This is a summary of links recently featured on Quantocracy as of Monday, 09/01/2025. To see our most recent links, visit the Quant Mashup. Read on readers!
-
Combinatorial Purged Cross Validation for Optimization [Trading the Breaking]Traditional grid or Bayesian searches conducted on a single path reward parameters that overfit to this specific historical path. This inflates performance metrics through selection bias and temporal leakage. Combinatorial Purged Cross-Validation (CPCV) addresses this flaw by generating a multitude of chronology-respecting train-test partitions. Crucially, it purges any overlapping information,
-
New open-source library: Conditional Gaussian Mixture Models (CGMM) [Sitmo]Ive released a small, lightweight Python library that learns conditional distributions and turns them e.g. into scenarios, fan charts, and risk bands with just a few lines of code. Its built on top of scikit-learn (fits naturally into sklearn-style workflows and tooling). Example usage: In the figure below, a non-parametric model is fit on VIX conditioned on the VIX level, so it naturally
-
The Reversal Tendency of Labor Day Week [Quantifiable Edges]In the subscriber letter over the last several years I have demonstrated that the performance during the week of Labor Day has been impacted by the performance in the month leading up to it. Interestingly, is has been somewhat of a momentum reversal week. When SPX has rallied up to Labor Day, then it has struggled that week. And declines into Labor Day have seen positive performance. Below is an