Statistical likelihood methods in finance
Abstract
It is known from previous work of the authors that non-negative arbitrage free price processes in finance can be described in terms of filtered likelihood processes of statistical experiments and vice versa. The present paper summarizes and outlines some similarities between finance and the statistical likelihood theory of Le Cam. Options are linked to statistical tests of the underlying experiments. In particular, some price formulas for options are expressed by the power of related tests. In special cases the dynamics of power functions for filtered likelihood processes can be used to establish trading strategies which lead to formulas for the Greeks Delta and Gamma. Moreover statistical arguments are then used to establish a discrete approximation of continuous time trading strategies. It is explained that Ito type financial models correspond to hazard based survival models in statistics. Also price processes given by a geometric fractional Brownian motion have a statistical counterpart in terms of the likelihood theory of Gaussian statistical experiments.
Cite
@article{arxiv.1310.4400,
title = {Statistical likelihood methods in finance},
author = {Arnold Janssen and Martin Tietje},
journal= {arXiv preprint arXiv:1310.4400},
year = {2014}
}
Comments
16 pages, corrected typos