Big Mt4 Bug | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Bug with MT4 Historic Data on Simulator 11 replies

Strange MT4 output - mql bug ? 2 replies

MT4 Bug? 4 replies

A bug in MT4, any work arounds? 3 replies

Unusual Bug with DDE and Excel Scripts in MT4 8 replies

Attachments: Big Mt4 Bug
Exit Attachments

Big Mt4 Bug

  • #21
  • Feb 16, 2007 8:08pm Feb 16, 2007 8:08pm
Tesla - did U check my last charts ? Do You still think that everything is OK?
  • #22
  • Feb 16, 2007 9:09pm Feb 16, 2007 9:09pm
  •  tesla
  • | Joined Oct 2006 | Status: Friendly Neighborhood Programmer | 533 Posts
I think I know what your problem is... you're not importing your data to history center you're replacing the tick calculated data used for backtesting. Am I right?
  • #23
  • Feb 16, 2007 10:26pm Feb 16, 2007 10:26pm
For whatever it is worth - before I discovered Forex Factory i had purchased an EA that gave you a demo only version to try. I downloaded the demo which only allowed back testing. I did the my back testing to my amazement the thing worked as advertised... after purchasing the system, I set it up on a demo account and no matter what i tried the system would crash within two or three days. (crash = loose all the money in the account... not break pragmatically)

I could not figure out why the back testing could be so different from actual results... so i began to forward test in live then go back and back test it for the same period of time that i had forward tested it on...

Guess what - the back testing preformed miracles, while the forward testing could not walk a straight line to save it's own life.

I do not know how the back testing results could differ so radically from forward testing the exact same data, same time, same currency pair, same date range, same everything...

Unless... like you suggest - someone found a bug in the system that makes back testing look like a dream come true.
  • #24
  • Feb 22, 2007 1:49pm Feb 22, 2007 1:49pm
Quoting tesla
... you're not importing your data to history center you're replacing the tick calculated data used for backtesting. Am I right?
Sorry for a delay.... winter holidays

Yeah You are right, but You must know that before replacing fxt files and testing script I'd erased all history data from MT4 - i wanted to be sure that this data wont be included in calculation proccess.

After tests once again i check if history is empty to be sure that MT4 hadn't download or reconstructed MQ history files.
  • #25
  • Feb 22, 2007 4:21pm Feb 22, 2007 4:21pm
  •  tesla
  • | Joined Oct 2006 | Status: Friendly Neighborhood Programmer | 533 Posts
Quoting PabloPL
Sorry for a delay.... winter holidays

Yeah You are right, but You must know that before replacing fxt files and testing script I'd erased all history data from MT4 - i wanted to be sure that this data wont be included in calculation proccess.

After tests once again i check if history is empty to be sure that MT4 hadn't download or reconstructed MQ history files.
If you've deleted all history data then I'm stumped. The conclusion I came to was that the tick data you injected into the backtest process was an hour off from the data in your history center. Because the indicators fetch data from history center, everything would be thrown off by an hour.

Since you say that's not the case... I have no idea what else it could be.
  • #26
  • Last Post: Edited 5:13pm Feb 22, 2007 5:11pm | Edited 5:13pm
I think we've all had similar issues with backtesting. Backtest is awesome and forward testing is terrible. One thing I think I've found and I'd like for others to reveiw this. When I backtest from the Jan 1 2006, the EA makes millions. Suddenly around 11-4-2006 the results change and barely break even. I've gone over the trades using the Visualizer and it appears that for most of the year up to around 11-4 the backtester is NOT using spread. It is opening and closing trades using Bid alone. In effect testing as if spread were 0. Can someone else backtest and see if they see the same thing? Backtest, use the visualizer, and run it very slow and see if it is using spread or only Bid to backtest. I guess I noticed it because I was backtesting with GbpJpy and it's easier to see due to the larger spread that before around 11-4, there is no Ask line and my trades close when Bid hits my parameter when it should be waiting until Ask hits. Around 11-4, it starts using Ask and the line is then shown on the charts.

Quoting megaleads
For whatever it is worth - before I discovered Forex Factory i had purchased an EA that gave you a demo only version to try. I downloaded the demo which only allowed back testing. I did the my back testing to my amazement the thing worked as advertised... after purchasing the system, I set it up on a demo account and no matter what i tried the system would crash within two or three days. (crash = loose all the money in the account... not break pragmatically)

I could not figure out why the back testing could be so different from actual results... so i began to forward test in live then go back and back test it for the same period of time that i had forward tested it on...

Guess what - the back testing preformed miracles, while the forward testing could not walk a straight line to save it's own life.

I do not know how the back testing results could differ so radically from forward testing the exact same data, same time, same currency pair, same date range, same everything...

Unless... like you suggest - someone found a bug in the system that makes back testing look like a dream come true.
0 traders viewing now