Welcome back Forex fans.Its a magical time to live in the Northeast USA. The air is cool and crispy. The squirrels are burying their nuts, and all woodland creatures and preparing for the winter ahead. The light takes on a special property where even the common place is incredible. Words can't describe the experience. Alright, back to business..
It was a fine week in the land of Forex Robots and things are moving along right on schedule.
First up, the IamFX account that was supposed to get moved to Forex.com was liquidated and they sent me check for $2054.33. This was a return of 2.7% on my deposit of $2000 after just 2 weeks of trading with my favorite 2 scalpers. Consistent with my 1% a week mantra and i'm happy with that.
Next up, Megadroid Live had a fine week in its new home with Forex.com US and picked up 2.59% for the week. This was milestone since the account topped +30% for the year for the first time. This system has now had 25 consecutive winning trades and is way overdue from some drawdown. Megadroid Demo gained +3.5% in a similar fashion.
Next up is Gods Gift Duo which is my first live account trading this system trading on EUR/USD and GBP/USD on a $1000 account with a risk setting of 1. I want to trade this account at a higher risk level, but i'm going to finish out 2010 at the 1% risk level to see how it goes. The account picked up just under 1% with an open sell in GBP/USD.
Next up is demo account Atinalla #1 which gave up about 0.9% for the week. Also in the same category was Atinalla #3 which I kicked off last week in demo mode. This system lost -2.12% for the week and i'm not impressed with the way its going. I back-tested 3 of the 4 symbols, and was profitable on only 1 of the 3 YTD. Also this system seems to loose slightly more than it wins, with a win/loss ratio of about even. You can't make much of a judgment after 1 week, but so far, i'm not impressed. Good thing its demo money.
Finally the Q-Diamond demo accounts were idle all week. With the switch to Forex.com US, they switched to a new symbol format for demo accounts: EURUSDpro. The new suffix defeats the Q-Diamonds and I e-mailed the vendor to fix this this last weekend, but got no reply back. This system/vendor is making things more difficult than necessary. They could just pick off the left 6 characters of the symbol and be done with it, but for some reason they choose to make this harder that it needs to be. I'll follow up and circle back.
Have a great weekend all.
Hello Chris,
ReplyDeleteThank you for the update :o) The Asirikuy accounts currently running Atinalla No.3 had a profitable week, along some of the accounts running separate instances of the systems within the portfolio. I also ran the backtests on the new data for the year and obtained that 3 out of the 4 systems have been profitable YTD. You might want to check your settings or send me an email (including your broker) with your current ones so that I can give them a look :o). Thanks again for the update Chris,
Best Regards,
Daniel
Daniel-
ReplyDeleteThanks. I have Atinalla #3 running on Forex.com US. I have the Sunday Candle set to true, and I have daily charts open for each of the pairs, with the WithoutSunday EA loaded.
I'll send you all my settings and an account statement via e-mail. Thanks again,
Chris
Hello Chris,
ReplyDeleteI just answered the email you sent me :o) It turns out that most of your settings were not correct because you didn't take into account the fact that Forex.com US is a GMT 0 and not GMT +2 broker (like the backtests) and most of the instances you were trading have time sensitive information which was very wrongly assigned. You also have the wrong settings as you did not choose the NFA compliant versions :S.
Well no harm done :o) This is just a matter of correcting the settings and restarting your demo account. I am glad I could help with this :o)
Keep up the great work Chris !
Best Regards,
Daniel
Daniel-
ReplyDeleteThanks for the follow-up and quick reply. We are in synch on the settings for all pairs. I'll restart the demo and we'll have some results next week. Regards,
Chris
Hello Chris,
ReplyDeleteIt's a pleasure to help :o) Make sure you also repeat the backtests and watch the other Atinalla No.3 accounts to compare your results,
Best Regards,
Daniel
Thanks for the interesting discussion around Atinalla #3, especially the correction about the GMT offset. There are several commercial EA's being marketed which require accurate GMT offset settings, because of the requirement that the EA trade within a fairly narrow window of time. What I am wondering is, can an EA be considered as truly robust if it requires such a narrow time window to make successful trades? And if so, doesn't that really seriously compromise the future potential of that EA because if very specific conditions (around time and trading conditions based on that time window) are necessary for success, then flooding the market with trades in that time period could easily change a winning strategy into break even or worse?
ReplyDeleteThe other question I have for both you, Chris, and Daniel, is this: I note that Daniel has obtained some very tight and accurate historical data to get around the Alpari UK data that had become tainted somehow, and I also note that significant developers like Funyoo have now gone to Dukascopy data which ostensibly provides 99% accuracy, and in light of this, why wouldn't you both, but particularly Daniel, not use this apparently better data for backtesting your EA's?
Thanks for all your good work gentlemen, you are advancing the cause for us all.
Redford.
Redford-
ReplyDeleteThanks for the comment.
The Megadroid for one trades in a specific time of day. I think time-of-day based trading mades sense generally since most traders are going to be active during their business day, and therefore meaningful moves tend to happen during the active part of the day. Conversely, after hours moves tend to retrace the move that was made during the active part of the day.
The Fish Forex system for another example, trades EUR/USD away from the price at 3PM EST looking for a move back to the price level at 3PM EST. And it turns out price does return to the 3PM level something like 80% of the time.
Regarding how time of day is used with WA No 5, i'll let DF answer since i'm not completely sure what the logic is there. I'll also ask him to comment on your question regarding the Dukaskopy data.
Take care buddy,
Chris
Hello Redford,
ReplyDeleteThe concerns you raise are actually well founded but not necessarily correct. Atipaq (Watukushay No.5) doesn't actually "trade" within a specific time window but it uses a time parameter to define a "breakout box" which is then used to set pending orders that are the actual entry points for both long and short trades. The system needs a specific hour because it relies on certain market hours to behave in a certain way to define a valid box, this makes sense as different sessions have different volume, price patterns and Atipaq bases its profitability on a very simple analysis of these patterns. I do not believe that this makes the system any less robust and adequate "timing" is just part of the inefficiency exploited by the system (due to the fact that it is based in session-pattern behavior, so to speak).
Regarding the 99% Dukascopy data, there are several reasons why I have considered not to use this data.
1. It is only available from 2007 so I can only get 4 instead of the 11 backtesting years I get with the data now in Asirikuy.
2. The strategy tester can only have Bid data loaded so - since there is no actual Bid/Ask stream- a large benefit of having real tick data (which is accurate real evolving spreads) is lost.
3. The gain in accuracy for Asirikuy systems is very low since they trade with very large profit and loss targets and with signals on higher time frames (hourly and above).
4. The process is VERY cumbersome and only one year can be backtested at a time.
As you see the problem here is that we are trading a slight gain in accuracy for a far more limited test which is ten times harder to do and does not even fully benefit from the real tick data Bid/Ask streams. The current data offers more than satisfactory simulations for Asirikuy systems and it is likely that any increase in accuracy (for these systems) through the use of tick data is within the limits of broker dependency (so even if more accurate simulations would be made the differences between brokers would be more important).
Right now the data downloadable within Asirikuy fulfills everything I wanted, full one minute data with no holes from 2000 till oct 2010, more than enough for adequately designed systems that do not depend on one minute interpolation errors and are only slightly affected by execution variables such as slippage.
I hope this has answered your questions ! :o)
Best Regards,
Daniel
Daniel-
ReplyDeleteThanks very much for the comments and further input, I appreciate it!
Chris
Daniel, you made a solid case for not using Dukascopy data vs the clean 1 minute data that you've been able to obtain, thanks.
ReplyDeleteAnd thanks to you Chris, for keeping this interesting blog going.