MetaTrader 4 Build 600 with Updated MQL4 Language and Market of Applications Released - page 46

Ex Ovo Omnia
2924
Ex Ovo Omnia  
disbellj:

MetaQuotes even gave us Capital Letters for New Folders. If New Folder Has Same Name As old folder, But New Folder Starts With A Capital Letter, Use <metatrader>/MQL4/(Capital Lettered Folder Name Which Is New Location).


In Windows, the file names and paths were never case sensitive, so the importance is really low.

Donald R Isbell Jr
224
Donald R Isbell Jr  
I was just talking about the mistakes that I have made, and have seen other people make, and is dawned on me that all the new folder names began with a capital letter, where old folder names didn't/don't. I pointed it out for those that don't know how to tell the new folders to use vs the old ones to not use.
Simon Gniadkowski
17237
Simon Gniadkowski  
disbellj:

"The docs have been updated, I think the F1 help in metaeditor has the most recently updated information, there is a section about the new version and compatability. I'm not sure it is the last word on all of that though"

Thanks SDC!

On another note:

Sorry RaptorUK, but blue panther is correct. MT4 99% backtesting can be searched on internet and then researched and achieved. At least I did and others I know. The quality of the data is everything. Using minute data with gaps vs tick data makes all the difference. Strategy Tester in MT4 gives 99% accuracy results with tick data made into .fxt files.


You should read what I wrote not what you think I wrote. You cannot do better than 90% modeling quality using MT4 alone . . . that is fact.
Simon Gniadkowski
17237
Simon Gniadkowski  
graziani:


well it try's and when it fails it continues with the prepared tick-data .fxt. So you can do better.

for everyone who uses strategies with small SL/TP tick data backtest is a must, as the gaps within M1 bars as well as the PA itself can change backtest results completely.

and this is one more shameful way MQ is trying to manipulate all traders, by denying this test capability in MT5. we will see if MQ will try to disable this testing possibility in future MT4 releases.

How do you prepare 99% modeling quality .fxt file using just MT4 ?
Ian Venner
2397
Ian Venner  
disbellj:
"SDC:

There is really only one type of strategy where tick modelling vs tick data could be reasonably debated. That is high frequency trading on 1 minute bars where trades are opened and closed on the same one minute bar. Such strategies are entirely dependant on the order at which ticks are received in the bar."

I must disagree with you SDC. When using tick data as I have been able to for free, it makes a difference no matter which mode you use.

Also, are you saying that all traders enter and exit trades on closed bar? Do you use open prices only in strategy tester? Does everybody else? And to get 90% quality it uses minute history with gaps?

I need better and faster backtested than what MT4 provides, and I don't see an alternative to test my EAs, and MT4 strategy tester takes HOURS (DAYS to optimize) just for one currency, forget multiple timeframe and multiple currency EAs which I write. Although I would love to sell an EA in the Market in MT4, I cannot with good conscience sell something I can't backtest.

Kindest regards,

Don


Well all I can say is I try to look at it logically, and logic tells me there can be no difference when all the modelled ticks do is fill in the gaps between OHLC. When someone explains to me with clear logic, why there is a difference and points out something I had not taken into account I might change my mind. The trouble is no one can ever tell me why there is a difference, they always say there just is. It is almost like a religeous belief. The way logic dictates it to me is like this, when a trade is closed on a subsequent bar, as long as the OHLC is accurate it makes no difference how the prices in between got there as long as they are there, tick data or modelled makes no difference.

There will obviously be a different set of results when comparing a set of tick data chart results with any other chart and vice versa, there is always a difference when using two different sets of test data. Sometimes very big differences.

In my earlier post when I mentioned 1 minute bar trading, no, I didnt mean trading on a closed bar price, I meant closing on a subsequent bar in other words, not the same 1 minute bar the trade was opened on. No I dont always use open prices, I rarely do, often I use prices that occur somewhere within the bar.

I would like to do a test, but I know I dont have the time for it right now. The test would be as this.

  • Take a 1 year set of tick data
  • Extract from it the 1 minute OHLC's
  • Backtest a test EA on the tick data
  • Backtest the same EA on those 1 minute OHLC's so the intermediate ticks are modelled
  • Compare the results.
Ubzen
5299
Ubzen  

This idea that 99% modeling quality is a marketing gimmick used by tick_software vendors.

If someone used Random-data and received 99% quality does that make their product better?

If someone knows the OHLC, they can decide to Buy@Low and Sell@High regardless of modeling quality.

If we have a choice of Importing Real-Ticks, then the real issue would be more transparent.

Everyone benefits with choices, Import-Ticks or Don't the choice would be yours.

NoTomatoes
40
NoTomatoes  

I do not know if this has been reported yet.

DoubleToStr() seems to have a intermittend memory corruption inside.

The code
<Image removed by moderator, Please use SRC to insert code>

produces this output:


Note that the orderModifyReliable() still receives the right SL and that the first call to DoubleToStr() made no difference.

Ex Ovo Omnia
2924
Ex Ovo Omnia  
ubzen:

This idea that 99% modeling quality is a marketing gimmick used by tick_software vendors.

If someone used Random-data and received 99% quality does that make their product better?

If someone knows the OHLC, they can decide to Buy@Low and Sell@High regardless of modeling quality.

If we have a choice of Importing Real-Ticks, then the real issue would be more transparent.

Everyone benefits with choices, Import-Ticks or Don't the choice would be yours.


I have no experience with the custom fxt files, so a question came to my mind. Could the fxt file contain different open/close prices than the hst file? It is common problem with the Renko charts, that the open/close price are pure artificial values, and sometimes they are affecting even the candle high or low.
Ubzen
5299
Ubzen  
Ovo: I have no experience with the custom fxt files, so a question came to my mind. Could the fxt file contain different open/close prices than the hst file? It is common problem with the Renko charts, that the open/close price are pure artificial values, and sometimes they are affecting even the candle high or low.

Have you seen this thread:https://www.mql5.com/en/forum/141075. I haven't gone through tick data back-testing.

If the fxt goes through the normal process then it'll match the hst because the fxt are generated from the hst.

If it goes through a 3rd party tool then its anybody's guess what its capable of.

Milanoss
3
Milanoss  

Here is bug in MetaTrader 4 build 610:

Steps to reproduce:

  1. Open any chart
  2. Edit example indicator SimplePanel.mq4
  3. Add this line into code:
 extern bool test=true;

  • Insert indicator into chart
  • Close chart
  • Open any chart
  • Insert indicator into chart but change value of extern "test" property
  • Check logs. You will see something like:

2014.02.17 22:43:50.983 SimplePanel EURUSD,H1: CAppDialog: find subwindow error

Now you cannot add this indicator into any chart. Only solution is to remove following line from terminal.ini:

Custom Indicator - SimplePanel=2

Any comment? I spent 4 days on this but without fix it is not possible to use CAppDialog.