Automated trading results - Looking for critiquing | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Automated trading vs Manual Trading 28 replies

Automated Trading & ONLY Auto Trading 3 replies

Automated or semi automated trading? 3 replies

YOUR results with live, fully-automated EAs?? 26 replies

Looking for a broker supporting automated trading 0 replies

Attachments: Automated trading results - Looking for critiquing
Exit Attachments

Automated trading results - Looking for critiquing

  • #1
  • First Post: Aug 10, 2011 3:18am Aug 10, 2011 3:18am
So, I have been testing my EA and I feel the results are fabulous. With them being this good, I am a little leery about them, and would like to get veteran traders opinions if they would "go for it" and start trading live with it. I am relatively new to the EA arena, and would assume there are some stats that may or may not look good, that would cause a person who is more wise to refrain from using their own hard earned money. Testing was done for the period between Jan 2010 and present. Any comments/suggestions/critique would be appreciated.

Cory
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Automated trading results - looking for critiquing.PNG Size: 25 KB
Click to Enlarge Name: Automated trading results - chart.PNG Size: 16 KB
  • #2
  • Aug 10, 2011 6:09am Aug 10, 2011 6:09am
I would check it again with real tick data, (of course it depends on how your ea trades but results may look completly different with it) , here is a how to

http://eareview.net/tick-data
  • #3
  • Aug 10, 2011 6:13am Aug 10, 2011 6:13am
DD is too high imo

Is this the MA in sync EA? Saw your posts @ forex-tsd yesterday, small world
  • #4
  • Aug 10, 2011 8:23am Aug 10, 2011 8:23am
blutkehlchen - I read that post at Birt's...the article and instructions were well written and detailed, though it seems pretty complicated. I will consider it in the near term. With that being said, can results change that drastically by going the other 9% to reach 99% accuracy. For example, can it go from 100k down to 20-30k?


wickedwin - no, its a different one with multiple MACD's and a non-lag ma. the lowest I have been able to get this down to is 18% and a profit of around 55k.
  • #5
  • Aug 10, 2011 8:39am Aug 10, 2011 8:39am
Quoting bigsteely101
So, I have been testing my EA and I feel the results are fabulous. With them being this good, I am a little leery about them, and would like to get veteran traders opinions if they would "go for it" and start trading live with it. I am relatively new to the EA arena, and would assume there are some stats that may or may not look good, that would cause a person who is more wise to refrain from using their own hard earned money. Testing was done for the period between Jan 2010 and present. Any comments/suggestions/critique would be appreciated.

Cory...
I suggest you to remove the money management algo in backtesting so that you can evaluate real performance of your system.
  • #6
  • Aug 10, 2011 8:45am Aug 10, 2011 8:45am
What do you mean? The money management is part of the strategy
  • #7
  • Aug 10, 2011 8:57am Aug 10, 2011 8:57am
Quoting bigsteely101
What do you mean? The money management is part of the strategy
if your ea is not profitable without money management then it is useless. You are not going to make it in live trading.
  • #8
  • Aug 10, 2011 9:11am Aug 10, 2011 9:11am
ah ha, can you elaborate? Why would money management not be considered and properly activated in automated trading?
  • #9
  • Aug 10, 2011 9:26am Aug 10, 2011 9:26am
Quoting bigsteely101
ah ha, can you elaborate? Why would money management not be considered and properly activated in automated trading?
I meant you shouldnt activate the money management during backtesting because the system performance can be easily skewed. Just look at your equity chart, the last 30 trades made more than 50% of your total profit. I am pretty sure that your profit factor will drop significantly without the last 30 trades.
  • #10
  • Aug 10, 2011 9:32am Aug 10, 2011 9:32am
so you are saying that my testing should be run clean; without automatically adjusting position size relative to the equity balance.
  • #11
  • Aug 10, 2011 9:32am Aug 10, 2011 9:32am
The results are realistic, sample size is good. Was there optimization done?
Does it include slippage and spreads?
The largest loss is greater than the largest win, do you know why?

My opinion is test the strategy with both money management and no money management. But keep in mind there are pitfalls with just testing on a 1 contract basis in some instances.


