Large language models (LLMs) have revolutionized many areas (e.g. natural language processing, software engineering, etc.) by achieving state-of-the-art performance on extensive downstream tasks. Aiming to achieve robust and general artificial intelligence, there has been a surge of interest in investigating the reasoning ability of the LLMs. Whereas the textual and numerical reasoning benchmarks adopted by previous works are rather shallow and simple, it is hard to conclude that the LLMs possess strong reasoning ability by merely achieving positive results on these benchmarks. Recent efforts have demonstrated that the LLMs are poor at solving sequential decision-making problems that require common-sense planning by evaluating their performance on the reinforcement learning benchmarks. In this work, we conduct an in-depth assessment of several state-of-the-art LLMs' reasoning ability based on the inductive logic programming (ILP) benchmark, which is broadly recognized as a representative and challenging measurement for evaluating logic program induction/synthesis systems as it requires inducing strict cause-effect logic to achieve robust deduction on independent and identically distributed (IID) and out-of-distribution (OOD) test samples. Our evaluations illustrate that compared with the neural program induction systems which are much smaller in model size, the state-of-the-art LLMs are much poorer in terms of reasoning ability by achieving much lower performance and generalization using either natural language prompting or truth-value matrix prompting¹.

Formulaic alpha synthesis has emerged as a central problem in quantitative finance, where the goal is to construct predictive signals in the form of formulaic expressions that capture asset return patterns. However, existing approaches typically overlook the non-stationarity of financial markets and fail to embed the market information appropriately, leading to brittle performance under varying market regimes. In this paper, we propose Lovas (Logic-enhanced Variational Alpha Synthesis), a novel neural–symbolic framework for robust stock prediction. Lovas integrates a symbolic regime analyzer with a neural alpha synthesizer: for a specific time step, the symbolic regime analyzer predicts the latent market regime and assigns ensemble weights for the alpha pool, while the alpha synthesizer learns specialized formulaic alphas conditioned on these regimes. Since the market regimes are hidden variables, we therefore formulate alpha synthesis as a variational inference problem and optimize it via a variational EM algorithm. This design enables Lovas to handle market regime shifts more robustly by explicitly conditioning alpha synthesis on different market categories, therefore achieving much better prediction results. Empirical results on multiple stock benchmarks demonstrate that Lovas consistently outperforms state-of-the-art alpha synthesis methods by achieving better prediction accuracy.