Hi guys,
Can someone shed me some light on the this section in the Quants? I don’t understand what is the purpose of it. From my understanding, this section simply says that Moving Average cannot be used for forecasting, because it provides the same results as the moving average calculations with random variables. The Example 14 states exactly that. I don’t see any relation between this section and real forecasting.
Thanks!
Anyone, please?
Financial time-series follow either an AR or an MA process, and can be modelled by an AR or an MA model accordingly. The order of lag of an AR or an MA is determined by examining autocorrelations.
For an MA(q) model, the first q autocorrelations will be statistically significant (the null hypothesis that there is no autocorrelation is rejected), and all autocorrelations beyond q will be statistically insignificant (the null hypothesis is not rejected). If an MA(1), the first autocorrelations will be statistically significant, and all correlations >1 will be zero. On the other hand, autocorrelations for an AR start large and decline gradually.
Example 14 refers to a situation were all autocorrelations presented are all equal to zero (null is not rejected) – even the first one, so that it is why it is an MA(0) and no lags are included in equation 14. Here therefore we established that S&P 500 follow an MA(0) process which is therefore not predictable – they cannot be predicted as you rightly said. However, if it was an MA(0
However, if it was an MA(q) process you use an MA(q) model to forecast.
However, if it was an MA(q) process you use an MA(q) model to forecast.
Cool, thanks. But I still don’t see any application for this.
Ummm, you’d use this for [short-term] forecasting. I’m not sure how many people actually do it in the industry unless you’re in the risk/analytics field. As with everything else, this thing is heavily dependent on inputs, so the forecast is only as good as the inputs and the data.
But I can see how this theory can complement/aid your fundmental analysis.