Using neural networks in trading - page 12
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
In short, it's early, I've still got a lot of work to do.
In short, it's early, I've still got a lot of work to do.
Saw, Shura, saw, she's golden!
Oh, yes! It's true that sometimes it looks a bit like ananism, but I'm sawing.
Ticket Open Time Type Size Item Price S/L T/P Close Time Price Commission Taxes Swap Profit
6598986 2013.01.27 22:15 buy 0.01 eurusd 1.34564 0.00000 0.00000 2013.01.28 06:00 1.34555 0.00 0.00 0.00 -0.09
6599063 2013.01.28 00:00 buy 0.01 eurusd 1.34657 0.00000 0.00000 2013.01.28 08:00 1.34316 0.00 0.00 0.00 -3.41
6599082 2013.01.28 01:00 buy 0.01 eurusd 1.34604 0.00000 0.00000 2013.01.28 09:00 1.34443 0.00 0.00 0.00 -1.61
6599116 2013.01.28 02:00 buy 0.01 eurusd 1.34548 0.00000 0.00000 2013.01.28 10:00 1.34554 0.00 0.00 0.00 0.06
6599176 2013.01.28 04:00 buy 0.01 eurusd 1.34529 0.00000 0.00000 2013.01.28 12:00 1.34366 0.00 0.00 0.00 -1.63
6599217 2013.01.28 05:00 buy 0.01 eurusd 1.34601 0.00000 0.00000 2013.01.28 13:00 1.34393 0.00 0.00 0.00 -2.08
6599412 2013.01.28 07:00 buy 0.01 eurusd 1.34467 0.00000 0.00000 2013.01.28 17:09 1.34509 0.00 0.00 0.00 0.42
6600763 2013.01.28 15:00 sell 0.01 eurusd 1.34667 0.00000 0.00000 2013.01.28 23:00 1.34466 0.00 0.00 0.00 2.01
6601001 2013.01.28 16:00 sell 0.01 eurusd 1.34596 0.00000 0.00000 2013.01.29 00:00 1.34483 0.00 0.00 -0.01 1.13
6601152 2013.01.28 17:00 sell 0.01 eurusd 1.34578 0.00000 0.00000 2013.01.29 01:00 1.34532 0.00 0.00 -0.01 0.46
6601251 2013.01.28 18:03 sell 0.01 eurusd 1.34544 0.00000 0.00000 2013.01.29 02:00 1.34481 0.00 0.00 -0.01 0.63
6601393 2013.01.28 19:00 sell 0.01 eurusd 1.34484 0.00000 0.00000 2013.01.29 03:00 1.34526 0.00 0.00 -0.01 -0.42
6601480 2013.01.28 20:00 sell 0.01 eurusd 1.34550 0.00000 0.00000 2013.01.29 04:00 1.34532 0.00 0.00 -0.01 0.18
6601520 2013.01.28 21:00 sell 0.01 eurusd 1.34539 0.00000 0.00000 2013.01.29 05:00 1.34580 0.00 0.00 -0.01 -0.41
6601563 2013.01.28 22:28 buy 0.01 eurusd 1.34493 0.00000 0.00000 2013.01.29 06:00 1.34468 0.00 0.00 -0.02 -0.25
6601571 2013.01.28 23:00 buy 0.01 eurusd 1.34469 0.00000 0.00000 2013.01.29 07:00 1.34420 0.00 0.00 -0.02 -0.49
6601621 2013.01.29 00:28 sell 0.01 eurusd 1.34483 0.00000 0.00000 2013.01.29 08:00 1.34391 0.00 0.00 0.00 0.92
6601675 2013.01.29 02:04 sell 0.01 eurusd 1.34482 0.00000 0.00000 2013.01.29 10:00 1.34422 0.00 0.00 0.00 0.60
6601818 2013.01.29 05:00 buy 0.01 eurusd 1.34580 0.00000 0.00000 2013.01.29 13:00 1.34294 0.00 0.00 0.00 -2.86
6601860 2013.01.29 06:00 sell 0.01 eurusd 1.34469 0.00000 0.00000 2013.01.29 14:00 1.34426 0.00 0.00 0.00 0.43
6601935 2013.01.29 07:00 sell 0.01 eurusd 1.34429 0.00000 0.00000 2013.01.29 15:00 1.34849 0.00 0.00 0.00 -4.20
6602048 2013.01.29 08:00 sell 0.01 eurusd 1.34375 0.00000 0.00000 2013.01.29 16:00 1.34778 0.00 0.00 0.00 -4.03
6602163 2013.01.29 09:00 sell 0.01 eurusd 1.34539 0.00000 0.00000 2013.01.29 17:00 1.34847 0.00 0.00 0.00 -3.08
6602265 2013.01.29 10:00 sell 0.01 eurusd 1.34403 0.00000 0.00000 2013.01.29 18:00 1.34813 0.00 0.00 0.00 -4.10
6602602 2013.01.29 12:00 sell 0.01 eurusd 1.34293 0.00000 0.00000 2013.01.29 20:00 1.34879 0.00 0.00 0.00 -5.86
