Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 2 down vote favorite

I need references on tools for testing the validity of technical analysis indicators on stocks market.

The tools I want could be anything, the technical analysis indicators are a list of values built from some stocks market parameters, such as price and volume - from these newly built indicators, there are some rules "if n happens, sell it, if b happens, buy it etc" these rules suposedly have "earning money" as their objective, is there something that could help me to test the connection between the rules and the objective?

share|improve this question
1  
I think you will need to be a little more specific to turn this into the sort of statistical question that people can help you with. The two questions that would be particularly helpful are: which stock market indicators (or at least type of indicators); and what do you mean by "validity". – Peter Ellis Jun 10 '12 at 10:33
   
Validity: It being true or not? – Gustavo Bandeira Jun 10 '12 at 10:39
The issue being, that something like the Dow Jones index is just an average of some sort of various stock prices, so it is by definition "true" for that particular average of stock prices. What is the underlying concept you want the index to be a valid sign of? What would it mean for the Dow Jones index to be "true" or not? - for example, it doesn't claim to have predictive power. I'm assuming you're not concerned about whether there is deception involved in the various published indices. – Peter Ellis Jun 10 '12 at 10:44
2  
(+1) As reformulated, this is a great question: the world is chock full of "technical indicators" designed to help make investment decisions in markets. Most of them are self-serving BS, based on illusions stemming from overfitting, lack of cross-validation, lack of confirmation, and so on. For instance, how would a statistician go about testing the "head and shoulders" indicator so beloved of "chartists"? (To begin with, it's impossible to find any unambiguous definition of this pattern, a sign that pseudo-statistics is in play.) – whuber♦ Jun 10 '12 at 13:41
1  
@whuber Yes. I started to think: "How can I define such a pattern?" the after exaustive trials, I started to doubt about it. – Gustavo Bandeira Jun 10 '12 at 13:51
show 3 more comments

closed as not a real question by Peter Ellis, onestop, Macro, Andy W, Peter Flom Sep 27 '12 at 0:34

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, see the FAQ.

2 Answers

up vote 1 down vote

Now that the question is a little less vague I think I can guess at what you are driving at and give a suggestion. Test you indicator using regression or correlation with the outcome they are suppose to predict. If they lack correlation then by your terms they are not valid.

share|improve this answer
2  
Sorry for the vague question, I naively/unconsciously presumed that you knew about the technical analysis indicators. – Gustavo Bandeira Jun 10 '12 at 13:31
up vote 0 down vote

Why not backtest them against a benchmark (e.g. for US stocks, how did it compare to the S&P500 in terms of risk/return)? If the indicator requires past data like moving averages, or has numeric terms, just make sure you're only calculating on previous data using a rolling window function to avoid 'data snooping'.

Simplest way is to use a software package like R (or excel for simpler things) and have a dummy variable in your data that equals 1 for a long position, 0 for no position and -1 for a short position. Then you can run calculations like sum, mean and standard deviation on the log returns of the actual asset multiplied by this dummy variable; and compare it against your benchmark. If it 'beats the market', its worth investigating further while paying close attention to transaction costs/market depth (if there is very little liquidity at the historical price point, you would likely have gotten a much worse price as your trade would have affected the market price adversely).

share|improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.