Neural Network Test | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Neural Network Trading, Serious people only! 127 replies

Neural Network trading using Matlab and Metatrader 8 replies

neural network based systems in forex 22 replies

Neural Network anyone? 12 replies

The "Better" neural network 53 replies

Neural Network Test

  • #1
  • First Post: Jan 14, 2010 3:09am Jan 14, 2010 3:09am
Hello All.

I'm now forward testing my neural network after some extensinve development period.

The neural network tries to find a trend by measuring volatility, natural and historic support/resistance areas, and accounting for different time of day/week/month based on historic movement.

Once the trend is established, the system will start trading heavily in that direction.

If it spots a change in the trend, it will start closing the open trades - once everything closed, it will trade in the direction of the new trend.

As many here, I've connected my account with Myfxbook to shown verified results.

This is my first version and try of a neural network so many changes could be made before I'll go live.

Here's also a live widget from Myfxbook:

http://www.myfxbook.com/widget?id=13...pe=2&color=red
  • #2
  • Jan 14, 2010 3:38am Jan 14, 2010 3:38am
Here's an excellent read for anyone interested: http://www.statsoft.com/textbook/neural-networks/
  • #3
  • Jan 14, 2010 5:17am Jan 14, 2010 5:17am
Just some more information of my system:

By natural support and resistance areas I mean to the round numbers areas and the middle's 1.4000, 1.4050, 1.5000 and so one.

Historic support and resistance are calculated by finding price bouncing in the same area with a predetermined distance, and of course in a certain timeframe.

Also, the historic movement is statistically analyzed to get another advantage by knowing the most likely time of day for the symbol to move the most and the least (in case of least, try to not open positions).
  • #4
  • Jan 14, 2010 11:31am Jan 14, 2010 11:31am
You might notice I have both buy and sell positions at the same time - the algorithm "tries" every possible trend change. So even with a clear trend up, it will open a few sell positions down the road just in case it changes.

I guess this is a kind of a grid system, but I'm still working it.
  • #5
  • Jan 15, 2010 4:10am Jan 15, 2010 4:10am
Ok, not much to report.

The system is trading well for now, closed 4 trades; 2 for a loss and 2 for a profit. Overall it's up 0.06%.
  • #6
  • Jan 15, 2010 7:52am Jan 15, 2010 7:52am
The eur is dropping but the system still thinks the trend is upwards, so it keeps buying at different levels.

Not too worried about the drawdown as with current postions sizing, it can withstand a trend change quite nicely (need to go over the numbers though for an exact figures).

We'll see how it goes...
  • #7
  • Jan 18, 2010 5:14am Jan 18, 2010 5:14am
Slow and steady wins the race!

It might look like the progress is very small or non existent even, but I do have a rigorous money management scheme, so the risk is kept minimal, and there's plenty of room for mistakes.

I have currently 3 buys open with a small floating loss.

Waiting for the trend to resume, the price is still in a tight range since the weekend.
  • #8
  • Jan 18, 2010 1:05pm Jan 18, 2010 1:05pm
Ok, I'll be increasing the position sizing.

The market is practically on pause, and it's like looking at paint dry - need to see some action!
  • #9
  • Jan 18, 2010 1:57pm Jan 18, 2010 1:57pm
im interested today is a holiday in us low volatility .. hope to see how this works im not to good in the whole genetic algorythm and neural network stuff so should be interesting... good luck does it take into account fundamentals at all?
  • #10
  • Jan 18, 2010 4:38pm Jan 18, 2010 4:38pm
Quoting domino
im interested today is a holiday in us low volatility .. hope to see how this works im not to good in the whole genetic algorythm and neural network stuff so should be interesting... good luck does it take into account fundamentals at all?
Thanks..

No, it doesn't take into account fundamentals, just pure price action.
  • #11
  • Jan 19, 2010 3:02am Jan 19, 2010 3:02am
Finally starting to see some movement. Although I've doubled the position sizing, it it still very conservative and I just might to increase it again.

Closed another trade in profit.
  • #12
  • Jan 19, 2010 7:19am Jan 19, 2010 7:19am
I think we're near a major crossroads with the eur - it just might stop up trending and plumb downwards.

My system might lose some money when the trend changes, but should make it all back and more when a new trend establishes.
  • #13
  • Jan 20, 2010 5:37am Jan 20, 2010 5:37am
Quoting wilkinson
I think we're near a major crossroads with the eur - it just might stop up trending and plumb downwards.
Looks like it's indeed breaking downwards - the move should be pretty big and fast, so I will keep my eye on it.
  • #14
  • Jan 20, 2010 12:35pm Jan 20, 2010 12:35pm
Quoting wilkinson
Looks like it's indeed breaking downwards - the move should be pretty big and fast, so I will keep my eye on it.
dont get caught in the 146 pip upmove comming.
  • #15
  • Jan 20, 2010 1:46pm Jan 20, 2010 1:46pm
Quoting domino
dont get caught in the 146 pip upmove comming.
And how do you know that if I may know?
  • #16
  • Jan 20, 2010 2:07pm Jan 20, 2010 2:07pm
Quoting wilkinson
And how do you know that if I may know?
psychological price action ... confirmation wave. of course we dont know what will happen in the future only what has happened the most.
  • #17
  • Jan 21, 2010 1:55am Jan 21, 2010 1:55am
Quoting domino
psychological price action ... confirmation wave. of course we dont know what will happen in the future only what has happened the most.
For when do you predict this move then?
  • #18
  • Jan 21, 2010 9:13am Jan 21, 2010 9:13am
The account is not progressing as expected..but maybe it's because of the rapid move. I guess I should let it trade for a few weeks before jumping to any conclusions.
  • #19
  • Jan 22, 2010 2:52am Jan 22, 2010 2:52am
Looks like the market can't decide where to go - so is my system
  • #20
  • Jan 22, 2010 3:27am Jan 22, 2010 3:27am
Quoting wilkinson
Looks like the market can't decide where to go - so is my system
I'm not sure what rules you presently use to determine the beginnings and ends of trends, but for your system to be successful it needs to be think like a successful trend trader.

Most successful trend traders use the rule that once price closes below a trendline, and then re-tests the trendline and is rejected, then odds are the trend is broken. For trends drawn on a Daily chart, this test is best seen on a 4H chart, for trends drawn on a 4H chart, this test is best seen on a 1H chart, and ect, ect.

It also wouldn't hurt to have your system analyze bars for price action to help confirm wheather the trend has been broken or not.

Also the market may seem like it's confused sometimes, but this only means that all the action is taking place on a smaller timeframe. Once you take a look at the smaller timeframe, everything will make sense.
0 traders viewing now