPredictor. It has been built in collaboration with and powered by Modulus Labs. The result is a decentralized, self-improving intelligence network optimized for model financial markets.
The Allora Network and Proof of Alpha ushers in a new era for financial markets––where significantly more sophisticated and efficient financial infrastructure can be built leveraging decentralized machine intelligence. For example:
ML-Powered Yield-Generating Strategies: managing tokenized vaults with more advanced machine-powered yield-generating strategies.
Long-Tail Derivatives: more efficient lending, perpetual, and other derivative platforms built for exotic or long-tail asset types with pricing powered by the network.
Automated On-Chain Index Funds: index funds that are optimized and actively managed by the decentralized network of machine intelligence.
DeFi-Native AI Agents: Entire economies consisting of AI agents that autonomously execute on-chain actions based on the machine intelligence network.
ML-Informed MEV: Leveraging the network to create new proactive Maximal Extractable Value (MEV) strategies such as predictive arbitrage or multi-block MEV strategies.
Leveraging ML for Risk Management: Harnessing the collective price signals and insights generated by the Allora network to create risk-adjusted strategies.