Question

Can I model past shocks in order to forecast future values using seasonal variation and trends?

Answer 1

Yes, you can and it is a pretty common analysis tool used in the industry. But the past events that are being used have to be cyclical in nature for the model to produce meaningful results. For example, if you are trying to model in seasonality of sales in certain industries, that would work really well. But as you move towards more unpredictable events such business slowdowns or recessions, the model will tend to become less effective. This is because even though business booms and busts are cyclical in nature, accurately predicting what phase of the cycle is currently going on is pretty difficult. Also, when creating such a model, always make sure that you focus on the industry rather than one company. That tends to vindicate the presence of seasonality by removing excess noise in data.

Additionally, you might want to research a few mathematical models such as the monte carlo simulation. This method uses previous stock prices and the concept of random movement to predict future prices. It assumes the stock price to be composed of 2 components namely the drift (the part that reflects historical trend) and a random component (that takes into account the uncertainty inherent in markets). It come in quite handy when building mathematical models to predict future events based on past data.