Daily breakout system

[SpiderX]

Hi ,

My personal gripe is that Steve's EA is a tat too bulky and complicated for someone else to modify, so it is nice to see an EA that is so simple.
Ironically, this time, i would be the one making things complicated.
I further modified Steve's version by putting in OrderReliable stuff and pipfactor stuff from the Shell.
Attached is my updated version.
There seems to be a missing "CountOpenTrades" in each cycle of the EA, which could result in a bug...hmmm
Have put it in the attached version

Cheers

[SteveHopwood]

Hi ,

My personal gripe is that Steve's EA is a tat too bulky and complicated for someone else to modify, so it is nice to see an EA that is so simple.
Ironically, this time, i would be the one making things complicated.
I further modified Steve's version by putting in OrderReliable stuff and pipfactor stuff from the Shell.
Attached is my updated version.
There seems to be a missing "CountOpenTrades" in each cycle of the EA, which could result in a bug...hmmm
Have put it in the attached version

Cheers

Thanks Spi, Great stuff. Can we have the crapq4 code, please.

[SpiderX]

Thanks Spi, Great stuff. Can we have the crapq4 code, please.

Sorry !
Posted the wrong file !

Cheers

[SteveHopwood]

Thanks Spi, Great stuff. Can we have the crapq4 code, please.

Sorry !
Posted the wrong file !

Cheers

Fantastic Spi, thanks. You even did the pips/points conversion thingy - a huge help.

V 1.01 with Siderx's adaptations and Stefan's set file hard coded into the inputs is in post 2.

[Stefan61200]

Hello Steve, hello Spi,

thanks a lot for modifying the EA. I will run both EAs in a parallel test. I'm curious about the result.

Stefan

[SteveHopwood]

V 1.02 is in post 2. Please read 'Revision history' for details.

Do me a favour coders please, and look over the Saturday/Sunday candle accommodation code in the block between lines 433 - 442 and check it is correct. I coded this after a long and hard day, a couple of glasses of wine and a rapidly decreasing level in the Scotch glass - 'nuff said?

[SpiderX]

Hi,

was wondering why this system was so familiar, until i saw this:
http://www.stevehopwoodforex.com/phpBB3 ... =12&t=3599

I requested Gary to rerun his simulation with these rules and the current set and he was kind enough to do so...apparently the results are worse than the original system in the link above.

Stefan please let us know how your test works out, but decidedly, i would not be running a demo on this.

Cheers

[garyfritz]

SpiderX, there might be more here than I thought.

The original "95 Pips" model was roughly breakeven, before costs.

This model (as described to me by Spidey, I didn't try to reverse-engineer the MQL) is nearly identical, except it enters 4 positions instead of 3, uses a 10pip SL instead of 30, and has larger TPs.

So the entry signal is identical, the TPs are larger, and the SLs are smaller. That's going to result in many more losses than the original 95 Pips model. The 10 pip SL is hit very frequently, and the large TPs are seldom hit -- 70-100 pip days are rare. The result is a pretty steady loser.

I poked at it a bit more and found it works a lot better with a larger SL. 20-25 pips actually works fairly well, before costs. With costs of $3 per 0.1 lost, it made about $1000 (on EURUSD) for the last year.

I tried adjusting the PT, setting the exits at 1, 2, 3, & 4 * PTval. Any value of PTval from about 10 to 30 pips worked well, with 20 maximizing the profits.

Then I looked at some other pairs. USDJPY actually works very well, making about $4500 in the last year. In fact most pairs were positive -- EURCHF and EURNZD are the only clear losers.

Many/most pairs have flattened out in the last 2-4 months. I don't know if that's a temporary problem or if something fundamental has changed. But if you had traded 0.1 lot on every pair for the last year, you would have made a nice profit -- *IF* your costs are low. The average trade size is pretty small. If your costs (spread/slip/commission per 0.1 lot) get much above $5-$7, there won't be much left.

So with a few tweaks it's much better than I expected. Still not a golden goose.

[garyfritz]

Time out, I made a mistake. I tested this on D1 bars, and by default TS doesn't look "inside" the bar to see when stops get hit. When you have both SL and PT in the same bar, it tends to get over-optimistic and assume you hit the PT first. I changed it to be more careful about checking the order stops got hit, and then the nice positive results all went bye-bye.

Which is what I expected. This model is way too simple-minded to get the results I was seeing.

[grat001]

In code on the line 639 o make small correction. Now i get "Alert" only if have any problem with modification order

Code: Select all

   if (err > ERR_NO_RESULT) {
      Print(Symbol()," SL/TP  order modify failed with error(",err,"): ",ErrorDescription(err));
      Alert(Symbol()," SL/TP  order modify failed with error(",err,"): ",ErrorDescription(err));
   }

Grat