In the previous post we look at two issues a) What are Indicators? b) How do we go about finding Indicators that would aid in our trading process or strategy?
In this part we will look at the question - Do Indicators have any predictive value?
Since we discussed that indicators are essentially representative of past, lets re-frame the question. Can past data predict future?
In other words, if I tell you that in 2016 - Monsoon started in June and in 2017 it did in May. That is past data. Will the mere knowledge of past data help you predict the future? Will knowing the above 2 data points tell you anything about the chances of Monsoon this year?. The answer is obviously No.
However, what we can do is to analyze past data of adequate number of occurrences and then calculate the probability of an event occurring in the future. Remember we are only calculating the "chances" that X would happen given that it has happened Y number of times in the past.
But it still does not mean, X will happen for sure.
So if you can create an Hypothesis based on the Indicators or without them even, you can test it for its validity.
So to sum it up, Indicators by themselves are not predictive in nature, however, you can back test the Setups or Hypothesis-es to know how often a particular Cross-over or a breakout leads to a specific outcome that you are looking for - So it's the back-test which determines the probability of an event happening not the Indicator itself.
Of course there could be different indicators with varying degrees of accuracy for a given hypothesis. Our attempt should be to test them all and find what works best for us.