An agent-based prediction marketplace implemented on a blockchain
Software Engineering Group Project 2019
Everyday, energy retailers face the challenge of purchasing an optimal amount of energy from energy wholesalers. We set up a decentralised prediction market, where autonomous agents can place bets on the energy demand for the next day, on behalf of individual households. Agents are motivated to bet accurately due to monetary rewards, and energy retailers benefit from the 'wisdom of the crowd'. These agents make use of IoT data and machine learning strategies to make intelligent bets.
This was a fairly complex project, so we split up to work on two main portions: implementing the prediction marketplace with Ethereum, and implementing the autonomous betting agents.
I worked on implementing various machine learning strategies and deploying the autonomous agents to interact with the blockchain prediction marketplace.
Tech stack: Solidity, Truffle Suite, Python, Keras