Discuss Forex-auto-trader.com

[Ken Long]

I also wonder how much Artifical Inteligence is actually in this program,.. for $60, Currently on sale for 50% off the already discounted price.

However, I have read some very good reviews on this robot, Forex-Systems-Reviews.com, this is what interests me and brings me here to see if there is any supporting information.

I appreciate your comments Roberto, at least they seem straight forward and honest. The screen shots of your user settings are especialy valulable, they show the very limited transparency of this robot.

Reviewing the trade results shown on the web site, on the short , 1 month, forward test it appears he is running a 25 pip TP, a 22 pip SL, and some smaller increment of TS.

His gains are not from large gains or many winning trades. He has neither. His results are from a winning average during his 1 month test period and a large position size, it appears he is trading with a 34% positon size. In combination with a 22 pip stop thats about a 7.5% risk per trade, a bit higher than recomended.

His 2.5 yr backtest results are a little strange as well. He starts off with a 43% position size and after 6 months he completely changes the system parameters, which is imposible to do in the middle of a backtest.

He starts with a 30 pip TP, a 27 pip SL, and a TS. He then changes to a 40% position size with a 15 pip SL, 50 pip TP, and a TS.

By the end hes running a 32% position size with an 18 pip SL, a 65 pip TP, and a TS.

These changes are not caused by the internal program as they dont change systematicaly like the position size. They change periodicaly when he does an update.

Overall these results are very interesting. Obviously his oversized gains come from an oversized risk alotment, but even with a reasonable risk setting it appears it can make a good profit with a very tightly controled loss and limited drawdown.

Impressive for a black box robot with no accessable user parameters.

I am buying this at their 60$ sale price and will continue the testing with you.

Bought, I'll now be joining your discussions on a more actively involved level. I'll also bother their techsupport for a detailed explaination of the internal workings. Its no good just having anouther robot that works if you dont know why or how it works.

Ken

 

[roberto]

here are some screenshoot from last weeks trading. So fare there where 4 winning trades and 4 loosing trades from Dic 28 -> january 09/09
son 50/50.
I still think the system makes stupid trades selling on bottom buying on top (or near)
Does anyone have some updates?

 

[Ken Long]

I still think the system makes stupid trades selling on bottom buying on top (or near)

I have yet to find a system that doesnt. At least when I make these kind of mistakes I have more control over the end results. I find it difficult using and trusting anyone elses system, especialy when its as concealed as this is.

Even a system such as the Steinitz HAS has enough transparency that you can tell what it is doing wrong. A system like this you have no idea what it is basing its decisions on. You have to be willing to trust it on faith, a very poor way of making any decisions.

Even at $60 this is a bad purchase. Theres not much I can learn from it and it certainly doesnt perform well.

I'll run a few more tests, but theres not much we can do with a system like this.

 

[Ken Long]

It's sure more than in some of its users.
Care to check the statement I posted?

I have a question. Whats with Mr Rude and condescending? I've gone over his past statements and he seems more like the company pimp than an EA tester. Is this the same Craig Reynolds who sells this EA?

If so how are you getting results that arent replicable? Backtesting and forward testing do not seem to support your statements.

Perhaps this is just anouther situation where any good reviews have more to do with the 60% affiliate payments than anything else. Unfortunately theres a lot of that going around...

 

[Ken Long]

I cleaned my data and am rerunning some tests. The results seem a lot better.

On a 1 year backtest this EA did very well up untill Dec. Using his risk settings, 0.3 risk, 30% of equity, it went from 10K to 134K, before giving back 56K. This seems to have been optimized for this one pair, on this one time frame, up until Dec, when it all falls apart. This is a pretty common occurance with these marketed EA's.

It doesnt backtest with any other settings. You can change the risk allotment, but trying to change the TP, SL, or TS doesnt work. This is a very controling program, apparently for the obvious reasons.

Perhaps this one will work going forward, but we cant even hypothesize without knowing what its system runs on. Regardless of any of these results, its not really usable without knowing why it does what it does.

Trusting in black box systems is for the blind faithful and has no place in inteligent trading.

 

[roberto]

crap software... doesnt work!
Again buy on top and sell on bottom.
Will ask refund

 

[Ken Long]

Roberto,

Have you done any backtesting on this EA. I have a feeling you may be wrong condeming this one too soon.

In spite of the arrogant and condesending attitude of Mr PipBluster, who appears to either be Mr Craig Reynolds, or is simply regurgitating his data.

This EA backtests quite well. I am rerunning some of my tests. In all cases the performance falls off in Dec. This is not only this EA, it is across all the markets. It has affected my own trading and many EA's.

