Portfolio optimisation and bootstrapping

Abstract

PurposeThe purpose of this paper is to develop a more robust methodology for asset allocation for the property investment market which takes into account inherent valuation and data issues.Design/methodology/approachThe methodology applied is that of a bootstrap, borrowed from Carlstein, and is applied to an investment universe consisting of UK equities, gilts and property. The bootstrap selectively re‐samples the return time series by maintaining the economic cycle. The resulting return series is then used in the standard mean‐variance optimisation (MVO) on an unconstrained basis. Finally, a “sanity” test is applied on the correlation matrix to ensure that spurious instances do not skew the results.FindingsThe bootstrapped optimisation provides a range within which the portfolio weights can be manoeuvred instead of a static point under the standard MVO. It provides a more robust methodology for asset allocation and without giving any undue significance to one year of extreme result.Research limitations/implicationsThe current analysis is based on unconstrained portfolio optimisation, with a very limited investment universe. Additionally, by conforming with the MVO methodology, normality of asset returns is implicitly assumed, which is clearly not the case in the data used. Future work will also focus on an all‐property portfolio.Practical implicationsThe proposed methodology will prove to be useful for making asset allocation decisions, particularly in turbulent financial markets.Originality/valueThe paper focuses solely on bootstrapping with the IPD UK annual index and is particularly significant after one year of extremely poor performance of UK property. The results will be of use to fund managers and portfolio analysts.