Asset Pricing with Contrastive Adversarial Variational Bayes
Authors: Ruirui Liu, Huichou Huang, Johannes Ruf
IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments show that CAVB not only significantly outperforms prominent models in the existing literature in terms of total and predictive R2s, but also delivers superior Sharpe ratios after transaction costs for both long-only and long-short portfolios. We evaluate the proposed CAVB model using the Open Source Asset Pricing dataset consisting of a variety of monthly firm characteristics. Sections 3.1, 3.2, 3.3, 3.4, and 3.5 are dedicated to 'Dataset', 'Baselines', 'Performance Metrics', 'Statistical Evaluation', and 'Economic Evaluation' respectively, providing empirical results and comparisons. |
| Researcher Affiliation | Collaboration | Ruirui Liu1 , Huichou Huang2,3 and Johannes Ruf4 1King s College London 2City University of Hong Kong 3Bayescien Technologies 4London School of Economics and Political Science |
| Pseudocode | Yes | The overall training process of the factor network is introduced in Algorithm 1 of Online Appendix B. It yields the estimated parameters of the factor network, i.e., ˆθ, ˆϕ, and ˆψ. Then the common latent factors ˆft+1 can be constructed by the estimated encoder. Given the estimated latent factor ˆft+1 and zt, we can train the beta network in the second step using SGD with the loss j=1 ρτj(ri,t+1 βτj(zi,t) ˆft+1) + Lcl where ρτj(u) = |u|(τj1u 0 + (1 τj)1u<0) denotes the check function with τj as the quantile [Koenker and Bassett Jr, 1978] and Lcl is the contrastive loss of (9). Algorithm 2 in Online Appendix C shows the min-batch training process of the beta network. Following Algorithm 1, we obtain the estimated parameters of the beta network ˆψ, and accordingly the estimated quantile-dependent beta functions ˆβτ(zt). ... The algorithm is shown in Algorithm 3 in Online Appendix D. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing source code, a link to a repository, or mention of code in supplementary materials for the methodology described. |
| Open Datasets | No | We evaluate the proposed CAVB model using the Open Source Asset Pricing dataset consisting of a variety of monthly firm characteristics. ... All in all, the dataset consists of 96 different observable firm characteristics and individual returns (obtained from the CRSP database) of stocks traded on NYSE, AMEX, and NASDAQ from January 1980 to December 2022. The paper mentions the 'Open Source Asset Pricing dataset' and 'CRSP database' but does not provide a specific link, DOI, or a formal citation with authors/year for direct access to the exact dataset used. |
| Dataset Splits | Yes | To capture the time-varying market states, we adopt a moving window of 31 years, split by a 20-year training sample, 10-year validation sample, and 1-year test sample, starting from January 1980. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., GPU, CPU models, or memory) used for running the experiments. |
| Software Dependencies | No | The paper discusses various machine learning techniques and architectures (e.g., VAE, GAN, neural networks, SGD, ELU activation) but does not provide specific software library names with version numbers that would be necessary for replication. |
| Experiment Setup | Yes | To capture the time-varying market states, we adopt a moving window of 31 years, split by a 20-year training sample, 10-year validation sample, and 1-year test sample, starting from January 1980. ... We evaluate the out-of-sample performance of all competing asset pricing models by two metrics in the test data. ... We use a standard 60-month rolling window to calculate ˆλt. ... Specifically, we sort the predictive returns ˆrpred i,t+1 by different models and select stocks within the top 10% and bottom 10% of predicted returns. ... All selected stocks are equally weighted in the portfolios, and the transaction cost for portfolio rebalancing is assumed to be 30 basis points (bps). |