Given 42 managers with an average return over theperiod of 37.59% and a standard deviation of 4.90%,normal probability theory tells us that 95% of all ofthe returns should lie within the range of about 29%- 46% which in fact they do.Equally it implies that only 5% of the funds or 2 ofthem should lie outside this range. In fact there are 3managers who did better and one who did worsethan this. But since at least one of the 3 managerswho did better could have been there by chancealone how do we decide which was lucky and whichwas skillful?A complicated t-test of the null hypothesis mightwork but would also almost certainly bore theaudience to death. Obviously if we had more dataone clue might be to see which of the 3 managerssuspected of being skillful also did well in priorperiods on the reasonable assumption that if hedoes well consistently then s/he must be good.Unfortunately here again the evidence isinconclusive. Such ‘persistence’ of managerperformance rankings has been both confirmed andrefuted by various studies. (Notable work in this fieldhas been done locally by Cadiz). However, mostsuch studies suffer from survivorship bias and theneed for vast amounts of data. In SA fewer than 10General Equity funds have been in continuousoperation for 10 years or more.As a result most studies are for 5 year periods whereI believe they may be biased by the effects of theunderlying business cycle which typically also runsfor 4 – 5 years. Any study of only this length maytherefore be dominated by the value and growthcycle inherent in the business cycle itself. This hasmost recently been seen in the strong localpreference for the value style of investing.Whilst I believe that value may be the superior styleof investing (at least from a long-term and riskperspective) I do not believe it is our naturalpreference from a behavioral point of view.Economic theory holds that investors are preparedto pay a premium for growth albeit at higher risk.Thus when risk aversion is low and economies pickup there is a natural preference for the growth styleof investing. We are just entering such a phase nowand I fully expect to see a spate of new listingswithin the next 18 months typifying such a growthcycle.The exhaustive studies of Fama and French haveconvincingly demonstrated that excess performancecan be explained by the following persistent effectspresent across markets. They are:1) Sector preferences2) Size and liquidity effects3) Style bias4) MomentumOf these the Size and Style factors appear to be themost consistent. This has given rise to a large industryof consultants and style analysts encompassing suchfirms as BARRA and APT’s style factor models as wellas the consulting industry’s benchmarking initiatives.While all of these initiatives help in attributingperformance correctly and in identifying risk factors,they once again do not entirely separate skill from luckalthough persistently good selection and timing can betaken as evidence of the presence of skill. Thus theydo give us a better idea of where to look.One method which may provide a possible answer isRon Surz’s,
Portfolio Opportunity Distribution(POD)
methodology.Extending the monkey analogy to that of 1000’smonkeys typing eventually producing the completeworks of William Shakespeare, the POD method usesthe brute force of Monte Carlo simulation to randomlygenerate thousands of possible portfolios from theopportunity set given and to determine the range ofoutcomes which were possible. By comparing theactual performance of the manager to this referencerange we can see exactly how well the manager madeuse of the opportunity set available to him or her. Realworld constraints such as position size limits can easilybe included.
“To be or not to bzw;afuklyqkulWE bebee”