## Abstract

This chapter investigates the profitability of a strategy based on short selling indices. It is constructed on the principle of the importance of leverage effects, which relates to the fact that increases in volatility are linked with falls in stock prices and vice versa for decreases in volatility. Several new techniques based on the use of forward-looking implied volatilities on two indices, the FTSE100 and the S&P 500 are used to generate short and long trading strategies in the two indices. These strategies employ variants of quantile regression-based techniques, including linear quantile regression, kernel-based quantile regression, and quantile regression random forests, to predict quantile intervals and employ changes in these to generate a trading strategy. Out of these, Kernel-based quantile regression methods appear to generate the greatest returns in the hold-out sample periods and dominate buy and hold returns. Kernel quantile regression is an evolving quantile regression technique in the field of nonlinear quantile regressions. As it is capable of modeling the nonlinear behavior of time series data, it proves to be more efficient in forecasting risk than other methods, including linear quantile regression.

Original language | English |
---|---|

Title of host publication | Handbook of short selling |

Editors | Greg N. Gregoriou |

Place of Publication | Waltham |

Publisher | Elsevier |

Chapter | 33 |

Pages | 479-492 |

Number of pages | 14 |

ISBN (Electronic) | 9780123877253 |

ISBN (Print) | 9780123877246 |

DOIs | |

Publication status | Published - 2012 |

Externally published | Yes |

## Keywords

- implied volatility
- Kernel-based quantile regression
- linear quantile regression
- machine learning-based methods
- ordinary least squares method
- quantile regression random forests