Scholarlink

<p>This study examines the efficacy of various range-based estimators in the model-free form, notably the Parkinson (PK), Garman–Klass (GK), Rogers–Satchell (RS), and Yang–Zhang (YZ) estimators, in predicting realized variance across G7 stock market indices. Empirical analysis reveals that the Garman–Klass estimator consistently outperforms others in robustness and accuracy, confirming its reliability across diverse market conditions. While the Yang–Zhang estimator shows strengths in specific metrics and markets, it is generally the worst-performing estimator and outperformed by simpler approaches such as the GK and PK estimators. The Rogers–Satchell and Yang–Zhang estimators, despite their complexity, do not provide significant predictive improvements. This highlights the potential of simpler range-based estimators in improving predictions. Overall, these findings emphasize the importance of considering market-specific dynamics when selecting among estimators, suggesting that no single model is universally superior; however, the choice of estimator should be well-established to the unique characteristics of each market.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Is complexity always better? A model-free assessment of range-based volatility estimators

  • 期刊论文

  • 发表日期：

    2026年2月20日

查看全文
  • Burak Korkusuz

摘要

This study examines the efficacy of various range-based estimators in the model-free form, notably the Parkinson (PK), Garman–Klass (GK), Rogers–Satchell (RS), and Yang–Zhang (YZ) estimators, in predicting realized variance across G7 stock market indices. Empirical analysis reveals that the Garman–Klass estimator consistently outperforms others in robustness and accuracy, confirming its reliability across diverse market conditions. While the Yang–Zhang estimator shows strengths in specific metrics and markets, it is generally the worst-performing estimator and outperformed by simpler approaches such as the GK and PK estimators. The Rogers–Satchell and Yang–Zhang estimators, despite their complexity, do not provide significant predictive improvements. This highlights the potential of simpler range-based estimators in improving predictions. Overall, these findings emphasize the importance of considering market-specific dynamics when selecting among estimators, suggesting that no single model is universally superior; however, the choice of estimator should be well-established to the unique characteristics of each market.