Quote from rickf:
Granted, while I've traded equities for years but only traded futures intraday for several months, please allow me to pontificate on the subject. After months of research, paper trading, reading, and live trading using different indicators and techniques with the goal of profiting from the futures market (to include looking for that *perfect* or *best* indicator to profit with in any market condition that everybody looks for when starting out in the futures market) the answer is simple:
You are your own best indicator..
I do a lot with intraday futures, but with predictive models.
So my findings are from that perspective as opposed to that of a technical or indicator trader.
Most indicators or simple combinations of popular indicators don't work very well, seldom getting a profit factor above 2.0.
Correctly used in a soft computing predictive model (neural network, GA, etc.) the PF goes considerably higher.
There are several reasons for this:
1) Markets are very complex semi-chaotic phenomena. The idea that all this can be reduced to one number and a few trading rules is wishful thinking or good marketing if you are a seller of trading tools to mass market traders.
2) Most are designed to be simple and easy to calculate. Several reasons here: A) Many were designed years ago when computing power was expensive or if you go back several decades when they were calculated by hand calculators.
B) Simple is easy to sell. If you are a system seller doing a seminar like Wilder did, you don't want to dampen sales with complex math. Averages and division is about all most people can remember from school.
3) They loose their edge as everyone looks at their historical results and decides that a given combination of a few is a good trading system and uses it.
So that said my approach for predictive modeling is this:
1) Along with the indicator give the model the component values used to calculate the indicator. Instead of one number for the RSI, give it the four or five values needed to calculate the RSI.
2) Create new indicators using a level of mathematical complexity
that will never sell to the mass market traders such as Wavelets.
3) Let the model figure out how to combine all of these as opposed to creating manual trading rules like if the MACD > X - Close AND RSI < .80, etc.
This approach works but only if you are an analytics trader using predictive models.
If you are making manual trading decisions the level of complexity required to integrate many dozens of decision points
is beyond human capacity. A neural network has no problem doing it but the human mind can't look at much more than 7 factors in making a decision.
Jerry030
Surely, you'll make money everytime...lolz~!