资产价格泡沫是全球学者一直关注的问题. 对数周期幂律奇异性(log-periodic power-law singularity, LPPLS)模型是预测资产价格泡沫破裂的有效方法之一. 在应用中, LPPLS 模型预测泡沫破裂存在误差. 主要原因是每一个泡沫形成时期都会有大量行政干预以及外部力量干扰, 极少存在``完美''的泡沫, 使得LPPLS模型在某些外部条件发生变化的情况下会发生失灵. 基于此, 研究了LPPLS模型预测有效性的条件, 选取创业板指数和沪深300 指数以探究LPPLS模型的适用范围. 结果表明, 在交易连贯、摩擦成本低, 且投资者结构稳定的条件下, LPPLS模型能够准确预测泡沫破裂的时间.

The research of asset price bubbles has long been a concern of global scholars. At present, the quantitative research on bubble burst in China mainly depends on the log-periodic power-law singularity (LPPLS) model. However, in the real investment market, there will be a lot of administrative intervention and external forces in each bubble frenzy, and there will be no perfect bubbles. Therefore, the LPPLS model will fail when some external conditions change dramatically. In this context, the article selects the GEM index and the CSI 300 index to explore the applicable scope of the LPPLS model. The results show that, under the conditions of coherent transactions, low friction costs, and stable investor structure, the LPPLS model can accurately predict the bubble burst time.