Adrian Millea
Adrian Millea is a machine-learning researcher affiliated with Imperial College London, working on deep reinforcement learning for financial trading. He is the sole author of “Deep Reinforcement Learning for Trading—A Critical Survey” (Millea 2021), an open-access review in Data (MDPI) 2021, 6(11), article 119, that surveyed 152 deep-RL-trading papers and is cited in this vault as the principal evidence for the field’s methodological fragmentation and reproducibility deficit.
Millea’s significance for this vault is that his survey is deliberately critical: rather than aggregating the literature’s many positive backtests into a profitability claim, it argues that the lack of shared benchmarks, common datasets and disclosed code makes such aggregation impossible. His subsequent research — including “Hierarchical Model-Based Deep Reinforcement Learning for Single-Asset Trading” (Analytics, 2023) and work with Abbas Edalat on deep RL with hierarchical risk parity for portfolio optimisation — pursues the constructive directions (hierarchical and model-based RL) that the survey identified as most promising. He is the named author the vault relies on for the position that RL-trading profitability claims should be treated with caution rather than pooled as an evidence base.
Connections
- Millea 2021 — proposes_model, 2021, source: https://www.mdpi.com/2306-5729/6/11/119
- Reinforcement Learning Trading Policy — replication_missing, 2021, source: https://www.mdpi.com/2306-5729/6/11/119
- Replication Crisis in Quantitative Finance — relates, source: https://www.mdpi.com/2306-5729/6/11/119
Adrian Millea [proposes_model] Millea 2021 Millea 2021 [supports] Replication Crisis in Quantitative Finance