Big problem I can't solve

Trading Systems Expert Advisors and Automated Trading
1

I have written an EA that I have spent a few hundred hours back testing with various settings. Because of these tests, I think I have come up with very specific settings that seem to work unbelievably.

Most of my testing has been from January 1st 2003 to December 31st 2005.

In this time, the EA masses over US$9M profit starting with just US$200…

MA Tester 2003-01-01 to 2005-31-12.jpg700×171 19.2 KB

I know that back testing results aren’t something to use, but those figures are mind blowing to me.

Because I know that the forex market can change a lot over the months and years, I tested the EA from Jan 2012 on wards, with very different results. It looses everything in the account in a pretty quick time!

My questions…

  1. Is back testing something I can actually use as an indication of the possible profits/losses in real trading?

  2. Is it really possible to make the profits I am seeing in the 2003-2005 back testing?

  3. What happened in the market to make the massive difference in profits after 2005? (I’m pretty sure that no one will be able to give me a really accurate answer to this one)

The basics of my EA… (1 Min time frame)

  1. Works on SMA of 30 Periods and 0 shift.

  2. If the Bid Price is 130 pips above the SMA place a Sell order (130 TP and 650 SL)

  3. If the Bid Price is 130 pips below the SMA place a Buy order (130 TP and 650 SL)

  4. Only open one order at a time, so if there is an un-closed order, there are no others placed.

2

  1. Backtesting is useful to understand how your code works and how it reacts in market conditions so that you know how it should work in the live market. I think if it loses in back testing, it will lose in live market.
    You have to appreciate that the live market is so much different than back testing in respect to spreads and requotes.

  2. In order to answer that, someone would have to detailed knowledge of market conditions 10 years ago, thorough knowledge of your EA and accurate prediction of the future. So no its not possible.

3

One thing that occurred to me in regards to your back testing results is the data quality. I know most historical data is full of gaps and spikes and it doesn’t take a lot of them to skew your results. Did you do any QA on the data to spot gaps and spikes?

4

Why don’t you just let it run in real-time on a demo and see how it works? Like CodeMeister commented, your historical data could be faulty.
:slight_smile:

5

I second Sweet Pip’s suggestion. Would it be possible to see the stats (in a Google spreadsheet or Excel file) so we can get more insights on it?

6

That is exactly what I have started. As a result, I immediately saw a few bugs that I fixed.

7

These are the results from my back testing…

MA Trader Results.jpg700×511 49.1 KB

How do I attach a spreadsheet?

8

Some helpful insights here as well:

FOREX Statistical Research Center/Backtesting & Data Mining

Cheers

9

my suggestion is to just do some forward testing and see where your at after a few months :slight_smile:

in fact I am sure a lot of us wouldn’t mind tracking this so please post a myfxbook link if you can!

10

[QUOTE=“Matrix Fx;529693”]my suggestion is to just do some forward testing and see where your at after a few months :slight_smile:

in fact I am sure a lot of us wouldn’t mind tracking this so please post a myfxbook link if you can![/QUOTE]

There’s your fifth post… The now you can post your link to your PAMM… waits with bated breath