Maurice’s Journal

[SaidBitar]

Maurice Petterlin wrote:

I was doing any stress testing and now I have to go back over the notes and find out what I need to do after the Monte Carlo. Will have to make some notes of the procedure of stress testing. Need to set up a cookbook method.

Maybe

1. Single run backtest
2. Monte Carlo
3. Walk Forward In Sample out of Sample
4. Rolling Analysis
etc etc

I am making a file in excel that will display the procedure for the stress testing and will take all the results that are generated by Amibroker then will analyze them and generate report in word and powerpoint.

Reporting is one thing that i do not like but still can not be avoided this is why i prefer to automate it.

when it is ready i will share it

[LeeDanello]

You’re too kind

[LeeDanello]

Got myself a new laptop – HP ProBook 470 G3.

These are the specs..

Processor: Intel® Core™ i7-6500U with Intel HD Graphics 520 (2.5 GHz, up to 3.1 GHz with Intel Turbo Boost Technology, 4 MB cache,)
Operating system: Windows 10 Pro 64
Memory: 16 GB DDR3L SDRAM (2 x 8 GB)
Storage: 1 TB 5400 rpm SATA
SSD: Kingston 480Gb M.2 SSD (3rd party)
Displays: 17.3″” diagonal HD+ SVA anti-glare flat LED backlit (1600 x 900)
Display and Graphics: AMD Radeon™ R7 M340 (2 GB DDR3 dedicated)
Ports: 2 USB 3.0; 2 USB 2.0; 1 HDMI; 1 headphone/microphone combo; 1 AC power; 1 RJ-45; 1 VGA

Basically it just needs wings. The biggest problem is migrating from a 32 bit enviroment to 64 bit which means that you can’t use an image to transfer from the old computer. Correct me if I’m wrong. The biggest pain was loading a new install of Amibroker and replicating all my old layouts. On the old computer I stored all my data on a separate hard disk, using the primary disk for the operating system. When you have a SSD you want everything on that and not on a seperate disk unless that is an SSD. Basically a lot of my old formulas didn’t quite work on the new computer because they had references to the old hard disk. This affected include files also. The old computer had only 2GB of ram which meant that a MC analysis could take hrs maybe 4 or 5. With 16GB it takes an hour and a half.

I now I have the capability of producing heaps of reports for Nick to read. I just hope it’s not all crap! Garbage in = Garbage out and sometimes that can be Garbage in = (Garbage Out)ⁿ where n = the number of output reports

[SaidBitar]

You will feel the power of new laptop when you will run Monte Carlo back testing, this is why I bought my work station at the end of last year because I got really pissed off waiting for mcs to finish running.

Regarding migrating the image it will not work, so the painful part will be configuring everything again.

Regarding files not working due to references you need to solve it one at a time.

[LeeDanello]

For the reasons you stated I bit the bullet and got this one. It had to be a laptop for portability.
It too me a little while to work out why things in Amibroker were different to the old machine.
MC analysis on the S&P 500 for 20 years takes about an hour and 10

[LeeDanello]

I found this zig zag indicator on the Amibroker website which was written for the S&C website called the The 1-2-3 Wave Count. It plots lines connecting highs to lows and vice versa but does not use the zig function to do this.

Code:

ZZPercent = Param(“ZZPercent”, 5 ); 
ATRPeriod = Param(“ATRPeriod”, 5 ); 
ATRFactor = Param(“ATRFactor”, 1.5, 0, 5 ); 

HLPivot = ZZPercent * 0.01 + ATRFactor * ATR( ATRPeriod )/Close; 

Ll = Low[ 0 ]; 
Hh = High[ 0 ]; 
Llb = Lhb = 0; 

if( High[ 1 ] >= Hh ) 
{ 
   Hh = High[ 1 ]; 
   Lhb = trend = 1; 
} 
else 
{ 
   Ll = Low[ 1 ]; 
   Llb = 1; 
   trend = -1; 
} 

