Neural network (deep learning) EA & some manual trades

$$=2bucks

where is your paypal

) well spending half a year on understanding neural networks, programming then it in python, then programming mql code in metatrader, paying for fast linkage between these two systems which cost $250, i think it is not the deal to give it for you for only $2

#6 trade. Buy GBPJPY hit SL:
1. Yesterday we got an announcement of UK-EU agreement for the divorcement bill. This is major new for GBP and my SL was shaked. Of course, can blame only myself as i put SL at 147.2 level while minor low was at 146.9 level.

#12 trade buy GBPJPY. Reentering it at lowe price. Why:
1. There is a clear bill negotiatied between UK and EU. Which is positive for GBP. Also GBP increase its interest rates.
2. Yen interest rate policy is still low, no one talks about increasing it.
3. GBPJPY yesterday formed candle with big spike to the bottom and volume is increased.

#13 trade. Sell EUR/USD. Why:
1. Yesterday we got an announcement of UK-EU agreement for the divorcement bill. Euro in long term will weaken without GB in its system.
2.Dollar is still on the trend to increase interest rates on December.
3. Formation of bar going down with closing below previous bar open price and volume increased.

#14 short Gold. Why:
1. Dollar started getting stronger
2. rejecting from 1300 level
3. H4 timeframe shows bigger bars to downside than upside

#8 trade. Sell EURJPY. - SL
Main thesis on entry:


So it passed almost 3 weeks with no move despite in divergence and price action to the downs, and broke my SL. If it closes today above 133.2 level, next target to it will be 134.5 level

#15 buy USDJPY. Why:
1. Next week will be interest rate number, which expects still interest rate to rise
2. Yen does not show strong signals up
3. D1 made high volume bar on Friday with long spike to the downside, which is bullish bar
4. RSI bounced from 30 level

Which could cause a bit of disturbance:
1. This week we will have NFP and estimate is much smaller which could lead to have weaker USD

Move this shit to commercial

Closing my all trades related to USD and JPY pairs before tomorrow NFP report

#16 buy CHFJPY(technical only). Why:
1. Price broke resistnace triangle level
2. D1 RIS was jumping from 50 level up, both price and rsi did not move lower than previous level
3. Momentum also started increasing

#17 bought EURUSD yesterday after rates announcement on dollar drop. Why:
1.market had expectation it will rise 25basis
2.From four hike be three hike in 2018
3.US treasury 10 year drop

#18 bought GBPJPY. Why:
1. Yes is still weak fundamentally
2. Some rumours on GBP that it is looking at distinct trade deal with EU
3. Price bounced from 149.5 level

What i don't like:
1. GBPJPY now boucning between 152-146 level, so it could go back to 146 for retesting its strength

Closing my GBPJPY, EURUSD and CHFJPY trades , taking +1.5% from them.
I think this year i won't open anymore new trades with bot or with manually.
So are the results for 3 month of trading:
1. Manual trades(do not count USDCAD here as it has auto trading) made 5% profit.
2. USDCAD auto trading with my bot made more than 25%.
3. Maximum drawdown of portfolio - only 10%

Feeling great and hyped for next year!

New year, new opportunities!
1. Buying GBPJPY. Price at resistance 153 level. However, it moved to that price without any divergence and it formed big triangle. So soon that price will move either to 160 level or back to 147. Pound is strong and there were rumors that they increase interest rate in first half, so trying to buy rumours

2. Selling little USDCAD. Price broke 1.245 level so will try to move in broke direction. Also on Friday CAD announced strong unemployment number, which could lead to increase in interest rate

3.Selling EURGBP. On the same strong GBP rumours and broken D1 consolidation area.

4.Selling AUDUSD. Against a trend, but it moved without any retracement 8 days in a row, so possible retracemwnt with good RR ratio/

5.Buying natgas feb contracts. In USA we are getting strong cold coming from Canada, also reserves currently are much lower that historical due to climate warming, so buying on expected consumption and possible increase of reserves

Hello, just passing by to tell you that it is a nice analysis and work you do, don't give up!

The main problem i see with neural networks and data mining in general is that the system has to be calibrated frequently to consider the recent structural changes and shifts. This is another sophisticated way to curve-fit the data, and that's the biggest pitfall of this methodology (IMHO). When it effectively catches a valid repeteable pattern you have to deal with the competition of other sophisticated traders that are catching the same patterns (impacting and eroding the edge, or even preying on it).

Sadly i don't see many ways around this, but to try another market with less ferocious neural net competition or play around with the risk management (increasing targets, trying other types of risk management, etc). Kudos!

so tell me about your programming knowledge?

I believe the reason why were your EA not working IS BECAUSE THE TICK DATA YOU"VE DOWNLOADED ARE IN FAULTY!!! TAKE A LOOK AT THE MISMATCH ED CHART ERROR!!!! GOOD QUALITY TICK WILL REMAIN AS 0!!!! HERE"S AN EXAMPLE!!!! BOLD!!!

PM me if you need a strong computer to backtest those data, with your EA, I most symbol data.

Hello,
Data was downloaded from broker. Are there some different sources which can provide data? not only OHLC prices, but also RSI/MACDs and so on

Hi not sure you are still working on it or not. Just to inform you some pitfalls of the automation trading system as well as manual trading.

Manual Trading :-
- Most of the manual trading emphasizing on trading levels/zones, and most important is reward risk etc etc stuff. Reward risk is fundamental for the system to get profitable. However, the things are made equilibrium, the higher the reward risk, the lower the success rate for your manual trading strategies, for eg hit s/l more frequent before it going to profit. That is why many gurus emphasizing on "psychology" as seriously, the too many tiny losses, quite often if feel very painful when being accumulated. Success rate of 5/10 will not guarantee you win 5 trades in all of the 10 trades, u might loss the first 10 trades, before can goes into winning streak. When you face the winning streak, due to the past lossing experience, you tend to exit earlier, making your system cannot cover back the losses eventually. Hence at the end, summary for it is this is too volatile, if you know what I mean.

Automation Trading :-
- For your system, I don't know which neural network layer or methods you applying. Do you use the DNN to direct predict the price? If yes, what I can tell you is it will never work. One concept you need to understand in trading is "stationarity". However, almost all of the financial market/instrument are not stationary, so we hard to predict it except "overfitting" it as mentioned above by other guys. So you will get impressive backtest result, no harm as it really pretty cool, but when being "test forward" or out of sample test it just failed. The theory is just like you give a kid to do an exam paper, initially he scored badly, however you provide him an answer scheme, and make him memorize the answer, so in the repetition of the similar test, he can score 100 marks with no surprise. However, when you give him another new test, he will score terribly again, because he don't exactly know what he's doing. The kid can be analogy to your current system, you can't just give feeding him answer scheme, but you need to "teach" him some knowledge, so he can perform in other new test as well. If you get what I mean.

p/s : I attend FF once in a blue moon, out of sudden google search result pop out your thread, think you might need some helps / advices. Hope my advices did helped. One thing is have faith and perseverance in anything you are currently doing, success will not come in odd, it's reserved for truly hardwork guys! Good luck!

Hi,
may I ask what deep learning system you use as your basis. I was thinking about including an AI in my own project, but not sure which one to choose.
Cheers...