David D

整理并合并美国公司债券交易的TRACE数据集，以便仔细研究市场做市商的价格歧视行为。根据Bessembinder的2018年的研究，制定一个新的流动性指标，即“半差价”，它将不对称信息和流动性的方面分开。提取流动性成分以分类债券流动性，并使用支持向量机（SVM）模型进行债券价格预测。增强SVM模型以提高预测精度，从而在要约（Ask）方面改善了债券预测误差，并将平均绝对误差（MAE）减少了20%

Revise the TRACE dataset of US corporate bond transactions, ensuring it is clean and consolidated, in order to scrutinize the price discrimination behavior of market makers. In light of Bessembinder's 2018 research, formulate a new liquidity indicator, the "Half Spread," which isolates the aspects of asymmetric information and liquidity. Extract the liquidity component for classifying bond liquidity and utilize a Support Vector Machine (SVM) model for bond price prediction. Enhance the SVM model to boost forecasting precision, resulting in a substantial improvement in the bond forecasting error on the Ask side, as well as a 20% reduction in Mean Absolute Error (MAE).