Line = Null; 

for( i = 2; i < BarCount; i++ )  {     if( trend > 0 ) 
   { 
      if( High[ i ] >= Hh ) 
       { 
          Hh = High[ i ]; 
          Lhb = i; 
          Curline = LineArray( Llb, Ll, Lhb, Hh ); 
          Line = IIf( IsNull( CurLine ), Line, CurLine ); 
       } 
      else 
      if( Low[ i ] < Hh - Hh * HLPivot[ i ] )         {            Ll = Low[ i ];            Llb = i;            trend = -1;            CurLine = LineArray( Lhb, Hh, Llb, Ll );            Line = IIf( IsNull( CurLine ), Line, CurLine );         }      }     else     {        if( Low[ i ] <= Ll )         {            Ll = Low[ i ];            Llb = i;            CurLine = LineArray( Lhb, Hh, Llb, Ll );            Line = IIf( IsNull( CurLine ), Line, CurLine );         }        else        if( High[ i ] > Ll + Ll * HLPivot[ i ] ) 
       { 
          Hh = High[ i ]; 
          lhb = i; 
          trend = 1; 
          CurLine = LineArray( Llb, Ll, Lhb, Hh ); 
          Line = IIf( IsNull( CurLine ), Line, CurLine ); 
       } 
    } 
} 

Plot( Line, “”, colorBlueGrey, styleThick ); 
Plot( Close, Date()+ ” Close”, colorDefault, styleCandle );

I’ve tested this using the bar replay function and the pivot points seem anchored ie the line connecting high to low or vice versa doesn’t plot unless the minimum percentage swing is met. I’m just wondering if any of the hard core coders out there can verify this also. You can see from the chart below that the new upswing line is still to plot. There are lots of people out there saying you can’t use the zig zag function in trading systems and I agree in some cases because sometimes the last pivot moves, but I think using the above formula could be OK.

[Nick Radge]

We removed the zigzag from later versions of the 20% Flipper

Code:

_SECTION_BEGIN (“Parameters”);
//=================================================================================
//Parameters
//=================================================================================
function flipper_func(trail)
{
trailArray[ 0 ] = C[ 0 ];
for( i = 1; i < BarCount; i++ ) { prev = (round(100*(trailArray[ i - 1 ])))/100; if (C[ i ] > prev AND C[ i – 1 ] > prev)
{
trailArray[ i ] = Max(prev,0.8*H[ i ]);
}

else if (C[ i ] < prev AND C[ i - 1 ] < prev) { trailArray[ i ] = Min(prev,1.2*L[ i ]); } else if (C[ i ] == prev AND C[ i - 1] < prev) { trailArray[ i ] = 0.8*H[i]; } else if (C[ i ] == prev AND C[ i - 1 ] > prev)
{
trailArray[ i ] = 1.2*L[ i ];
}

else if (C[ i ] < prev AND C[ i] > 1.2*L [ i ])
{
trailArray[ i ] = 0.8*H[i];
}

else if (C[ i ] > prev AND C[ i] < 0.8*H [ i ]) { trailArray[ i ] = 1.2*L[i]; } else if (C[ i ] < prev AND C[ i - 1 ] > prev)
{
trailArray[ i ] = 1.2*L[ i ];
}

else if (C[ i ] > prev)
{
trailArray[ i ] = 0.8*H[ i ];
}

else if (C[ i ] < prev) { trailArray[ i ] = 1.2*L[ i ]; } else { trailArray[ i ] = 1.2*L[ i ]; } } return trailArray; } trail = 1; trailArray = flipper_func(trail);

[LeeDanello]

Thanks I’ll check out the difference

[LeeDanello]

ignore

[LeeDanello]

I’ve lost count of the number of Monte Carlo runs I’ve done on my system. Keep screwing them up for one reason or another because I forget to adjust the parameters, wrong start dates, periodicity etc etc. For example I adopted the following commissions given here for US stocks
https://www.amibroker.com/guide/w_commission.html.

I think I should have left the max amount as 0 instead of 0.5. Can anyone tell me otherwise?

Can this be hardcoded?

[Nick Radge]

Max amount should be 0

[LeeDanello]

Thought that just as I was about to send you my results.

[LeeDanello]

Ha, I’ve got over my 1st hurdle and have submitted a stress test. The system looks OK. Now that I’ve done it once, there’s no need to be as daunted doing it the 2nd time. This experience has shown me the method and now I will continue to tweak and see how the results fare.

[LeeDanello]

Finally started some simulated trading on IB. Going to test the system on the ASX and Russell 1000. Navigating around the menu the IB menus can be a bit daunting if you haven’t done in a while so I’ll be sticking to the classic interface.

[TrentRothall]

I think Nicks done a webinar on the IB platform, might be in the ‘Community”

[Author]

Posts