Strategy Tester Produces Vastly Different Results

I tested my EA on two different laptops and was totally perplexed by the vastly different results. On my Acer running Vista, I get a profit of …… drum roll pls…… 215562.22! On my eeePC running Linux, I get only 46772.18….. about 5 times different.

I analysed the two result files and soon realised that the historical data used for testing differs on the two laptops, even though they were both downloaded from Alpari. I have no idea why the discrepancy… and which is a better reflection of my EA’s earning power…. Anyway, I reinstalled the data file on both laptops, using exactly the same procedure… and re-tested the EA… This time, the results were the same…. So, can I consider problem solved?

Maybe not… Cos I have yet to test the EURUSD pair, since I was unable to download the zip file from Alpari just now… seems like the server’s down…..

In addition, I remember my EA caused MT4 to crash when I tested it on a third laptop, running Windows XP…. but I’m going to ignore this at the moment, cause I highly suspect the crash is due to the limited virtual memory on that laptop… I’m not quite prepared to run the test on that 512MB RAM laptop again….. the frustration…

Ok, so, what to do now?

1) Download EURUSD from Alpari and test again….
2) Set up a forex and a VPS account to forward test the EA
3) Try running the EA on a forth laptop and see what happens. The EA is a bit ‘dangerous’, cause it produced too extremely difficult results…. It’s not robust… so how much can I trust the Backtesting results?????

Tested the 2 Hypothesis

Tested the 2 hypothesis mentioned in a previous post….

Among the two, filtering trades based on whether current open trade is profitable is a better option. Have yet to compare that against no filtering…

Tired now…. will resume testing tmr….

MetaTrader on eeePC

Yay!!! I finally managed to get MetaTrader and MetaEditor working on my 0.92kg eeePC…. which means I can now do my programming even when I am on the move…. spent three nights getting it working…. there are still some rough edges, but on the whole, MT4 on Linux rocks….!!!!!

For those who are interested, you need to first install Wine, which is a program that allows users to run Windows applications on Linux. Here’s a wiki with a step-by-step guide

http://wiki.eeeuser.com/howto:wine

After installing Wine, you can proceed to install MT4. Follow the steps here

http://www.moneyfanclub.com/forex-forum/3809-metatrader-4-linux.html

Note:

1. When installing MT4, instead of using the cp command to transfer the needed Fonts and DLLs to your linux drive_c, you can actually just navigate to  your Fonts  directory in Windows and copy and paste the files to Linux manually.
2. To get the latest DLLs (the needed ones are mfc42.dll and maybe mfc40.dll), you can download it at http://www.dll-files.com/dllindex/dll-files.shtml?mfc42
   Metatrader did not work when I used the dll from Vista, so I recommend downloading from the website above.
3. After installing MT4, you can click the .desktop file generated to execute MT4 or key the following command in a terminal
   wine ~/.wine/drive_c/Program\ Files/MetaTraderXXX/terminal.exe
   (Replace MetaTraderXXX with the name of your Metatrader Folder)
4. Click on F4 in MetaTrader to launch MetaEditor. When prompted to install Gecko, I would recommend that you choose ‘CANCEL’. This was the problem that caused me 2 sleepless nights. Whenever I choose to install Gecko, MetaEditor would crash. To get MetaEditor running, I finally had to give up on Gecko. The only drawback is that “Help” would not work.

That’s all…. If all goes well,  the whole installation should take less than  two hours…. If you encounter any problems, you can post your questions here and I’ll try to help…. but I’m no Linux expert, so no promise….

Improving my Expert Advisor

ok, here’s my ‘hypothesis’… gotta program them to check if they help me avoid losses:

1) Skip a BULL trade if the current bull trade is losing money (or if the current bear trade is making money – optional)

2) enter trade mentioned above only if pair closes above high by 5 pips within 4 (or 20) bars

OR

1) Skip a BULL trade if close is below MA(50).

2) enter trade mentioned above only if pair closes above MA(50) within 4 (or 20) bars

Test all EAs without Pyramiding, all ticks, M15, $3000 initial capital

Target:
Consistent Profit from 2004 to 2008.
More than 10k profit in 2008.
More than 3k profit per year from 2004 to 2007.

Focusing on my Expert Advisor

Been focusing on developing my Expert Advisor lately… basically still using the same concept – CANDLESTICKS. But I’m trying to improve the ‘hit rate’ and trying to increase the average winning profit…. When that is done, I’ll try to improve the money management aspect, so that I will have a lower drawdown…. has been pretty enjoying the programming, and learning more and more about MetaQuotes… I’m glad I love programming….

I realised I’ve been shifting from task to task… At times, I develop my EAs… then I got bored.. and swtich to writing codes for Amibroker… then I change my mind and decide that I should do some fundamental analysis on stocks so that I can do some value investing since we are in a bear market… then, I find a book at Borders on commodities and decide to learn more about it…. then I decide I want to learn more about Economics…. etc etc…. I guess I am trying to do too many things at the same time…. maybe because I’m afraid if I don’t do it now, I’ll miss the party (like maybe the strong trend in Forex is gonna end soon… or the Bear market is gonna end and stocks are going to get expensive…)

Anyway, I know I gotta prioritize.. else I’ll end up achieving nothing…. so for now, I’m gonna be concentrating on developing my EAs…. till I get one satisfactory one out, I’m not going to switch to another money making task…..

Bogie-NN-v8b

I backtested this EA, and man, it’s good. But I have yet to forward test it, since I have not set up my VPS. Anyway, I’m pretty inspired by the results. I did wonder if I should purchase the EA, but I can’t convince myself that I can make money the lazy way. What if I purchase the EA and it starts performing badly? Do I have the conviction to hang on to it? Or will I abandon it at the exact moment when it is about to turn profitable? How can I trade an EA confidently if I do not understand the logic behind it? Moreover, this EA is charged on a monthly subscription basis of $49.90. A bit steep, and it’s recurring.

But like I said, I’m inspired. So I refused to sleep till now (3.33am), trying to figure out a better way of entering trades. I’ll try this tomorrow:

1) Stock breaks previous high.
2) Previous high is above MA(50).
3) Incorporate a turn-sour limit to help deal with wipsaws.

Ok, gotta go meditate now and zzzzz…..

Believe in myself, the simpler the system, the better. Didn’t Nicholas Darvas make millions with a simple system…..