An early indicator for anomalous stock market performance
提出一个实时检测股市异常估值的指标，通过非参数趋势识别和SETAR模型预测短期异常，对1970-2022年标普500数据测试，能提前一年识别80%以上的异常事件。
We propose an indicator for detecting anomalous stock market valuation in real time such that market participants receive timely signals so as to be able to take stabilizing action. Unlike existing approaches, our anomaly indicator introduces three methodological novelties. First, we use an endogenous, purely data-driven, nonparametric trend identification method to separate long-term market movements from more short-term ones. Second, we apply SETAR models that allow for asymmetric expansions and contractions around the long-term trend and find systematic stock price cycles. Third, we implement these findings in our indicator and conduct real-time market forecasts, which have so far been neglected in the literature. Applications of our indicator using monthly S&P 500 stock data from 1970 to the end of 2022 show that short-term anomalous market movements can be identified in real time up to one year ahead. We predict all major anomalies, including the 1987 Bubble and the initial phase of the Financial Crisis that began in 2007. In total, our anomaly indicator identifies more than 80% of all – even minor – anomalous episodes. Thus, smoothing market exaggerations through early signaling seems possible.