Self Assessing EA

Hello, i am about to write my first EA and i am currently planning it out. I am new to mql4 but have a good programming background, so much of my examples will simply be my thought processes with which i will base the actual EA. I do this so that experienced programmers can point out where i do not have the right/ logical ideas.

I plan on making a 5 min scalping system that works at a 1:1 RR ratio. The goal here is to trade more profitably than negatively, for 10 pips each side, in a liquid market.

The system will be based off a combination of the indications of 2 strong 5 minute indicators,a strong swing indicator, and a candlestick indicator. when these indicators are positive and in the same direction i enter a trade.

The system will also be self assessing. To achieve this the system will record the number of take profits vs. the number of stop losses and plot a graph of:

^y - axis
|
| take profits
|
|------------------------>time
|
| stop loss
|

graph A


the assessments will be based off of the ratio of take profits : losses, and will determine whether or not the program continues live trading. if the ratio begins to favour losses over a certain period of time(I'm thinking using 15 minute periods and assessing over last 5 periods) it stops trading and enters 'demo' mode.

Demo Mode:

in this mode, the program will not open any new positions, and will only close outstanding trades based off of the rules of the system.

the program will operate as normal but instead of opening a position where it normally would, it would now create a variable. the program would then compare this value to the last candlestick of the live chart to determine if a profit or loss would have occurred.

graph A would now be plotted using these values, and the system will continue to assess itself. If successes surpass losses over a certain time period, it enters 'live' mode and trades again.

this is the basic principle of the EA, because i am new to this, i don't have a very wide range of good indicators. i already plan on using Imran's strategy for 1 5min indicator, but don't know what to do about the other 5 min, swing, or candlestick.

This is why i come to you my more experienced forex brothers and sisters for guidance and wisdom as i try to create my first successful EA

What you choose will ultimately be up to you. It sounds like you have a programming background, so put down the MQL language reference and start focusing on technical analysis instead. That will help you understand what the indicators tell you (leading/lagging/momo/etc) and you can choose from there. From there, you can start to build in some of general concepts and start the EA process. Having said that, the 1M and 5M choices will be interesting - my guess is that you're going to be jumping from demo to live quite a bit since the 10-pip swings are fast and will be hit-and-miss. Keeping the 1:1 ratio will be hard.

I am also consider my as an experience programmer.

In order to program a profitable EA, you must write down all the detail of your manual process and program it.
Later you will find out, the more you program, the more confuse you will get because of all the possible scenarios that could occur during robot trading.

But I am with you. PM me or keep posting. I will keep contributing to this thread.

You also forgot to mention backtesting.

Backtesting in visual mode is the programmer's way of "debugging". It can help you understand what's going on with your program and how it reacts in situations. You'll need to find charts that simulate easy stuff for your EA - there's a buy signal, does your program react? And you'll also have to test harder stuff - do trailing stops work properly?

Don't skimp on the backtesting.

LongToBeFree, i thought backtesting was understood, my apologies. also i am currently doing my own testing and know that it will take time to develop my own system, but if i am shown a strong indicator or system with with to make my 10 pip trades, then i will be able to test this and thus get a nice first step into the system.

I plan to develop this system much more, but i would like to develop the first version and have it running on demo. as you can see this is a system which will require at least 6 months of backtesting to validate, but i may begin live trading after 4 if the results are positive, and i can determine the best trading zones.

Here's what i'm thinking regarding trading conditions:

- to determine consolidation regions, the program will only trade if the MACD was above a certain value within a period of candlesticks.

- the program will compare current price to previous days highs and lows, and not trade when approaching these regions by a certain pip range, but will finish outstanding trades according to basic rules.

- the program will not hold open more than 2 trades at a time, for minimising of losses.

- if there are 3 all loss turns in a row the program enters demo mode until re-entry into live mode is verified.

- if total losses exceed 30 pips for the day it stops trading.

-the system sets profit targets for the days trade, at 50, 100, 150 then stops trade.

- if the indicator suggests a VERY strong gain potential, the program will leave the order open and use take profits and trailing stops to end the trade at 15 pips ->10 pips->10 pips

- the system will trade with the day's main trendline, and will will send mobile updates when in 'away' mode, and using a remote access program i could manually change the direction the system trades in, as well as modify other parameters if need be, and also put it into demo mode.
Also thanks to wifi, i could read news and react accordingly from my phone.

--------------------------------------------------

This next step may seem like a risky idea, but i think i can out enough measures in place to make it worthwhile:

After the system has reached a certain profit target for the day, it will begin to make trades based off of positive signals from selected indicators at a certain percentage. in other words i would let the system make a 'gamble' on the trade. this would obviously be done with a tight stop loss, like 4 pip or so, and use profit targets of 10->5->5 on the position. it would allow for 2 gambles for every 6 'perfect' trades.

So for example, i usually trade only if all indicators are positive, but after i make 50 pip profit for the day, the candlestick indicator, and 3 price indicators are positive, and the projected move is in the direction of the day's trend, but the remaining price indicators are negative.At this point, the program will random a true or false value, with say 20% chance of true, hence a 1:5 ratio of entering a trade. above 100 pip the ratio would increase to 2:5.

--------------------------------------------------

Right now my goal is to design a system which can make the biggest profits once given the right indications. in a sense the indicators used are inconsequential of the system and can be varied freely.

This thread is about creating a program that can make the most essential thinking processes a good forex trader makes, that is, analysing its performance, protecting investments, and trading in a disciplined manner.

--------------------------------------------------

If you're good at something the reactions become natural. this is why you don't look at your hands and feet when you're driving.

Think of a racing game, think Grand Turismo 5. now when the rival cars controlled by the computer is racing against you, there is a driver pictured in the car. now does the driver actually drive the car? the answer is obviously no, the car moves based on the basic rules of the road; don't get into accidents, approach a corner like this, accelerate when you're on a stretch, etc. When you and me drive our cars, we respond to situations with our knowledge of the road, while the AI does the same thing with the instructions its given.

You may be thinking this is true but the forex market is not based solely off of rules. well again to the case of GT5, if an AI car is sent off of its main course and into the side walls, it performs sufficient countermeasures to get back on course, all the while trying to avoid these accidents in the first place. By providing the trading system with a way to correct mistakes, it is easier to keep it trading profitably.

--------------------------------------------------

These are my 2 cents on the system thus far and welcome you the readers to comment on what i have said so far.


Thank you and congratulations on reading this lengthy post.