Sunday, February 28, 2010
Cheers to all Canadians including my buddy Sir Redford - who probably had nothing do to with the games. But I don't care, anybody who makes his blogger icon his national flag has something to be proud of.
I have lot cooking under the covers which hasn't been showing up this blog lately. So if you think this blog has been a snoozefest lately - I dont blame you. But stay tuned because some day, somehow, it will all come out. But for now the words of the Canada national anthem....
Our home and native land!
True patriot love in all thy sons command.
With glowing hearts we see thee rise,
The True North strong and free!
From far and wide, O Canada,
We stand on guard for thee.
God keep our land glorious and free!
O Canada, we stand on guard for thee.
O Canada, we stand on guard for thee.
Saturday, February 27, 2010
Sir Redford suggested that I take a look at the differences between the live and demo Megadroid results. This lead to some interesting findings.
I was overstating the results of the Megadroid demo account by a few percent since my results accidentally included the last 2 weeks of 2009.
The Demo account traded 26 times, and the live account traded 17 times. That's almost a perfect 0.65 Fib Ratio!
The demo account is on FXDD and the live account is on Forex.com UK so different brokers might account for some differences in results.
Anyway, I saw several times when (right at about 4PM), the brokers widen the spread to 4 pips. This activates the spread-monitoring technology and puts Megadroid out of business for about 45 minutes or so. I've seen the brokers do this on Monday and Tuesday, but allow trading on Wednesday. Its almost like they know they have to give you something, or you won't remain a customer.
The net result is that the live Droid only makes about 65% of what the demo droid makes. That coupled with the infrequent trades using Droid, makes the results flatish versus what they could be. So the Droid algorithm is great, but you don't get enough good changes to use it.
What's the answer? Apply Megadroid to multiple pairs and you get EAKAIN.
EAKAIN is currently #2 on the Forex Peace Army Expert tests and was up a whopping 14% for the week. This one is really spectacular.
And the #1 robot on the list is ......... the Fusion EA Bot from the guys over at the Forex World Cup. This one was up 18% for the week! This is what I call an EA and leaves Megadroid in the dust!
It's clear that I can do a whole lot better than 1% a week with the super EAs.
This is getting good.
Friday, February 19, 2010
The US Fed increased the discount rate 0.25% to 0.75% in a surprise move at about 4:30 PM on Thursday Feb 18th.
This was during the Megadroid trading window and Megadroid had just gone long the EUR/USD which quickly shed 42 pips in a few minutes time. Apparently, the Fed didn't get the memo that you're not supposed to change interest rates between 4-7PM EST when the scalper robots trade!
Just goes to show you that in trading anything can happen, and usually does given enough time. Anyway, Megadroid has had 14 straight winning trades, so it was due for some drawdown.
Its a healthy reminder that even good trading systems can have period of drawdown and you have to be ready for them. Taking a look at Megadroid's back-testing results for 2009, we see it went into drawdown 3 separate times. At one point it went down to a balance of about $800 and had a top-to-bottom drawdown of 40%
This brings up another good point.
My buddy Sir Redford suggested (a while back) that back test Megadroid with a 30% risk setting versus the 5% risk setting I used in my tests. I tried that today for 2009, and I found that it blew out the account (exceeded 50% drawdown) before the year was half way done!
In the FRWC competition, they stopped any Robot that had more than a 50% drawdown from the initial deposit and I think that's a good rule of thumb.
Its makes perfect sense in that there's a linear relationship between the percent risked, and the amount or drawdown or profit you will experience. And that's another factor to consider to determine how much to risk. If your expected drawdown for your selected risk exceeds 40% of your account size, you're getting too close and you have to dial down the risk.
What's to say that drawdown won't go any higher than 40%?
Well it certainly can, and all we can do is hope that the account balance will have grown enough to survive the larger drawdown when it eventually comes.
Megadroid showed a 40% drawdown in 2009 and 2008, which was the highest in its 10-year history of back-testing. So I don't want to go any higher than that 5% risked per trade.
Enjoy your weekend and check back later.
Tuesday, February 16, 2010
Today is the biggest day ever in the history of Automated Forex trading. The team over at http://www.forex-robot-world-cup.com/ finally unveiled their Robots and the price tag.
