Reproduce “Sparse Autoencoders as Implicit Diversification Regularisers for Cross-Sectional Cryptocurrency Returns” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Weijia Han, Lisha Qu

The authors use a sparse autoencoder (SAE) to encode daily cross-sectional cryptocurrency returns, reconstruct them, and rank assets for equal-weighted long–short portfolios. Under pre-test selection, the selected K=128 SAE produced 20–26% lower annualized long–short volatility than matched-dimension PCA in each of five confirmation folds, averaging 23%. This is rank-scale realized-risk evidence, not alpha or an executable strategy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.