Low-Turnover Rebalancing for Sparse Index Tracking

Battle Count: 62

Directly relevant to passive portfolio management and index-tracking fund operations. The paper addresses the practical problem of maintaining sparse portfolios with low turnover, which is a core operational concern for quantitative portfolio managers. The uncertainty-gated rebalancing framework provides a principled decision-support mechanism for when to trade versus hold. However, it focuses on tracking (passive) rather than alpha generation, and lacks explicit transaction-cost modelling, which limits direct applicability to active quantitative trading strategies.

The pipeline involves multiple computational components: SAPG empirical-Bayes calibration, proximal gradient/FISTA optimisation, Moreau-Yosida smoothing, preconditioned MALA posterior sampling, loss-scale grid search, posterior support screening, directional gating, recovery-mode triggering, and bounded local implementation with funding rules. Each component requires careful numerical implementation. The decision logic has multiple interacting thresholds and asymmetric gates. However, the paper provides detailed algorithmic descriptions, a GitHub repository, and an appendix with computational details.

GitHub repository with replication code and experiment logs is provided. The paper details all hyperparameters, grid settings, thresholds, and decision rules explicitly. However, the workflow involves many tuning constants (loss-scale grids, posterior thresholds, recovery triggers, turnover budgets) that are presented as example specifications rather than universally optimal values. The computational pipeline (SAPG, proximal MALA, FISTA) requires careful implementation. Supplementary Material contains additional diagnostics and sensitivity experiments.