Do Technical Indicators apply to all stocks?

Many people have asked me how to find out which indicator to use on which stock and why don't some stocks follow the basics of technical analysis while others follow it so neatly.

Well, the answer is hidden in basics of how technical anaysis have been founded and the fundamentals of prabibility and statistics. Often, many traders forget these fundamentals that define the very basics of any forecasting methodology including stock trading.

In this article, I've tried to show, one of the way to filter out stocks that will most probably follow the basics of technical analysis. The article is fairly mathematical in nature and I expect that you already know basic statistics concepts like Std. Deviation, mean, averages and technicals used in trading like Fibonacci, etc.

So, in this article, I'm focusing on "3 Month Avg. Volume" as our filter. The reason is that more is the volume of shares of a particular stock, higher will be the probability that the stocks is traded by more number of people, and hence it's not biased by only a few set of people. Technical indicators are often thrashed and bruised when only a few set of people are manipulating the stock price. So the whole idea is to find only those stocks that have enough number of shares in the market. But the questions is "how enough is enough?".

To find this, I analyzed all the 8000+ stocks traded in the US markets and can be found at http://finance.yahoo.com.

The first thing to find was the "distribution" of the 3 month avg volume of the stocks so that I can find out what's the type of statistical tool I should use to further analyze the stocks.

And here are the results:
Std Dev = 3,474,398.44
Avg = 905,364.23
Mean = 57,043,505.88

From this it is clear that the stocks are spread in a very large band and it's very difficult to prove that most of the stocks are in/around the avg (because avg and mean have a lot of difference).

To prove this, I plotted a histogram of the stocks and here are the results:
11 to 5,704,361 = 7981
5,746,440 to 11,450,789 = 112
11,542,300 to 17,246,649 = 44
18,230,600 to 23,934,949 = 14
24,889,500 to 30,593,849 = 8
34,054,600 to 39,758,949 = 10
67,585,400 to 73,289,749 = 4

This gives a hint that we should be focusing on the below 5M range and dig deeper into that range to find out that "cut off" of the 3 month avg. volume.

So, I started plotting the percentile chart to find out how many stocks have above n 3m avg. vol.
Here are the results:
90% ptl = 9,510
80% ptl = 24,160
70% ptl = 54,002
60% ptl = 106,053
50% ptl = 187,646
40% ptl = 310,041
30% ptl = 499,732
20% ptl = 845,622
10% ptl = 1,863,100

From this we see that 90% of the stocks have 3M avg. volume more than 9,510, 80% have more than 24,160, and so on.

This analysis along with the histogram clearly shows that if we are concentrating on stocks above 1.8M range, we are really focusing on a very small number of stocks (only 10%) and if we are focusing on above 54K range, then we are looking at 70% of the market.

I meshed this analysis with the Fibonacci magic numbers in "two dimensions". First, I used the 38.2% Fibonacci ratio and hence used the 40% percentile, which suggests that we should be focusing on stocks above 310,041 range.

Then I padded it with a risk factor, adding the second dimension, to say that instead of using the 40%, I'll use the 30% percentile, because for one, it's safer than the 40% range anyways, and secondly, the 30% number is 499,732, which is incidently the 50% Fibonacci ratio of the avg. (1M).

==================================================================
Hence, for US stocks, there are better chances of your technical analysis to work properly on stocks that have 3 month avg. volume more than 500K
==================================================================

Hope this analysis helped you! Feel free to drop in your comments!

cheers and Happy Trading!
GT