In the stock market, a successful investment requires a good balance between profits and risks. Recently, stock recommendation has been widely studied in quantitative investment to select stocks with higher return ratios for investors. Despite the success in making profits, most existing recommendation approaches are still weak in risk control, which may lead to intolerable paper losses in practical stock investing. To effectively reduce risks, we draw inspiration from adversarial perturbations and propose a novel Split Variational Adversarial Training (SVAT) framework for risk-aware stock recommendation. Essentially, SVAT encourages the model to be sensitive to adversarial perturbations of risky stock examples and enhances the model's risk awareness by learning from perturbations. To generate representative adversarial examples as risk indicators, we devise a variational perturbation generator to model diverse risk factors. Particularly, the variational architecture enables our method to provide a rough risk quantification for investors, showing an additional advantage of interpretability. Experiments on three real-world stock market datasets show that SVAT effectively reduces the volatility of the stock recommendation model and outperforms state-of-the-art baseline methods by more than 30% in terms of risk-adjusted profits.

翻译：在股票市场中，成功投资需要实现收益与风险的良好平衡。近年来，股票推荐在量化投资领域被广泛研究，旨在为投资者筛选具有更高回报率的股票。尽管现有的大多数推荐方法在盈利方面取得了成功，但其风险控制能力仍然薄弱，在实际股票投资中可能导致难以承受的账面损失。为有效降低风险，我们从对抗性扰动中汲取灵感，提出了一种新颖的分裂变分对抗训练（SVAT）框架，用于风险感知股票推荐。本质上，SVAT促使模型对高风险股票样本的对抗性扰动保持敏感，并通过从扰动中学习来增强模型的风险意识。为生成具有代表性的对抗样本作为风险指标，我们设计了一个变分扰动生成器来建模多样化的风险因素。值得注意的是，变分架构使我们的方法能够为投资者提供粗略的风险量化，展现出额外的可解释性优势。在三个真实股票市场数据集上的实验表明，SVAT有效降低了股票推荐模型的波动性，并在风险调整收益方面比最先进的基线方法提升了30%以上。