Question -> 1k to 10Millions in 15 months
Z
I've had a couple of those!
The 'golden period' can be a few weeks or many months then... SPLAT :(
To generalise, this points to a system that is just too... simple, one with a Profit Factor that was a bit low.
IMHO, any EA with a PF under 2.0 over a long period is asking for trouble during the more unfavourable periods.
Try & develop or find a system that would or is working as a manual system then code it up...
..
> going crazy trying to figure out what's so special in period 2004-2006
By all means do a post mortem - it may be one simple thing but its unlikely
Dont spend too much time or self-blame on it, just move on as soon as you can
..
NB - you can experience the same effect when moving from MT data to live broker data
Sometimes just moving from a brokers demo to a live account will cause a marginal EA to come unglued...
Good Luck
-BB-
Z
I've had a couple of those!
The 'golden period' can be a few weeks or many months then... SPLAT :(
To generalise, this points to a system that is just too... simple, one with a Profit Factor that was a bit low.
IMHO, any EA with a PF under 2.0 over a long period is asking for trouble during the more unfavourable periods.
Hi BarrowBoy!
Tnx for the answer. I totally agree on your point.
I have to amend what I wrote about the timeframe of simulation... from 01/01/2004 to 30/03/2006 it's in fact 27 months and not 15!!! :)
...this makes the thing even more curious cause my EA suddenly stops producing profit after more than 2 years of awesome results.
PF over 2 years is 1.91 and Expected Payoff 351.8... note that these figures are pretty meaningful as number of trades is 31266.
As written before % of wins is 85.18%.
I find pretty strange that a strategy whit such robust history can suddenly stop working. I think there must some change in Market's behavior... something I cant see by naked eye but maybe a skilled analyst can find out or recall.
Just for the records, in the following you can see the report.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
| Symbol | EURUSD (Euro vs US Dollar) | ||||
| Period | 5 Minutes (M5) 2004.01.01 01:40 - 2006.03.29 23:55 (2004.01.01 - 2006.03.30) | ||||
| Model | Every tick (the most precise method based on all available least timeframes) | ||||
| Parameters | minLot=1; MM=2; Risk=1; PatternIN=220; Aperture=180; tppoint=30; StopLoss=20; SafeProfit=3; mpbo=5; MaxTrades=3; MagicNumber=789; | ||||
| Bars in test | 169065 | Ticks modelled | 5062166 | Modelling quality | 90.00% |
| Mismatched charts errors | 11 | ||||
| Initial deposit | 1000.00 | ||||
| Total net profit | 11005142.31 | Gross profit | 23131732.60 | Gross loss | -12126590.29 |
| Profit factor | 1.91 | Expected payoff | 351.98 | ||
| Absolute drawdown | 548.01 | Maximal drawdown | 162448.71 (1.87%) | Relative drawdown | 58.37% (633.75) |
| Total trades | 31266 | Short positions (won %) | 0 (0.00%) | Long positions (won %) | 31266 (85.18%) |
| Profit trades (% of total) | 26633 (85.18%) | Loss trades (% of total) | 4633 (14.82%) | ||
| Largest | profit trade | 8235.76 | loss trade | -17367.90 | |
| Average | profit trade | 868.54 | loss trade | -2617.44 | |
| Maximum | consecutive wins (profit in money) | 89 (120182.33) | consecutive losses (loss in money) | 18 (-6799.41) | |
| Maximal | consecutive profit (count of wins) | 120182.33 (89) | consecutive loss (count of losses) | -89972.51 (17) | |
| Average | consecutive wins | 12 | consecutive losses | 2 | |
Looking at the strategy tester report:
Your EA buys low and sells high only.
It places only long trades. No short trades are ever placed. All 31266 trades were buys.
The behaviour of the market for the time period tested is conducive to such a trading strategy.
There have been no major pull backs between April 2006 to August 2007. You can confirm this for yourself by pulling up a weekly chart.
Test your EA for the time period between July 2007 to August 2008 and report the test results to prove my point.
Post back
jab
Looking at the strategy tester report:
Your EA buys low and sells high only.
It places only long trades. No short trades are ever placed. All 31266 trades were buys.
The behaviour of the market for the time period tested is conducive to such a trading strategy.
There have been no major pull backs between April 2006 to August 2007. You can confirm this for yourself by pulling up a weekly chart.
Test your EA for the time period between July 2007 to August 2008 and report the test results to prove my point.
Hi Kengen,
tnx for reply.
...sorry to disappoint U but the EA doesnt simply buy low and close trades high.
On the contrary, although it is not a scalper the profit on a single trade is often less than 10 pips. You can see that form AVGwin and AVGloss knowing that StopLoss is set at 20 pips.
Yes, the EA only buys and this is cause I found out that the particular indicator I created is not efficient to individuate entry and exit points for sell... so I simply decided to avoid selling.
On the other side the buy strategy is performing so good that I dont miss selling at all :)
I'm following your suggestion and looking at the weekly chart... interesting to see that from November 2005 the market kept raising till today with no big fluctuation. Strange is that I start losing profit from October 2006 while before it works perfect [ yes, I tested month by month... crazy, uh? ;) ]
I see a remarkable fluctuation between September 2004 and July 2005 but this does not affect at all my system as it works on small periods (M5 and trades not longer than 24 hrs).
...sill I find no evidence in changes on the market after 10/2006 to justify why the EA should stop working after 27 months of awesome results.
As BarrowBoy said it happens that EA performs good for a certain period the they drift... I've seen that before... but 27 months are IMO too long period to think it was just a fortunate series of trades... again, we are talking about 26633 trades won on 31266.
Maybe there is something I miss, maybe it's just the way it has to be :)
I'm running now an optimization on EURGBP to test the EA on that symbol on the same period (2004-2006) and see if I get matching results.
I'll keep you guys informed;)
Cheers,
Zyp
Ok,
now I really start to freak out.
I finished testing EURGBP. No wonder the optimizer suggested almost the same settings than on EURUSD.
The result on 01/01/2004-30/03/2006???
...well... I had to stop the simulation after it reached 30 Million$... I wanted just to prove a point, we all know that trading with those figures is impossible in real life.
Anyway, I made sure that the EA was working even in EURGPB. Then I tested from 09/2006 up to today. I stopped it after a few months of simulation... see the result in the picture.
What to say???
U can clearly see October 2006 coming and spoiling the game... after that it's all crap. Tested in EURUSD and EURGBP.
Coincidence again???
I start to believe something has changed after mid 2006... which btw matches the period of the first championship... is it us??? :)
...are we trading live so much to affect Market's behavior??? :)
Zyp
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hello everybody.
Here is the thing: after a few months of studies and coding hard I finished my last EA and it shows impressive results on the tested period from 01/01/2004 to 30/03/2006.
...as I wrote in the header 1000$ reach 10.2 Million$ with 85.18% of wins. If you think it's good (damn if it's good) U'll be surprised to hear that StopLoss is set just at 20 pips!!!
Where is the problem???
...problem is that after April 2006 something changed in the Market and the EA does not give profit anymore... actually it is kind of useless crap.
I've been comparing Market's graphs but I do not see any relevant difference in Market's "behavior".
I'm kind of going crazy trying to figure out what's so special in period 2004-2006 to make my EA work super and what's so wrong aftr 2006 to make my it look like crap.
Any suggestion appreciated :)
Zyp