Hello everyone,
I’ve been learning to program in mql4 and enjoy building robots and backtesting them. I do it because I enjoy it, not because I expect to earn tons of money with it, but if I could then that would be a nice bonus.
Anyway, I tried googling, but couldn’t find much information, so I figured I’d start this post. What I’m looking to create is a list of characteristics and/or guidelines of auto trading robots. I’m a newb, so to be honest I don’t even know if it is possible to generalize these things to a list…
I’ve only figured out a few things so far:
- Keep the number of indicators used to a minimum
- Use leading indicators instead of lagging (do you agree?)
- Ride any trend as long as possible instead of focusing on fixed take profits
- Cut losing trades as quickly as possible -> i.e. monitor open trades
I however am still looking for answers to a lot of questions:
A. Do auto traders work mainly with short time frames? Which?
B. Are the number of trades limited to 1 at a time? Or how many are open at once?
C. How many currencies do auto traders work on at once?
D. How many signals should an auto trader produce per day/week/year?
It be awesome if you guys could share your insight about how successful forex robots work and contribute to this list.
Thanks!
First of all, you need to understand the basics and some advanced concepts of trading to be able to build an auto trader which makes sense. You should probably spend 10 or 20 times as much time trading as coding. Once you have enough practice trading, most of your questions will be answered. To help the learning process, you should follow other traders’ threads on this and other forums. You will learn a tremendous amount about trading and get the benefit of others’ insight. What you need to look for is a mechanical trading system that can be automated. Discretionary systems (the majority) can never be made into an auto trader although you might be able to automate the trade and risk management parts. Do not take the “rules” of a trader’s system too seriously. I think you will see that many of them are broken when the trader doesn’t feel the market or something else isn’t quite right.
As for your questions, I can see that you seem to believe that complexity and sophistication is the goal. There is an understandable tendency to believe that if you add one more feature or filter, some of the losers will become winners. Instead what happens is some of the winning trades are either rejected or turn into losers. I have been there and in the end found that complicated auto traders work as badly as simple ones. The benefit of simple auto traders is that you spend less time before abandoning them.
This is a bit of a contradiction to CodeMeister, but it is from my experience.
Building and testing robots helps you to trade. You can see which bots work and why they work. You are continually analyzing charts and robots sort out systems that work to ones that dont work.
I will try to answer your questions.
-
Robots are varied and come in many different types. Using your imagination helps you to try different systems. So the timeframe differs from bot to bot. Some bots need regular data - 1 min, some look at long term trends H4 or daily. Generally for Moving Average based robots, The higher the timeframes are better H1, H4.
-
You can open as many trades at once as you want. I generally refer to a single trade bot as a scalper.
-
My best bots work with multi-currencies. Using more then 1 currency opens up more options for you. Hedging etc.
-
How many signals? Really depends on the system. 1 per minute - 1 per month
Also on the subject of Lagging indicators. I have a very good bot that works with the standard mt4 zig zag indicator. It repaints all the time, but at least it is honest and tells you that it isnt sure what the market is going to do in the future.
One thing most ea’s lack is the ability to deal with the market when it becomes a monster and dosent act nicely. You can build into your code procedures to deal with this. But it all depends on the type of system as to how you would do this. eg. look at your free margin as an indicator. or change take profit and stoploss based on drawdown or use other ea’s to compensate or use hedging or use percentage takeprofit,stoploss etc.
As I said, use your imagination and try everything. You will have 100 failures for every win, but it is worth it.
Also very important - start demo testing. If it works on a demo, it is likely to work for real.
Also you want to see how it fails. It is more important then how it wins.