Yesterday, they dropped the news that the best logic of the top-5 robots were combined into a single "super" robot called "Fusion V 1.1a" that incorporates all 5 strategies. Here are the short-term performance stats on the robot:
- 354.37% real-money, live trading NET profit in 19 days
- Averages 12 trades per day
- Impressively SMOOTH Equity Curve
- Exclusive money management and risk proportion between the robots designed by the FRWCA true MULTY-STRATEGY "all in 1" profit pulling Forex robot.
- Contains a combination of five proven strategy types: - Scalping - Swing Trading- Short Term Trend Trading - Short Term Reversal Trading- Price Action-Only Based Trading
Another entry in that category is EAKAIN Scaplper pro over at: http://www.besteaforex.com/. This robot goes for $579 and sources indicate its a Megadroid on steroids and turned loose on 7 different currency pairs but with the same deadly accuracy and trade management. Whether it actually works as advertised remains to be seen. The test results from Driven2Success.com indicate its up about 69% in just 3 weeks time, averaging 18% return per week. So much for my 1% a week mantra.
The question is, how stable are those systems over a long period of time? How likely are they to blow up an account? Will the FCM's push back somehow and refuse to trade these robots?
Anyway, FRWC put together a "super bundle" which includes:
The Top 8 Robots:
- LMD MultiCurrency: 145.60% Net Profit In 60 Days
- HiRider: 88.11% Net Profit In 60 Days
- SuperVolcano: 35.25% Net Profit In 60 Days
- VREM2: 19.39% Net Profit
- Nutcracker: 14.93% Net Profit
- Simple: 8.00% Net Profit
- Neg-Correlation -USDCHF: 6.07% Net Profit
- ZOOP-EURJPY: 4.34% Net Profit
- The Fusion V Super Robot spitting out an average of 12 trades a day and returned 354.37% real-money, live trading NET profit in 19 days.
- Access to FRWC's "EA Lab" which has investor passwords to view results of EA's being tested with real money.
For now let's suspend judgment on whether its possible and see what happens. After all, the FRWC's "Live money" test period was a relatively short time, and I would like to see how these robots test over 10 years of history.
And finally, what you've been waiting for is the price tag.
A single payment of $999 buys the entire FRWC Mega-bundle.
Who knows what's going to happen next, its going to be fun to watch.
Saturday, February 13, 2010
I took Expert Advisor Forex PIPZen, (available for free trial at http://www.forexpipzen.com/) for a spin through some back testing. I used all the default settings on EUR/USD - 1 hour timeframe which were a takeprofit of +17 pips and a stop loss of 93 pips.
The results for 2009 and 2008 were amazing, with +100% wins and drawdowns in the low 20's. From there on back, the results were medoicre at best, with drawdowns much worse.
This could be the case that the EA was optimized to show great results for recent years. The readme recommends optimizing on StopLoss, Trailing Stop, and TP4, TP5 (Take Profit) values which I did not do.
So I can't say i've been completely fair with this evaluation, but I don't want to have to bother with optimization and I would rather trade a system that is stable over 10 years with the same set of parameters.
Another issue with this robot is the risk management. With a Percent Risk setting of 3.0, it will trade 0.03 lots on a $1000 account. But instead of just one trade, it puts on 5 trades at once signal opening up 0.15 of the account in a single position. The lots are then closed out with all the same Take Profit, Stop Loss and trailing stop values which are displayed to the broker.
So the risk management seems a little agressive, and inconsistent with my philosophy on position sizing discussed here Account and position sizing.
Thanks all for now, enjoy your weekend.
Friday, February 12, 2010
It was a quiet week in the land of the trading robots. Megadroid traded only once this week when it took a pair or sell signals on the EUR/USD on Tuesday evening.
Megadroid trades 2 strategies concurrently and its rare to see two trades at the same time. In fact seeing any trades is an event since its been trading about 2 or 3 trades a week so far this year. But i'm not complaining since the trades have been 100% profitable so far. Megadroid does have periods of drawdown - some as much as 44% - as we saw from backtesting in my prior blog entry Megadroid testing and 1% a week.. So lets hope the equity continues to build so that when the drawdown comes, we will have an equity cushion established.
Just close your eyes and say "1% a week", "1% a week". At 60% a year it really adds up. I'm looking for more robots in that category, so let me know if you have any tips.