Are you testing on a portfolio or just 1 currency?
I have not been using the tester in MetaTrader for quite sometime, but if I recall correctly there is no portfolio testing capability

I don't agree the last 30 trades can justify the money management factor.
It could be that his "edge" manifested for the last 30 trades
  • #12
  • Aug 10, 2011 9:35am Aug 10, 2011 9:35am
Quoting bigsteely101
so you are saying that my testing should be run clean; without automatically adjusting position size relative to the equity balance.
Yes, because money management cannot turn a losing system into profitable in long run.
  • #13
  • Aug 10, 2011 9:48am Aug 10, 2011 9:48am
Quoting bigsteely101
so you are saying that my testing should be run clean; without automatically adjusting position size relative to the equity balance.
basically whatever your strat is based on like you "take longs when it breaks the high".. check the probabilities that price goes to your TP level, probability it goes to SL.. figure out all that stuff
  • #14
  • Aug 10, 2011 10:08am Aug 10, 2011 10:08am
Quoting ash234
The results are realistic, sample size is good. Was there optimization done?
Does it include slippage and spreads?
The largest loss is greater than the largest win, do you know why?

My opinion is test the strategy with both money management and no money management. But keep in mind there are pitfalls with just testing on a 1 contract basis in some instances.


Are you testing on a portfolio or just 1 currency?
I have not been using the tester in MetaTrader for quite sometime, but if I recall correctly there is no portfolio testing capability

I...

I did not adjust optimization settings, they remain the default in the tester; "balance". Regarding slippage and spreads, I have it set at slippage of 2 pips, but I am not sure if it accounts for spreads (nothing built into the EA, maybe in strategy tester).

I am just testing on the Eur/Usd pair; I will try to expand on my data to further test. Also, you brought up a good point about largest loss versus largest gain, I will have to dig deeper into that to find out what happened.

I am currently back-testing with money mngt off, I will update
  • #15
  • Aug 10, 2011 10:33am Aug 10, 2011 10:33am
Oh you mean this EA was bought off the shelf?
Is your test period "Out of Sample" aka after the product has been made available for sale and the settings not tampered with after?
If I am a borderline criminal I would have overfitted the EA to the in sample data and would not have told you which period I optimized the EA for.
  • #16
  • Aug 10, 2011 10:36am Aug 10, 2011 10:36am
Quoting ash234
Oh you mean this EA was bought off the shelf?
Is your test period "Out of Sample" aka after the product has been made available for sale and the settings not tampered with after?
If I am a borderline criminal I would have overfitted the EA to the in sample data and would not have told you which period I optimized the EA for.

This is a EA I had built by a soon to be brother-in-law. The test was done using Dukascopy data. I have not played around with optimization settings as of yet.
  • #17
  • Aug 10, 2011 10:40am Aug 10, 2011 10:40am
Here are the updated results with no money management incorporated as suggested. 30k lot size
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Automated trading results - revised with no MM - 30k lot size.PNG Size: 19 KB
Click to Enlarge Name: Automated trading results - chart - no MM.PNG Size: 16 KB
  • #18
  • Aug 10, 2011 10:44am Aug 10, 2011 10:44am
Sorry I didn't expand the view. Here is a good shot.
Attached Image (click to enlarge)
Click to Enlarge Name: Automated trading results - revised with no MM - 30k lot size.PNG Size: 25 KB
  • #19
  • Aug 10, 2011 1:35pm Aug 10, 2011 1:35pm
In my opinion, your trading test period is too short. Try it with bigger period, at least 10 years. This will show you how your EA behave in different market condition. Also test it with other pairs.

OOT: How did you run the test in Dukascopy? Did you just opened Dukascopy account and ran the test there or with other way?
  • #20
  • Aug 10, 2011 1:55pm Aug 10, 2011 1:55pm
Quoting Special One
In my opinion, your trading test period is too short. Try it with bigger period, at least 10 years. This will show you how your EA behave in different market condition. Also test it with other pairs.

OOT: How did you run the test in Dukascopy? Did you just opened Dukascopy account and ran the test there or with other way?
If you've done your own backtesting, you're well aware that good quality historical data is tough to come by. I have to disagree with the need for ten years of backtesting...a trend is a trend. There will be bumps along the road but it is not going down, or sideways, maybe I could extend it another year or so though. Within the past 1.5 years I believe the market has thrown just about every pattern you can think of at us. Not to mention, I would have to say the past year could be argued that it was one of the most volatile, so if it could do it during these times it will likely do well in trending times as well.

Regarding Dukascopy account I have a live account with them and I got the MT4 platform through my Introducing Agent. You can also open a demo account through Fair Trading technologies. http://www.fairtradingtech.com/ they have the Dukascopy data feed.
0 traders viewing now