I scanned that article and would simply say that if there is a positive conclusion it is extremely subtle. The article, like all these articles, is convoluted and idiosyncratic and in fact the authors recognize that their result is at odds with established conclusions of the majority of studies.
"
many earlier studies concluded that technical analysis is useless"
The two very simple techniques they studied (moving average cross-over and support/resistance ) are the easiest techniques and the most obvious. These techniques are just barely included in the term 'technical analysis' and people usually mean much more complex rules by that term.
In fact when I look at a chart the first thing I recognize are support and resistance levels. I can't help myself. When I first looked into TA I was very distressed to find that most (not all) academic studies dismissed support/resistance along with the rest of the TA claptrap. I searched and searched for a published justification for my use of support and resistance but never found one. You may like this technique or that but you can't necessarily prove it works.
I think there is a rational reason for that.
When I look at a chart I have already drawn at least 75% of my conclusion from information that cannot be included in the statistical study e.g. :The industry and what the status of that industry is in the current market; The status of the stock: earnings reports and expectations of earnings. etc.
i.e. when I look at a chart I am looking to confirm my bias.
These issues are not included in the studies that have been done on technical analysis.
As for the data set they worked on it is certainly large. Probably too large. (1897 to 1986) and 30 years out of date.
Data from before WWI/WWII is hardly representative of the current economy or the current market. Using such a large data set with data stuffed in there from before WWI makes the study unusable...at least to me.
?
