xlitang, Interesting system, not sure how the promise indicator works (guess that's the point), but the system does look good. Here is my demo: https://www.myfxbook.com/portfolio/mt4-824319/758288 Anpar
Follow me to the Promised Land
- | Commercial Member | Joined Aug 2011 | 1,121 Posts
{quote} Answered already in post 48 here: http://www.forexfactory.com/showthre...19#post7093419 Oops! Didn't see xlitang's answer above.
I appreciated your voucher. Thanks.
Follow me to the Promised Land
- | Joined Feb 2012 | Status: Grand Ultimate | 825 Posts
{quote} Hey Man, I appreciated your voucher. Thanks.
I discovered a while back the same thing that you seem to have discovered. Martingale is only appropriate when an action (reversal) can be assured with 90%+ accuracy.
Price, as you said, always reverses (at some point! ;^) The objective, then, would be to get close enough to the reversal point so that the number of layers isn't too great, and to have a way to limit loses if the entry point is too early. Not exactly a mechanical thing, unfortunately!
When I worked at a prop shop, traders used a pairs trading strategy (statistical arbitrage) & put on "layers" when price moved against them. Basically, it was like martingaling but without doubling the bet.
All layers were the same size as the original. Only three layers were used except in special circumstances when 4 was allowed. Once price reversed, they took off each layer as price reached BE for that trade, eventually leaving the original trade for making a profit. If it didn't reverse after 3-4 layers, they took the SL.
They were careful to adjust the trade size according to the probability of a successful trade. Higher probabiity, increase the bet size, lower probability, decrease the bet size. The current president of that shop, a former $1 million/yr trader, insists that it is the only way to be consistently profitable.
There are many variations of martingaling, as you probably know, including using a different multiples than doubling. For example, instead of doubling the bet, increasing it by 1.2 times the original.
A programming system, such as Zorro or Neuroshell, might be able to optimize the multiplication factor over past data, but as you point out, backtesting has limited application to future trading.
Good luck with future developments.
First you must know the rules, then you must know when you can break the rules. The answers often lie under our biases.
The mind thinks, the heart knows...