Journal of Risk Model Validation

Risk.net

Nonparametric tests for jump detection via false discovery rate control: a Monte Carlo study

Kaiqiao Li, Kan He, Lizhou Nie, Wei Zhu and Pei Fen Kuan

  • Combining multiple nonparametric tests via p-value pooling approach with dependence adjustment can improve the performance of jump detection in high frequency financial data.
  • The reproducibility of the proposed framework was assessed via correspondence curves (local independence) and irreproducible discovery rate (reproducibility across replicates).
  • False discovery rate adjustment is recommended to account for multiple hypothesis testing in jump detection.

Nonparametric tests are popular and efficient methods of detecting jumps in high- frequency financial data. Each method has its own advantages and disadvantages, and their performances may be affected by underlying noise and dynamic structures. To address this, we proposed a robust p-value pooling method that aims to combine the advantages of each method. We focus on model validation within a Monte Carlo framework to assess the reproducibility and false discovery rate (FDR). Reproducible analyses via a correspondence curve and an irreproducible discovery rate were analyzed with replicates to study local dependency and robustness across replicates. Extensive simulation studies of high-frequency trading data at a minute level were carried out, and the operating characteristics of these methods were compared via the FDR control framework. Our proposed method was robust across all scenarios under reproducibility and FDR analysis. Finally, we applied this method to minute-level data from the limit order book system: the efficient reconstruction system (LOBSTER). An R package JumpTest implementing these methods has been made available on the Comprehensive R Archive Network.

To continue reading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: