stevehopwoodforex.com
https://www.stevehopwoodforex.com/phpBB3/
Print view

Zigzag Bollinger Band
https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=3237		 Page 9 of 28
Author:  nonlinear [ Sat Feb 08, 2014 8:32 pm ]
Post subject:  Re: Zigzag Bollinger Band
I have had this happen all too often. If you are going to test on Empty4, best to test on a patched version of Empty4 that in not connected to the internet. There are few floating around out there. PM me if you want a copy of the one I found (it is just for testing, on the old build -- need to find a new on for the new Empty4/5).
Author:  snailbeard [ Sat Feb 08, 2014 9:06 pm ]
Post subject:  Re: Zigzag Bollinger Band
nonlinear ยป Sat Feb 08, 2014 8:32 pm wrote:I have had this happen all too often. If you are going to test on Empty4, best to test on a patched version of Empty4 that in not connected to the internet. There are few floating around out there. PM me if you want a copy of the one I found (it is just for testing, on the old build -- need to find a new on for the new Empty4/5).
Well, it seems this was a blind upgrade to 600.
Is this essentially MQL5 in Empty4 clothing?

I've already made my mind up to move this code to open(ish) standard C#.Net.
The new MQ all singing all dancing sheep is another proprietary trap not to be falling into.

Rather than work around these issues it is better that I keep getting banged on the head, otherwise I might get too lazy to migrate since there is a significant effort to taking that step. :uff:
Author:  snailbeard [ Sat Feb 08, 2014 11:17 pm ]
Post subject:  100+ compiler errors
Just when you thought things could not get any worse:

Empty4 4 no longer compiles previously working mq4 scripts:
Author:  snailbeard [ Sat Feb 08, 2014 11:30 pm ]
Post subject:  Thank heavens for scite-mql IDE
scite-mql - still does the job
Author:  snailbeard [ Sun Feb 09, 2014 11:23 am ]
Post subject:  back on the tracks
Despite getting derailed yesterday by MetaQuotes disrespect for their end users, we need to get the old pip train back on the tracks.

Having had some historical data (usdcad) wiped away by the blind upgrade, I can now compile and run and test SB again.

I am investigating 60% failures of sells in contrast to 20% failures of buys.
Is it the market or a bug?
SB has multiple views of the market and I don't make use of everything that is calculated.
It could be a case of weighing a combination flags.
Due to the fickle repainting aspect of zigzags we can get false interpretations of this indicator, in the future I plan to make it work differently, for now I'll look at other data to complement the zigzags.

Example:

The following trade is entered in the wrong direction
1 2012.12.28 01:56 sell 1 0.30 1.03720 0.00000 0.00000 0.00 10000.00
2 2012.12.28 01:56 modify 1 0.30 1.03720 1.04024 1.03124 0.00 10000.00
3 2012.12.30 23:59 close at stop 1 0.30 1.03902 1.04024 1.03124 -54.60 9945.40

Resulting in an unnecessary loss
In the first chart we can see that there are plenty of lower highs and lower lows:
audusd-m15-zigzag-false-entry-28Dec214.png
However, there is strong evidence on other time frames of an emerging buy setup:
From the log file:
3M15 Bias: 0 , Priority: ;
3H4 Bias: 1 , Priority: i3H4_YCYL: +1
Y1Y2 Bias:1 YB0 YR+ Y><0

D1[Hslp: (PPD:-22), Lslp: (PPD:24)],
YD1[Hslp: (PPD:2.29), Lslp: (PPD:3.69)];
D[Y1D0]: 0, D0[R:11, C:6] ; , T~>:(Y)

M15[G: -1.00,W: 37.2,dL: 27.9,dH: 9.4];
H1[G: 0.00,W: 42.2,dL: 26.1,dH: 16.1];
H4[G: -4.00,W:213.5,dL: 27.9,dH:185.7];
D1[G: 16.00,W:261.1,dL: 20.8,dH:240.3];

Useful bits to look at here are:
H4 candles going up: 3H4 Bias: 1 , Priority: i3H4_YCYL: +1
Daily range moving up : YR+
D1 channel proximity: D1[G: 16.00,W:261.1,dL: 20.8,dH:240.3];
- especially: dL: 20.8 & W:261 - which tells us we are just 21 pips from a lower channel bounday of width 261 pips.
- meaning there could be a 200+ climb upwards from here!

This is easily understood from this H4 chart:
Author:  snailbeard [ Sun Feb 09, 2014 6:44 pm ]
Post subject:  D1 and H4 bias filters are working
The Daily and H4 candle bias have reduced short sell failures by about 25%.
Buy trade success remains high at about 80%.

But the disparity between buy trades and sell behaviour requires further investigation.
Only 6 sell trades in 12 months can only mean there is something wrong in the sell detection or sell filters code.

I plan to abandon the partial trades and replace them with stacking reentry trades instead.
This will help simplify profit locking and managing overlapping trades.

Somewhere along the way something went wrong. The profit factor was reaching 3.6 and now it has dropped down to below 2. The annual growth is also down from 20% to less than 10%.
:?
Author:  snailbeard [ Sun Feb 09, 2014 7:32 pm ]
Post subject:  Two Left Feet
In mid May 2012 there was a run easy sell trades.
I am pretty sure the old NanningBob EA would have hoovered up the pips on these swings.
SB is doing something strange, missing the easy money, then getting caught at a reversal.

This H4 chart offers treasure for the taking, until the reversal, NanningBob EA would also get clobbered at the reversal repeatedly which is why I put it to one side.
Author:  snailbeard [ Sun Feb 09, 2014 11:55 pm ]
Post subject:  oh no they've gone and done it again
Really you thought MetaCrap could not get any worse?

I am thinking of posting a Youtube video of this sabotage of the history data.

Watch MetaCrap rewrite you history data whenever it feels like it.

I deleted the rewritten history data.
Re-copied the original carefully prepared history data from a zip archive.
Restarted MetaCrap and yes it trashes my data again!
What the F****!!!!
:arrrg:
Author:  nonlinear [ Mon Feb 10, 2014 2:11 am ]
Post subject:  Re: Zigzag Bollinger Band
Fun, isn't it? :roll: Need to make your own bars...
Author:  snailbeard [ Mon Feb 10, 2014 7:12 am ]
Post subject:  Re: Zigzag Bollinger Band
Are you guys having fun with Build 600?

Here is some info from their release notes:
Terminal: Changed RateInfo history data storage format. RateInfo structure was presented as follows in the old version:
struct RateInfo
{
unsigned int ctm; // bar open date
double open; // Open price
double low; // Low price
double high; // High price
double close;// Close price
double vol; // volume
};
In the new format, RateInfo structure features fields for storing spread and trading volume:
//--- Standard quote presentation in the new terminal version
struct RateInfo
{
INT64 ctm; // open date and time
double open; // Open price (absolute value)
double high; // Low price
double low; // High price
double close; // Close price
UINT64 vol; // tick volume
INT32 spread; // spread
UINT64 real; // trade volume
};
Thus, if MQL4 programs contain DLLs for passing/accepting price data, the corresponding functions in the source codes should be rewritten and recompiled considering format changes to ensure proper operation.

Old EX4 applications and DLLs based on old RateInfo format will not work in the new terminal. Conversion to the new format is required.
Was it legal of MetaQuotes to rewrite data which I created and own without asking my permission
and not asking if I want to save the existing data first?

What is the UK law regarding the matter of an external party modifying files on a user's system without the user's knowledge or permission?
All times are UTC Page 9 of 28

Powered by phpBB® Forum Software © phpBB Limited