
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Reshetov,
Thanks for the effort, but I notice in the code, you have made a big mistake, that in every perceptron, you have "Close[0]".
You CAN'T use Close[0] in the code, because you can't get it when "start()" is called when a new bar is opened. This is a serious mistake that will make the system only work in backtest, and loose real money in trading. So please don't use this system for living trading.
You can't see the future, because close[0] is in the future.
When Close[0] is changed to Close[1] or Open[0], the result is very bad. But if any correction is made, please let me know. Thanks!
There is a discussion about Close[0], which can be found here. Hope this helps.
Waigua2,
Close[0] is simply current price. It's price without the spread. He's not actually trying to predict the future. He's simply checking the relationship between opening price and current price at any given tick. My bet is that you are not familiar with Reshetov. The odds of him making that basic of a mistake are very slim. Hope this helps.
Was anybody successful in using this EA ? I have to admit that my results are not looking pretty. I might have missed something here. Can you use it on any pairs?
Anybody could explain further? Thanks P.
Have you read the link I gave in my first post? When I check the code from Close[0] to Open[0], the result is aweful. The reason why nobody can get Close[0] used for EA is that when using forward testing, EA can't make a decision when the bar is just opened in start(). The warning is that if "Close[0]" is predicted from Open[0] using whatever method outthere, the test by Reshetov gives the best performance possible IF Close[0] is correctly predicted.
So maybe you are right if this is the intention of the author of this system. So this EA is lack of a prediction component which "foresees" Close[0] accurately for real trading, but this system illustrates very well the idea of using feed forward NN in trading. For short term trading it seems that lagged price difference can be used. I only read some papers talking about using price percentage for trading long term with NN.
So, if anything needs to be added, one good regression method is needed to be added to replace Close[0] to finish the EA for real trading.
Hi,
Which TimeFrame is best for this EA??
TIA
Hi,
Which TimeFrame is best for this EA??
TIA
This EA can't be used in real trading, as I said, you will probably loose money for trading using this system. So it does not matter whichever time frame is going to use.
i tried this combo for two weeks, pretty impressive in the beginning, many trades, but in last high floating account and then it just wipe out my demo account.
but i noticed that, every deals that were open on the evening and before the next morning was very good.
maybe you have a good ea, but it can make extreme losses on the bull and bear periods.
can you program a starttime and stoptime for trading on the ea and a automatic trailing stop?
i tried this combo for two weeks, pretty impressive in the beginning, many trades, but in last high floating account and then it just wipe out my demo account.
but i noticed that, every deals that were open on the evening and before the next morning was very good.
maybe you have a good ea, but it can make extreme losses on the bull and bear periods.
can you program a starttime and stoptime for trading on the ea and a automatic trailing stop?
i tried this combo for two weeks, pretty impressive in the beginning, many trades, but in last high floating account and then it just wipe out my demo account.
but i noticed that, every deals that were open on the evening and before the next morning was very good.
maybe you have a good ea, but it can make extreme losses on the bull and bear periods.
can you program a starttime and stoptime for trading on the ea and a automatic trailing stop?
Waigua2 and Warrismrazvi,
Combo has to be "trained" quite often, I found that it does fairly well for it's simplicity in programming, but it does need some improvement. It seemed to work best when it was trained often and using short time periods (1 to 2 weeks max). The longer the time period used to train it, the more erratic it became.
I think this same priniciple applied to indicators instead of price action has a better chance of being successful. Price action is too random at times to be lerned (in my humble opinion). Indicators have more "legible" patterns (crosses, distances, etc).
Waigua2, combo is "learning" the relationship between current price (Close[0]) and previous opens. It's basically making it's decision based on the position of current price in relationship to several previous opens. It's not predicting (Close[0]). It's trying to predict direction based on Close[0]'s position relative to past Open's. Close[0] exists and is a very useful piece of data if you don't want to make decisions using Ask and Bid everytime you want to establish current price. I know it exists, because I have used it extensively in my programming. If you simply look on your chart at any given moment and see the price line moving up and down, that is Close[0]. You can know when Close[0] is above, below or equal to any value and make a decision based on the position of Close[0] (or current price).
I suspect one of us may not be understanding the other, because this is a simple and basic concept as far as I know. It has been my experience that when two people over-discuss a simple concept, usually they are saying the same thing in different ways, or they are not understanding what each other is saying.
Again, I truly hope this helps in some way.
Just want to bring attention back to NN.
IT IS PROFITABLE and CAN bring good income:
Check for yourself - IT IS NOT test from the past it is todays data:
Sorry the picture did not come up well.
To make it even more real: here is the same EA runs from adifferent broker platform:
Account: ffffffffff
Name: ccccccccc
Currency: USD
2009 January 21, 17:04
Closed Transactions:
Ticket
Open Time
Type
Size
Item
Price
S / L
T / P
Close Time
Price
Commission
Taxes
Swap
Profit
58463032
2009.01.20 05:56
balance
Deposit
1 000.00
58464689
2009.01.20 06:11
sell
0.10
gbpusd
1.4276
1.4258
1.4170
2009.01.20 06:58
1.4209
0.00
0.00
0.00
67.00
58465143
2009.01.20 06:16
sell
0.10
gbpusd
1.4281
1.4258
1.4175
2009.01.20 06:58
1.4209
0.00
0.00
0.00
72.00
58465807
2009.01.20 06:20
sell
0.10
gbpusd
1.4270
1.4258
1.4164
2009.01.20 06:58
1.4209
0.00
0.00
0.00
61.00
58466851
2009.01.20 06:25
sell
0.10
gbpusd
1.4262
1.4258
1.4156
2009.01.20 06:58
1.4209
0.00
0.00
0.00
53.00
58469293
2009.01.20 06:30
sell
0.10</s
Kudos to you Reshetov. An astonishingly simple and elegant example! Many thanks.