6602803 2013.01.29 13:00 sell 0.01 eurusd 1.34294 0.00000 0.00000 2013.01.29 21:00 1.34923 0.00 0.00 0.00 -6.29
6603200 2013.01.29 15:00 sell 0.01 eurusd 1.34824 0.00000 0.00000 2013.01.29 23:00 1.34883 0.00 0.00 0.00 -0.59
6603422 2013.01.29 16:00 sell 0.01 eurusd 1.34757 0.00000 0.00000 2013.01.30 00:00 1.34888 0.00 0.00 -0.01 -1.31
6603696 2013.01.29 17:00 buy 0.01 eurusd 1.34842 0.00000 0.00000 2013.01.30 01:00 1.34872 0.00 0.00 -0.02 0.30
6603906 2013.01.29 19:00 sell 0.01 eurusd 1.34817 0.00000 0.00000 2013.01.30 03:00 1.34943 0.00 0.00 -0.01 -1.26
6605292 2013.01.29 22:15 buy 0.01 eurusd 1.34902 0.00000 0.00000 2013.01.30 06:00 1.34872 0.00 0.00 -0.02 -0.30
6605895 2013.01.29 23:00 buy 0.01 eurusd 1.34884 0.00000 0.00000 2013.01.30 07:00 1.34898 0.00 0.00 -0.02 0.14
0.00 0.00 -0.18 -36.99
Closed P/L: -37.17
Open Trades:
Ticket Open Time Type Size Item Price S / L T / P Price Commission Taxes Swap Profit
6608678 2013.01.30 04:00 buy 0.01 eurusd 1.34901 0.00000 0.00000 1.35434 0.00 0.00 0.00 5.33
6609644 2013.01.30 05:00 buy 0.01 eurusd 1.34891 0.00000 0.00000 1.35434 0.00 0.00 0.00 5.43
6609776 2013.01.30 06:00 buy 0.01 eurusd 1.34890 0.00000 0.00000 1.35434 0.00 0.00 0.00 5.44
6610623 2013.01.30 08:00 buy 0.01 eurusd 1.35131 0.00000 0.00000 1.35434 0.00 0.00 0.00 3.03
6610769 2013.01.30 09:00 buy 0.01 eurusd 1.35064 0.00000 0.00000 1.35434 0.00 0.00 0.00 3.70
6610990 2013.01.30 10:00 buy 0.01 eurusd 1.35314 0.00000 0.00000 1.35434 0.00 0.00 0.00 1.20
6611218 2013.01.30 11:00 buy 0.01 eurusd 1.35568 0.00000 0.00000 1.35434 0.00 0.00 0.00 -1.34
0.00 0.00 0.00 22.79
Floating P/L: 22.79
I lost a lot of money using NS in trading. Reason - re-training the NS.
Can you elaborate. What type of network, with back propagation of error or not, with a teacher or not... ?
Anything you think you need to do to draw conclusions and not make the same mistakes...
Thank you.
I lost a lot of money when using NS in trading. The reason is overtraining the NS.
the reason is different. redundant network.
Can you be more specific? What kind of network, with or without error back propagation, with or without a teacher... ?
Anything you think you need to do to draw conclusions and not make the same mistakes...
Thank you.
MLP network type, primitive learning, take random weight and random step. e.g. there was a profit on sample 0.434, take random weight (-0.13), take random step (0.1). Take the first step in the random direction, let it be plus (0.03), profit has decreased, divide the step by 2 and go the other way (-0.18), profit has increased, divide the step by 2 and go the same way (-0.205) and so on. But it is very slow and inefficient. So that's the way it is for now.
I used 1000 networks with different parameters (number of layers, neurons, it's also called Ensemble Learning) trained and tested on different indicators and data, so the forecast error would be reduced by averaging. The learning algorithm is BP, genetic algorithm. The result is very bad, I figured out that prediction model should be as simple as possible. Conclusion - use rule set + genetic algorithm + Ensemble.