Cross Slopey Bob

[spotdespot]

V 1d is in post 1. Copied from the amended post 1:

Probably means we will get one trade every 6 months, but as coded the bot has been a guaranteed loser so it is worth trying.

Hi Steve,

Yes it hasn't been looking good but where it is supposed to make pips is when there is some trending movement and we haven't had a great deal with retracements Thursday/Friday (& NFP) and a public holiday tomorrow in the US probably doesn't help. There has been some movement this afternoon. I guess it is just a matter of seeing if the strong movements can make up for the inevitable losses when markets go sideways. In theory the losses should be small relative to the gains (per trade) but whether that's enough or not remains to be seen.

The change I mentioned is not quite as you have programmed above. I don't think what you have done there will ever allow a trade. The terminology of CSS means that "down" is below zero and "up" is above zero - so if the currencies are UP and accelerating VS DOWN and decelerating then they crossed some time ago and are moving apart (I think anyway!!).

What I amended was this:

Code: Select all

         //Detect a cross of the currency pairs
         CssCrossStatus = nocross;
         //Long cross where currencies diverging
      if (CurrVal1[2] <= CurrVal2[2])
         if (CurrVal1[1] > CurrVal2[1])
            if (CurrDirection1 == upaccelerating || CurrDirection1 == downdecelerating)
               if (CurrDirection2 == downaccelerating || CurrDirection2 == updecelerating)
               CssCrossStatus = longcross;     
                        
         //Short cross where currencies diverging
      if (CurrVal1[2] >= CurrVal2[2])
         if (CurrVal1[1] < CurrVal2[1])
            if (CurrDirection1 == downaccelerating || CurrDirection1 == updecelerating)
               if (CurrDirection2 == upaccelerating || CurrDirection2 == downdecelerating)
               CssCrossStatus = shortcross;
 

To give a glimmer of hope this is what I have open right now

Cheers,
Dave.

[SteveHopwood]

Good point Dave. Your latest adaptations adopted into V 1e in post 1.

[SteveHopwood]

I left Dave's updated version (in post 1) running overnight and awoke to a basket of trades showing +10% profit on the start balance of $500. A few losers had closed, taking the balance down to $490, so bagging the pips in the open basket left the account at $541.

[spotdespot]

I left Dave's updated version (in post 1) running overnight and awoke to a basket of trades showing +10% profit on the start balance of $500. A few losers had closed, taking the balance down to $490, so bagging the pips in the open basket left the account at $541.

He he - doing better than me then I think I am trying to be too clever with the trade management and managing the opposite! Now there is some movement it has some promise - it was looking like a very damp squib.

[SteveHopwood]

I checked some of the snapshots to see what is on the chart at the time the trade is taken. The order of events as coded is: new candle; read the indicators and discover a trade signal; send the trade; take the snapshot; refresh the chart feedback.

This means the picture is always out of date.

Anyone wanting a fix, go to lines 1483/4 and replace them with:

Code: Select all

            if (TakeSnapshots)
            {
               DisplayUserFeedback();//Ensure the chart feedback is correct.
               TakeChartSnapshot(TicketNo, " open");
            }//if (TakeSnapshots)

[SteveHopwood]

Dave, your idea to trade only diverging currencies is working so far.

XAGUSDH1.png

[spotdespot]

Dave, your idea to trade only diverging currencies is working so far.

Excellent. I am trying something else at the moment - no code change just a different approach. I am essentially looking at it on M5 and just scalping 10 pips or so. It seems that during non-Asian hours almost all pairs that meet the cross criteria move enough to bag 10 pips a cross. I tried on M30 but even though I built a 1000 pip basket I had some significant losers as the re-cross to close was too slow.

Unfortunately I can only test one idea at a time as I still have that weird memory issue.

Could you do me a favour please Steve and check your libCSS file and see what version is says it is? I downloaded the last one from Paul's thread - which is supposed to be v 1.1.3 but the version number in the code header say 1.1.2 Clutching at straws a bit, as probably just a comment oversight, but worth a look as this memory situation is painful.

Cheers,
Dave.

[SteveHopwood]

Paul's release notes say v1.1.2, 9/6/13

[spotdespot]

Paul's release notes say v1.1.2, 9/6/13

Ok - I was clutching at straws then. His LibCSS thread says:

"August 19, 2013: new version v1.1.3

Changes:
Optimized code. Tighter, stronger, faster
Fixed bug in caching mechanism"

So as your version is the same as mine it was a dead end. Thanks for checking.

Cheers,
Dave.

[SteveHopwood]

My demo is up just over 10% on its starting balance.

The problem I am seeing is that there are lots of little losers closed that hit the bottom line, then a decent run allows me to close to overcome the losers. The result is that I only move ahead very slowly, and that one day the losers might overwhelm the winners and wipe out all of the profits.

Not sure this bot has a future.