In statistical arbitrage, pairs trading is usually considered a risk-neutral strategy. However, the methodologies in existing literature on choosing thresholds and calibrating cointegration coefficients could be arbitrary and insensitive to market changes. This research discovers that static parameterization in pairs trading could result in undesirable systematic risk and potential losses. We apply Kalman Filter to intertemporally estimate cointegration coefficients and the absolute standardized residual (ASR) threshold, and relate the ADF-threshold with stochastic discount factors. Backtests confirm our superiority in attaining better risk-to-reward ratios and lower systematic risk.

在统计套利中，成对交易通常被视为风险中性策略。但是，现有文献中有关选择阈值和校准协整系数的方法可能是任意的，并且对市场变化不敏感。这项研究发现，成对交易中的静态参数化可能会导致不良的系统风险和潜在损失。我们应用卡尔曼滤波器来跨时估计协整系数和绝对标准化残差（ASR）阈值，并将ADF阈值与随机折现因子相关联。回测证实了我们在获得更高的风险回报率和更低的系统风险方面的优势。