A Distributed Method for Cooperative Transaction Cost Mitigation

Battle Count: 88

Highly relevant to quantitative trading, particularly for multi-manager funds and systematic firms. The paper directly addresses transaction cost mitigation, which is a critical practical concern in algorithmic execution and portfolio management. The ADMM-based protocol provides a principled, privacy-preserving mechanism for coordinating trades across portfolio sleeves. The backtest demonstrates substantial Sharpe ratio improvements (from 1.60 to ~2.05) through cooperation. The work is by Stephen Boyd's group (cvxgrp), well-known in convex optimization for finance. Directly applicable to firms managing multiple independent portfolios with centralized execution.

Moderate complexity. The ADMM algorithm itself is well-established and straightforward to implement. The main complexity lies in: (1) designing the PM optimization problems with appropriate constraints, (2) choosing hyperparameters (rho, phi, D, K), (3) setting up the communication protocol between PMs and central entity, (4) implementing the transaction cost model with market impact parameters, and (5) integrating with existing portfolio management infrastructure. The mathematical formulation is clean and the algorithm is clearly specified in pseudocode. The GitHub repository provides reference implementation.

Code is publicly available on GitHub (https://github.com/cvxgrp/coop_t_code). The methodology is fully described with mathematical formulations and algorithm pseudocode. However, the LSEG (Refinitiv) market data is proprietary and not freely available. Synthetic alpha generation parameters and backtest setup are well-documented. The paper explicitly states it is a mathematical contribution and does not reference specific firm practices.