I'm working on some material for a post regarding order handling in MT4. Check back later for that. Enjoy your weekend,
Saturday, February 6, 2010
Well the Forex Robot's World Cup competition hype- machine is in full swing.
Yesterday, they blasted out e-mails about a "Top Secret" report which can be downloaded for free here: http://www.forex-robot-world-cup.com/.
I read the report and I don't think any "shocking truths" were revealed. But it did expose some of the rules of the competition which are worth a read if you haven't seen them already. Let's go through the interesting ones here:
B(13) - The ownership and Intellectual property rights of an EA will remain with the entrant. At no time will the FRWC organization assume ownership of any EA submitted.
Okay, that sounds pretty good.
B(14) - Should the entrant of a submitted EA be announced as a finalist of any prize category, said entrant (as the acknowledged owner), hereby agrees to unconditionally grant to the FRWC organization sole, exclusive rights to market the EA as it sees fit and for its own exclusive benefit for a period of 6 (six) months from the end of the competition - this is deemed to be 14 days from the end of the live trading phase.
So there's the catch, the FWRC gets the rights to sell the EA and bank all the profits for 6 months! The robot authors get the prize money and bragging rights but thats all!
B(16) - At the end of the 6 month exclusivity period, the FRWC organization shall acquire the non-exclusive rights to market the EA for its own benefit in perpetuity. With the commencement of the non-exclusive period, the owner shall thereafter be free to do with the EA as they wish (sell it, market it, license it), save that the owner may not sell or dispose of any rights to the EA if such sale or disposal would infringe, in any way, upon the rights previously granted to the FRWC organization - for example, the entrant of a winning EA could provide marketing rights to a third party on condition that such provision would not restrict the FRWC organization from continuing to commercialize the EA.
I think this means that at the end of the 6 month exclusivity period, the FWRC has the rights to market the robots themselves "in perpetuity". The EA developers can also market the EA along with the FWRC, as long as it doesn't interferes with the FWRC's rights to market the robot themselves.
B(17) - Any improvements to a winning EA at any time after entry into the competition shall be included within the scope of the marketing rights granted to the FRWC organization.
This one is just plain unfair to the EA developers. FWRC gets the right to future improvements in the EA?
B(18) - The FRWC organization may use any information and/or methods disclosed by competition entrants in any way, shape or form for its own gain.
More of the same.
B(25) - All disputes that cannot be resolved will be governed by the laws of Sheung Wan, Hong Kong.
B(26) - If a dispute arises regarding the rules of the competition then entrants agree to first attempt to resolve any issue(s) with the help of a mutually agreed-upon arbitrator in the following location: Sheung Wan, Hong Kong.
Looks like this matter has been carefully researched for the benefit of the FWRC!
C(2) - Entrants must submit the raw MQL code of each EA within the submission period. If DLLs are required by an EA then entrants must also submit the source code and compiler project files for the associated DLL files.
Talk about handing over the keys to the kingdom!
Okay, enough about how the FWRC gets all the rights and most of the money and let's get on to more specifics about the robots and testing parameters.
D(1) - The Forex Robot World Cup (FRWC) organizers will conduct a back-test of each submitted EA on historical data using the MT4 (MetaTrader) platform.
D(2) - The back-test will have a starting balance of US $1,000 and 200:1 leverage.
D(3) - Money management is permitted in the back-test.
D(4) - If money management is *not* enabled and no recommended fixed lot size has been specified in the submission notes by the entrant then the back-test will be made using 0.01 lots as the initial position size.
All looks good and reasonable here. We learned how to do money management earlier in my blog and its well understood at this point.D(5) - The back-test will be run over a 12 month period which is specified on the Submit EA page within the Developer area of the website. Please be aware that we will perform our own independent back-tests over the same and additional, random date ranges between January 2005 and the current date to prevent attempts at "curve-fitting".
D(6) - The back-test will be a "straight run" approach (the entire 12 month period), rather than "month-by-month".
I would rather go with Daniel Fernandez here and look at an entire 10-years of back testing. I agree with no parameter changes and no optimization during the test period.
D(7) - Any EA demonstrating a Relative Drawdown (floating loss or actual draw-down) in excess of 50% during *any* back-test period will be automatically disqualified.
D(8) - Any EA achieving less than $500 net profit will be automatically disqualified.
