The most well-known type of forecast tool uses the same single assumption for the returns on assets. This is known as deterministic forecasting and is highly likely to be misleading.
Even using the best estimate it will give results, which half the time will be too high and half of the time too low. With only one figure it gives no understanding of the risk taken and gives an unrealistic feeling of accuracy.
A solution to all these problems is to show a range of possible out-comes for an investment instead of a single figure, using stochastic mod-elling. This solution demonstrates the risk and reward of assets, as well as not emphasising a single result.
The two main types of stochastic model used in the market today are mean/variance/covariance asset models (MVC), which have a number of limitations, and economic sce-nario generators (ESG).
MVC models take no account of how long an investment is held for, as they assume the same average return and volatility irrespective of invest-ment term. In reality risk and return characteristics of assets vary by the term held, just as a car journey for a short distance will tend to have a lower average speed than for a longer journey. An MVC model would assume a single average speed no matter the length of the journey.
In contrast, an ESG model is able to ref lect reality in its outputs for both long and short-term fore-casts. The way such models can be designed means that fundamental real life characteristics of assets can be reproduced, giving sensible and realistic forecasts.
An ESG stochastic model works by building many possible economic scenarios from first principles and looks at the range of returns that are given as a result. As an example, let us imagine this method was being used to predict the time of a car journey. We would set up a model to measure the speed achieved by a large number of cars to reach various destinations in a model road network. The range of speeds at different times of day and routes would be measured to provide the information to work out the likely journey time.
The model would allow for random events such as accidents, road works, congestion levels, etc. These events are also likely to affect surrounding roads and could be inf luenced by other factors such as time of day, weather or holidays.
Between any two destinations there will be choices between a number of different routes – the shortest route perhaps along B roads or a longer faster journey mainly along a motorway. Which is likely to be quicker? Is one more certain but slower? The model could pre-dict a range of times for each option and a probability of one route being quicker than another depending on factors such as the length of journey or time of day.
Regular maintenance of the model is required to take account of road improvements, new more technically advanced cars and other major structural changes to jour-neys e.g. the introduction of new speed limits or road tolls.
A good ESG stochastic model can be complicated because it needs to take account of numerous fac-tors which can affect one another just as an incident on the roads can have knock-on effects to other jour-neys. What a good model can do is tell you what route is likely to be quickest depending on the time of day, the chances of a delay and the potential time lost if there is a delay. Depending on the importance of arriving on time, the driver might choose a slower less congested route with a higher probability of arriving precisely on time.
Investors deciding on how much risk to take with their investments are making precisely the same sort of decisions when deciding on the chances of achieving their objec-tives and how big a shortfall they would be prepared to accept if things don’t do well.
