Discussion in 'Technical Analysis' started by robinxing, Apr 2, 2008.
how to measure it with statistical tools ?
any books talking about it ?
The lack of ways of measuring an indicator's efficiency is one of the leading defects of current software. About all the software allows you to do is build a system, and that's jumping way ahead of what you want to know about the indicator. There ARE ways to standardize the size of the moves "predicted" by the indicator. You could use Std Deviations or AvgRange. I have done it, and I did find the system I sought. But for starters, your best bet is to go thru miles of history and mark S or s (big success, little success) or F in a paper tally for each signal the indicator gave. This will tell you more than any math formula. I will save you some trouble and tell you that most published indicators are worth less than you paid for them. But reliable patterns DO exist.
Backtest it for yourself and see how well it works out.
You need to put it in context, an indicator by itself means nothing. Work out how you are going to use it, that is put it in a system, and then backtest it. Sharpe, sorinto etc will give you a good understanding of whether you ideas are valid.
thank you linda
do you know any books talking about trading patterns?
I think it would be preferable to ask, How do I define the "efficiency" of one indicator?
Once you've defined "efficiency" in a quantitative way, measuring it is a small matter of programming and historical testing.
spreadhseet a large sample of historical prices. Put in your indicator. If it outperforms, your averages/medians should tell you what you need to know.
thank you MGJ
then how to test it with a quantitative way?
how do you define "efficiency" ?
I want to design a indicator to calculate the support and resistance level similar to CMO
I hope the incator is more sensitive than CMO
I do not know how to improve it
Separate names with a comma.