Okay, looks reasonable.
D(10) - No EA may contain code which has the sole objective of avoiding specific (losing) trades. An example of this rule, but not limited to, is "do not trade on Wednesdays at 18:23 GMT" when it can be shown that one Wednesday during the back-test period would have resulted in a large losing trade and dramatically affected the final result.
Wonder if this would disqualify EA's like the MegaDroid which never trade on a Friday.
D(11) - No EA can place fewer than an average of 2 trades per week, every week, for the entire back-test period.
This would definitely disqualify the Forex Godfather!
Okay, the next step we are all waiting for is the phase at which the EA's will be offered for sale.
Personally, i'm a fan of the one called "Simple" which had a slow and steady rate of gain during the competition and put in an 8% gain with a 4% drawdown. Will they sell those robots as well?
We'll find out! Stay tuned for more fun with Automated Forex Trading!
Friday, February 5, 2010
Its certainly been a wild week in the forex and equity markets. EUR and GBP broke down to multi-month lows versus USD. Global equity markets took a slide as the risk aversion trade was back on. Funny how when things get dicey, the money floods back to the USD - as though the US is some type of paragon of fiscal responsibility!
In the end, we don't care because the best robots will make money in nearly any market conditions, right?
On that topic, let's take a weekly look at robot's i'm following. The Megadroid pulled in 3.5% for the week on the demo side, but some indications are that live accounts only turned in about half that much. Even at half the demo account, a 1.6% weekly return is pretty good versus my new "1% a week" mantra.
The Forex Pip-Zen has a series of small losses after about 5 straight winners. It has now put up a gain of 0.08% (8/10 of 1%) for the 3 weeks I've been following it. My 30-day trial is going to expire in another week or so, and at that time I will probably drop it an add another commercial robot. I'm strong considering springing for "Forex Nightfox" which I've been following with some interest over on http://www.driven2success.com/.
The 2 God's Gift EA's from Daniel Hernandez have been pretty unspectacular. God's Gift (GG) EUR/USD is down at 3.5% year-to-date and GG GBP/USD is up about 3.5% year-to-date. GG GBP/USD had some good sell trades on the GBP selling first at 1.5913 and again at 1.5764. Both trades were exited with a 15-20 pip gain and would have made much more if they held on. I think these trades are being closed by the trailing stop value which is pretty tight at 30 pips. Anyway i'm of the mind that a small winning trade is always better than any type of loosing trade. Even with well timed signals place in the proper trend direction can turn into losers. So better safe then sorry.
I continue to be fascinated by the Megadroid, and how it can be so bloody consistent. So let's see what other information we can glean from this robot.
First, we know it trades only EUR/USD in a narrow 3-hour window from 4PM to 7PM EST. As we saw last week, this the least busy part of the forex trading day. It clearly uses the server time normalized for GMT to determine when to trade.
Next, the system goes thorough a series of messages that go something like this:
Reviewing Price Range
Assessing Market Volume
Upward Impulse detected
Upward Impulse fading 00:47:43
Downward Impulse detected
Assessing Market Volume
And so on. Now its quite possible that these messages have little to do with operation of the robot, and are there just to entertain people like myself.
One interesting thing is that it detects "impulse" moves up and down, then starts a counter which counts down to zero starting from some value (1 hour?). I've also seen similar time-based behavior on a trade where if the trade is on for some period of time (and perhaps starts to get close to the other 21-hours of the day it doesn't trade), it will close the trade at a small profit rather than let it stay open.
That's what little I can report about the mystery that is the Megadroid. I have some more ideas, but that's enough for now.
Enjoy your weekend and check back later,
Monday, February 1, 2010
The Forex Robot's world cup, sponsored by FXCM, wrapped up this past Friday. My buddy Redford did such a good write-up, I'm going to re-post his comments to my recent post.
"The Forex Robot World Cup ended on Friday, and the results are out-
LMD Multi-currency wins the $100k prize with a gain of 145% in two months; HiRider was second with a gain of 88%. But the real story is pips, where HiRider blew everyone away with 2,700 pips while LMD was only 573, and did it with minuscule drawdown. These top EA's will go on sale this week, very limited quantity.
There were hundreds of robots, including commercial robots in the first test, and only a handful made the cut. The organizers say that 99% of commercial EA's are complete crap, and that none would come remotely close to the winners in this competition.
