RBC Capital Markets deploys Aiden deep reinforcement learning platform to minimize VWAP slippage in European markets

Traditional pre-set algorithmic trading strategies struggled with slippage and alpha erosion during periods of volatility. Historical models could be upended by shifting market conditions, requiring frequent manual re-coding to maintain effectiveness. Traders needed a solution that could adapt in real-time to dynamic liquidity patterns and market microstructure changes.

RBC and its Borealis AI research institute developed Aiden, an electronic trading platform using deep reinforcement learning to adjust to market conditions in real-time. The Aiden VWAP algorithm processes more than 200 market data inputs through a deep neural network, learning autonomously from performance both during and after each trade. A companion Aiden Insights explanation system provides clients real-time visibility into how the algorithm is adapting its execution decisions.

After debuting in North America in 2020, Aiden demonstrated the ability to swiftly adapt to volatile market changes while preserving performance without frequent manual re-coding. The platform was subsequently launched in UK and European markets in 2024, extending its adaptive execution capabilities to those regions. A second algorithm, Aiden Arrival Price (launched in North America in 2022), expanded the platform's capabilities with over 300 data inputs and a more flexible execution trajectory.

• Deep reinforcement learning enables trading algorithms to self-optimize continuously, removing the need for manual recoding during volatile periods.
• Explainability tools (Aiden Insights) are critical for client adoption of black-box AI trading systems, giving coverage teams a way to communicate real-time decisions.
• Expanding from 200 to 300+ inputs across algorithm generations reflects a deliberate evolutionary roadmap, not a single deployment.