Mean reversion pays, but costs

A mean-reverting financial instrument is optimally traded by buying it when it is sufficiently below the estimated ‘mean level’ and selling it when it is above. In the presence of linear transaction costs, a large amount of value is paid away crossing bid-offers unless one adopts a strategy in the form of a ‘buffer’ through which the price must move before a trade is done. In this article, Richard Martin and Torsten Schöneborn derive the optimal strategy and conclude that for low costs the buffer width is proportional to the cube root of the transaction cost, determining the proportionality constant explicitly

A difficult problem in trading algorithm design is linear transaction costs. This is quite distinct from, and much less analytically tractable than, purely quadratic costs (Garleanu & Pedersen, 2009), and unless very large positions are being traded it is the major source of slippage.

Please click on the link below to read the full version of this article.

Mean reversion pays, but costs