Interesting to get the straight goods from an independent testing group. Confirms what most of us suspect but are afraid to confront. We live in hope that some $99 robot will somehow make us money. I'd rather spend $1,000 on the winner in that competition than buy 10 $99 crapola EA's." (end quote)
Will be interesting to see the long term results of these systems. 2 months is not much of a time-frame to get an idea of the long-term performance of these systems. The fact that they started out with 329 submissions, ended up testing 24 robots indicates they have pretty tight entry criteria.
Keep in mind that there's a bit of selection bias here. Since FXCM is trading these accounts with real money, its in there interest to trade a small number of systems.
The more I think about it, this whole thing is an effort by FXCM to grab the tail of the tiger of the industry which has become Automated Forex Trading - a game which they have been largely deal out of thus far. We know they are a very recent entry into the game with their partnership with Boston Technologies. So they are putting this whole program together in an attempt to reset the agenda of automated forex trading industry.
"Thought I would add some of the FAQs from the link above (World Cup). It is most illuminating. I mistakenly said that commercial EA's were permitted in the competition...apparently not."
Will the FRWC robots be available for purchase?
Yes. After the two-month live (real money) trading phase ends on January 29th, 2010, we will be preparing the winning EAs for sale. We plan to allow a very limited number of copies of the top ranked competing robots to be available for purchase starting February 16th, 2010.
Why is it that you do not accept commercial robots (those being marketed on-line presently) into the competition?
Simply because we believe that 99% of Forex robots currently being marketed are out there because they sell well for those who market them, not because they are good robots.
As a Forex trader you have to separate truth from fantasy. The fantasy about most commercial, heavily marketed FX robots is that they perform as advertised by the OWNERS of the product. However, the reality is that very rarely does anyone make money from these robots.
The FRWC's objective is to allow EA developers to prove that their EA is the best without needing to rely on marketing gimmicks.
Why are the FRWC's EA qualification rules so demanding and tough?
Well... why not?
Why would you expect less than this when it comes to a robot that handles your real, hard-earned money?
We are currently part-way through the first FRWC competition and, as you may already know, 329 robots were submitted during the submission phase.
From those 329 submitted robots, only 24 qualified. Yes, 24 - that is not a typo.
Why are the FRWC robots trading on live, real money accounts and not on demo accounts?
Simply because "forward testing" (the technical term used for this type of trading) is worthless when judging the performance of an EA. It is widely known and accepted in the industry that prices, spreads, spikes, fills, slippage, etc. is quite different on "real money" live accounts compared to demo accounts.
A robot can, and most of the time does, show a profitable equity curve on a demo account but later, when applied to a live account, the opposite happens.
How many Forex robots were submitted in the EA submission phase?
There were 329 EAs submitted from 63 countries.
Why are the winning FRWC Forex robots the best robots in the world?
Because of the "survival of the fittest" concept - plain and simple!In every competition, no matter what it is for, the winner is the best and with competition qualification rules as tough as the FRWC's qualification rules, the quality of the winning robots is furthered magnified.
As a side note, this is a mistake many people make when comparing commercial robots available for purchase on the Internet with the FRWC robots. These commercially available robots, have been designed to make money for their owners by selling them - the fact remains that 99% of them do not work in practice. On the other hand, the FRWC robots have been designed by NON-COMMERCIAL developers with purely one thought in mind: real, long-term, high performance results on live, real money accounts." (end quote)
Again, interesting stuff. I think FXCM rejected the entire existing universe of commercial EA's because it wasn't their product and they couldn't make any money from it. It doesn't mean there are no good commercial EA's. It just means there are no good commercial EA's that FXCM could make money out of.
I don't know about their assertion that 99% of commercial EA's are no good. But I definitely agree that out of 100's or 1000's of EA's there are only a handful worth trading. And its our job as EA consumers to separate the wheat from the chaff so to speak.
Anyway, the drama continues, and a limited quantity of these robots go on sale in about 15 days. What the quanties are, and how much they cost is the next subject of speculation. This is all great stuff, but just keep in mind that FXCM is going to great lengths here to be sure that - if you're going to give your money to anyone in automated forex trading - its going to be them!
Cheers and let us know what you think of the Forex Robot World Cup competition!