Help add extra parameters to the EA

Quote from: Turbo on December 19, 2018, 06:19:20 PM

  i have attached a backtest i made using the old version for the past 30days

I have backtested with different settings and I don't get results not even close to yours. I would be interested to know the exact settings you used and the broker.

Quote from: Turbo on December 19, 2018, 06:19:20 PM

Thanks for making its results constant and all the effort, I appreciate every single thing you've done, the backtests showed that the entry strategy changed
 for example
i.in backtest the first trade is always opened at asia open,
ii. those first few trades in the beginning are always profitable in every repeated backtest, you will need to download history from history center for the current 4Hr chart of the pair you want to backtest and restart mt4 before backtesting the original ver. (try audusd 5Min/1min chart TP25 SL65 Trail 4pips. use 0.1 fixed spread  20points or 2pips)

Please review the attached back tests. I have some points I want you to understand.

• There are 2 back tests from the modified EA. The results are the SAME. There are also 6 back tests of the original EA. They are all DIFFERENT. So the new version is consistent, but the old EA is more random. So it makes it difficult to claim the strategy logic has been changed if there is little or no foundation for comparison.
• You can see that the original EA has only sell orders. This is because the EA can't get values from the file and data[0] and data[1] are always ZERO! The modified EA has BUY and SELL orders, because data[0]=iTime(NULL, PERIOD_H4, 0) and data[1]=iClose(NULL, PERIOD_H4, 0) and these values are NOT zero. This is not a change in strategy, but a fix.
• You mentioned the first trades always opening at Asia open. The original EA placed the first trades at random times. So I think you mean the new EA opens trades at Asia open. The region is irrelevant to the EA. If the broker would be based in EST or GMT time, the EA would still open trades the same.

Quote from: Turbo on December 20, 2018, 02:54:31 PM

1. The modified one is consistent so you are right. I see you are using less than a month of history the original needs at least 2 months of history data plus the current 4hr bar's price details  before it calculates every entry, 
2. The original had both buy and sell working, but after the update of mt4 on 14 dec to build 1160 the buy trades stopped on the old version. you are right on the fix but i'm trying to figure out why the stable version is not very profitable. Since its stable, thats a major mile stone. .
3. at my end, the original usually open its 1st trade at asia open,

1) I'm not sure what you mean by less than a month of history. Are you talking about the duration of the back test only being for about a week? If so, the back tests I uploaded are only an example to make a point. I have also back tested on FxChoice for a month and even 2 months using different settings. The results are not even close to what yours are though no matter what.

2) The original EA is not very profitable in all my tests.

3) In the back tests I uploaded, you can see that the original EA made it's first trade at random times. For example, at 7:05, at 20:58, and so on. These times are so far apart and do not correlate to a specific time zone. The code of your EA has nothing in it to suggest regions such as Asia are considered.

i want to ask you a question that have been confusing me, in the code of the EA were you able to understand the calculation it does which is related to price details and time. ive been trying to figure it out since today. it really surprised me in my backtest

Quote from: Turbo on December 22, 2018, 10:33:47 PM

i want to ask you a question that have been confusing me, in the code of the EA were you able to understand the calculation it does which is related to price details and time. ive been trying to figure it out since today. it really surprised me in my backtest

The EA checks for live orders and if there is a live order then the EA Trails it. If there are not any live orders then the EA checks entry conditions:
1. if Close of last candle of H4 is MORE than Open of last candle of current TF then the EA opens BUY order;
2. if Close of last candle of H4 is LESS than Open of last candle of current TF then the EA opens SELL order;

Quote from: Turbo on December 30, 2018, 10:27:02 PM

hi admin, just a reminder, i never heard back from you. If you can add the options i suggested i will appreciate

Sorry for the delay. Here is what I think.

Your back test showed that in only a single month, the EA took $1000 to about $115000.

So I see two possible scenarios:

1) If your EA performs this well in real live trading (even just a small percentage of this profitability), it is a Gold mine. In this case, I would not change anything because the results are already impressive and there is no reason to change anything.

2) This EA is performing well because of unrealistic settings used in back testing (eg. small unchanging spreads) that do not mimic real live trading. Or because of other things like Trailing Stops which causes the profitability to be more in back testing, but not in live trading. I have seen other EA's like this before too. Even with 99.9% back testing data, unless your data is real tick-by-tick data, I do not think EA's like this will perform the same in live testing.


In either of the above scenarios above, I do not see any reason to add extra modifications right now. That is just my honest opinion based on my experience over the years. I would suggest you run the EA live to see if it has real life potential. If you have already tested live, please upload results here.

Quote from: Turbo on December 30, 2018, 11:32:55 PM

I never ran it live due to the issue it had regarding updating of history as i explained previously.

Please test one of the updated versions I gave you since that history problem will be gone now.

Quote from: Turbo on December 30, 2018, 11:32:55 PM

the spreads used were from 9-20 points and average profit per trade was between 7pips using a fixed spread test. its possible to achieve just a percentage of the result i posted if those details are added.

The back test results cannot be trusted because of two things you mentioned. First, live testing will not have a fixed spread. A true market will have changing spreads. Even if tested on a so-called fixed spread broker, they will most likely change their spreads at some point. Secondly, a small Trailing Stop of 7 pips will cause many EA's to show extreme profit. I have seen EA's just like yours before in my experience. This is because the back tester doesn't use tick-by-tick real market data. So it basically makes up the ticks. The price action that occurs in the back tester doesn't match the price action in a real market. So you will see that EA's that use Trailing Stops, especially small Trailing Stops, will show a lot of profit. However, when they are tested live, the real market price action is too random and volatile that the EA will not perform even close to how it does in the back tester. So I would suggest you run the EA live before proceeding with any more changes.

at the time of testing using what pair and what time frame, has it been tested for all pairs and the time frame, surely one of them will be able to make a positive result