智能投顾能够为个人投资者制定传统资产定价理论中最优化的投资策略，为研究个人投资者的投资行为提供了丰富的研究数据潜力。然而，以往研究智能投顾与机器决策的文献缺乏对投资者用户界面信息与投资行为之间联系的研究。本文以国内某头部第三方智能投顾企业真实交易数据，开创性地研究用户投资行为对用户界面中投顾建议以及产品信息突出效应的反应。首先，一方面1/4以上的有效样本用户初次购买时没有采纳智能投顾推荐的最优产品策略，其中绝大部分选择了风险等级更高的产品策略；本文发现初次购买用户是否采纳投顾建议取决于增加产品风险等级带来的收益增量，而产品下行风险信息的影响并不显著。另一方面，理论上冗余的市场比较基准信息对用户加仓与赎回有显著正向促进作用，并对经验与财富少的用户影响更大；由于智能投顾策略已考虑市场行情信息，这表明投资者对市场行情信息过度反应，做出非理性投资决策。其次，本文识别了不采纳投顾建议行为导致产品策略错配。本文发现不采纳投顾建议用户业绩总体显著劣于采纳的用户；高业绩层次、投资经验多的不采纳用户在短期内表现出比采纳用户更卓越的投资能力，但是业绩优势会逐渐消散；低业绩层次、投资经验少的不采纳用户业绩恶化得更加严重，拉大了投资者之间业绩的不平等。进一步地，通过外生监管事件形成的准自然实验场，本文识别了不采纳投顾建议导致投资业绩变差的因果关系，填补了以往文献对该因果关系识别的空白。最后，本文通过田野实验识别了冗余市场行情信息通过信息突出效应渠道对用户加仓与赎回行为产生影响。在不改变投资者信息集、仅降低冗余市场信息的信息突出效应后，用户加仓与赎回行为受市场行情的影响显著下降，从而投资者对市场行情信息的过度反应显著降低；对于投资经验少、收入水平低的投资者，降低市场行情的信息突出效应对投资行为改善更加显著。从投资业绩来看，处理组业绩层级低、投资经验少的用户业绩得到显著提高，进一步增强了上述结论的稳健性。综上，本文利用智能投顾平台真实交易数据，发现并识别了用户界面信息与投资行为以及业绩表现的联系和因果关系，并通过田野实验发现用户界面信息通过信息突出效应渠道对投资者的交易行为以及业绩有显著影响，填补了以往文献的空白；此外，本文发现通过合意信息展示，可以有效纠正缺乏经验或财富投资者的交易行为、提高其业绩表现，对有效发挥智能投顾金融普惠功能具有政策指导意义。

Robo-advising virtually manages to generate optimal portfolio strategies with respect to traditional asset pricing theory, providing enlightening and fruitful research material on investment behavior of retail investors. However, little in robo-advising literature is known about the nexus between investment information on user interface and investor behavior. This paper, utilizing retail investors’ data from a first-tier third-party robo-advising platform in China, innovatively focuses on how retail investors react to robo-advising recommendations and saliency effect of investment information on user interface.First, it is found that more than one quarter of effectively sampled users diverge from recommended portfolio strategy risk rank on their initial purchases, and most of them invest in more aggressive portfolio strategies than recommended. This paper hence researches on how portfolio information on user interface would impact on investors’ reaction to robo-advising recommendations and on addition to position and redemption afterwards. It is found that while investors’ divergence from recommendation is significantly driven by marginal history revenue from one higher portfolio risk rank, investors do not significantly react to downside risk. This phenomenon also varies between users in different experience and wealth cohorts. On the other side, in analysis of how information of portfolio strategies impacts on users’ addition to position and redemption, it is found that the theoretically redundant market performance significantly positively affects the chance of addition to position and redemption. Since the portfolio strategies from robo-advising has already captured past market information, this phenomenon reveals that investors overreact to market information and make irrational investment decisions.Second, this paper explores the causal relationship between divergence from robo-advising and portfolio strategy mismatch. It is found that overall, those divergent from recommendations have worse investment performance than those following advice. In those that are in top tier of investment performance and are more experienced, investors who diverge from recommendations did perform better than those who following advices in short time period; however, that advantage would fade away in longer estimation window. Among investors in bottom tier of investment performance or with less experience, those investors divergent from recommendations have much worse investment performance than average, which indicates that divergence from recommendation exacerbate the disparity between investment performances. Furthermore, in a pseudo experiment based on exogenous policy event, this paper manages to find out that divergence from robo-advising recommendations indeed causes deterioration in investment performance and hence portfolio mismatch.Finally, this paper carries out a field experiment on the robo-advising platform to discover whether historical market performance affects investors’ addition to position and redemption via saliency effect. Without materially altering the information set of investors, by mitigating the saliency of market performance information, users’ addition to position and redemption are significantly less affected by market and hence investors less overreact to market performance. Investors with less experience or wealth make more rational investment decisions than average if reducing market information saliency; as for investment performance, the less experienced or wealthy enjoy a significant improvement in their gains, which reinforces the robustness of the finding so far.In all, with the help of real account level data from robo-advising, this paper finds the nexus and causal relationship between user interface information and investors’ behavior and investment performance; this paper also manages to identify that user interface information functions significantly via saliency effect. In addition, it is revealed that with proper information display, the irrational investment behavior of individual investors would be efficiently revised, which would illuminate the authority and relevant supervising policy.