Past performance is not always indicative of future results
A hush descended over the expectant crowd, gathered around Augusta’s 18th green, as Sergio Garcia addressed his ball.
Seventy-three times before Sergio had tried and failed to secure a major golf championship win. Was this to be his day?
Many hoped but few expected that he could shrug off past demons and putt his way to glory – but that’s exactly what he did to earn himself a world famous green jacket. Even Sergio’s most ardent supporters may have been hard pressed to predict this outcome.
However, as Sergio’s success has shown us, relying on past performance, whether good or bad, is not exactly the best method of forecasting the outcome of a golfing championship. The same is also true when it comes to forecasting the level of returns on our investments.
Stochastic forecasting is a mathematical method used to predict outcomes that involve a certain degree of randomness or unpredictability. Not only does it help to determine which outcomes are expected to happen but can also show those which are less likely to occur. Such models are used for a wide range of purposes but are particularly effective at illustrating the trade-off between investment risk and reward. The idea is that assumptions are made about a variety of different inputs from which a wide range of possible future outcomes are produced.
Golf performance, for example, can be affected by numerous variables – what the weather is like, the loft of the club being used, the type and weight of the ball, the nature of the golf course and the underlying ground conditions, to name but a few. If all of the starting conditions were completely known then an accurate outcome of how a particular golfer will perform could, in theory, be produced. Unfortunately, it is impossible to know the exact starting conditions and so assumptions have to be made in order to generate a possible forecast of future performance.
However, not all kinds of stochastic models give a realistic view of future outcomes. The two main types of models used widely today are mean / variance / covariance asset models (MVC), and economic scenario generators (ESG), such as that provided by EValue.
MVC models are based on the assumption that past performance and variability can be used to predict the future. The quality of an MVC model is, therefore, significantly influenced by the time period over which performance is measured.
However, choosing a period that is too long or too short will lead to flawed and unrealistic results. Too short a period may not give a true indication of performance over the long term and choosing too long a period may include certain factors which are no longer relevant. Take Rory McIlroy, for example. In 2014, Rory won both the Open championship and the PGA championship holding the number one player’s ranking spot for 54 weeks, from 3rd August 2014 to 15th August 2015. Despite this, a spate of injuries throughout 2015 and a change of clubs in late 2016 have resulted in a decline in his notable golfing achievements. Using an MVC model to predict his future performance based solely on his 2014 and early 2015 results would not have given an accurate forecast of what was to come.
In contrast, EValue’s ESG model does not depend on a specific period of historic performance. Instead, the model has been designed to reproduce the fundamental real life characteristics of investments thereby enabling more sensible and realistic forecasts to be produced in a way which is not dependent on a specific period of historic investment data.
As with any sport, advances in new technologies and continuous improvement have helped to revolutionise the game of golf. Fully adjustable clubs that help players hit the ball further and straighter, golf balls with better flight that can gain more distance, better fitness regimes and access to sports psychologists have all had a significant impact on how the professional game is played.
Stochastic models also need to evolve in order to continue to be successful and remain relevant. Changing economic circumstances, greater availability of data, technical developments and improvements in computational efficiency will all lead to better stochastic models and even more accurate forecasting, including predicting whether or not you will need your golf brolly at this week’s BMW PGA championship!