Optimal multifactor trading under proportional transaction costs

TL;DR

This paper develops an optimal trading strategy under proportional transaction costs for assets with stochastic risk factors, deriving explicit formulas for buffer zones and position displacement.

Contribution

It provides a general solution for multi-factor diffusive models, revealing how transaction costs influence trading buffer widths and position adjustments.

Findings

Buffer width scales with the cube root of transaction costs.

Faster trading strategies are buffered more due to volatility effects.

Displacement of the target position is proportional to the 2/3 power of transaction costs.

Abstract

Proportional transaction costs present difficult theoretical problems in trading algorithm design, on account of their lack of analytical tractability. The author derives a solution of DT-NT-DT form for an arbitrary model in which the the traded asset has diffusive dynamics described by one or more stochastic risk factors. The width of the NT zone is found to be, as expected, proportional to the cube root of the transaction cost. It is also proportional to the 2/3 power of the volatility of the target position, thereby causing a faster trading strategy to be buffered more than a slower one. The displacement of the middle of the buffer from the costfree position is found to be proportional to the square of the width, and hence to the 2/3 power of the transaction cost; the proportionality constant depends on the expected short-term change in position.