About Amber Lab
We develop tools for collecting financial market data and study how neural networks perform on it.
Our approach
Our first project, AMBER, combines two kinds of work: building a data collection system and designing experiments that can be reviewed and repeated.
We use research notes to document the questions we ask, the checks we plan to run and the decisions that shape our experiments.
Read the research notesFuture research
One direction we’d like to explore is training a neural network to compete with top players in Mobile Legends: Bang Bang. The idea is to study decision-making in a team game and evaluate the agent in an authorised, controlled setting.