Still, on a 3 month test, 10K starting balance, 1.0 fixed lot size, (tests run faster with a fixed lot), it ended the 3M period with a 6K profit.

It took plenty of losses to achieve this, 42 losses vs 45 wins, but it keeps its losses small and allows its winners to run. This is a proper trading strategy and a difficult one to implement in an EA. Most EA's run on a very large stop, and a small take profit. A much easier strategy to design and program, with a lot more leeway for error and inacuracy. And in most cases a few bad trades can destroy a lot of gains.

This robot program barely broke 50% wins, but its average win was almost twice the size of its average loss.

On a 6 month backtest, it cleared 10.7K net profit, on a 10K balance, with a 1.0 fix lot size.
Doubling the account in spite of the Dec slump.

On a 1 year backtest it cleared 12.4K on the same account. Not quite as good.

On a 10 year test it had a few more problems, and really showed its strengths and weaknesses. Yet it cleared 38.4K net, using the same account and fixed lot size. The very tight stop used resulted in more losses than wins, 735/607, and a lower average profit to loss ratio, 267/168. It also had a very slow start, only taking a 50% profit during the first 2 years and then giving it all back and some during the next 2. After that it took off climbing from a low of 9K to a high of 50K, from 1/03-1/09.

This is all with a fixed lot size.

With a variable lot size, using the preset 0.3 risk, 30% of equity, it cleared $9,352,759.16 net during the 6 year period from 1/03-1/09. Rising from 10K to over 10M, before ending at 9.4M.

This is not simply a crap software. I have seen crap EA's. This is a fairly well constructed EA. Totaly capable of producing a nice profit. It does take a large number of losses, and that has to be figured into the risk profile. However, running on a dedicated account, allocating a percentage of equity it can produce some incredible numbers. Hardly the crap EA.

Bearing in mind that this is all backtesting results, and forward testing with real money is liable to be quite different. I would say that this is the kind of EA that I could put on a virtual server and run 24/7 for a number of years. In fact I am going to keep this EA to do just that.

It is not the kind of EA I would want to follow, or use as a trading assistant on a daily basis. It is too inacurate, and slow, for my trading style. I prefer the style of trading you mentioned, trend and range, trading the range, buying near the bottom, selling near the top, following the trend. I have looked at robots claiming to be able to do this, but none are accurate enough to run a 30 pip stop, and still produce any profit...at all. Most run without a stop, or with a stop many times higher than its gains. Just try running this EA's settings in any other EA. I just tried a 6 year test with ForexAutoPilot, 220 TP, 30 SL, 30 TS. It only lasted 3 years, account gone. A 1 year test only made it for 4 months before destroying the account..

 

[Ken Long]

I've stoped trading this bot. It did well in backtests with very large position sizes, but the new version stoped that. They lowered the position size and increased the stop loss just so the robot could survive, but ended up killing any posibility of good profits. And other than the position size none of these settings are user adjustable.

In forward testing the losses are simply too much. It might work on a dedicated account, if you left it for a couple of years. With the new settings it would take a couple of years to make any money, if it didnt loss it all first..

RIP, anouther one down. Theres no intelligence like artificial inteligence..

 

[Ken Long]

I've updated my opinion of this EA with a rating of 1. The only positive comments here are from a person called PipBuster who is either the developer himself, which I suspect, or anouther worthless affiliate.

This EA has no user adjustable parameters. Even the ones featured dont work for the user. It simply reverts to the developers preset settings.

This supposedly artificial inteligence EA only carries a file of 20kb, less than most comparable EA's, and certainly not enough for an actual AI program. The only artificial inteligence here is the developer.

Throughout all my testing this EA consistantly returned over a 50% loss rate. When I first recieved it it ran with a 30 pip stop and a 30% position size. This allowed it to turn a profit in backtesting, however it was never confirmed in forward testing. Now it will only run with a 60 pip stop and 10% position size, not enough to produce any profit at all, and theres no way for the user to actualy adjust these parameters, it simply reverts to the defaults in use.

But the final blow was the refusal to refund the $60 cost. This is not a ClickBank product or there would have been no question, either produce a product the customer values and wants to keep or give the money back. This was sold through Plimus, who forwarded my request for a refund to the developer, who flatly refused.

End of story. This is not a company to do bussiness with.

 

[RLR]

Stop loss

I habe found that if you have too large a stop loss youlose big when you lose and if you don't have a large enought stop loss you lose more often. In any event I will make money some times and then lose it all back. What are the approreate stop losses when using 5k

RLR