Do Robot EA's Work?

[megamuel]

The backtest is all well and good but do you have any forward test results? The reason I ask is because, like I alluded to in my previous post, it's pretty easy to get good backtests. Pretty much any system can be curve fit or over optimised with a little bit of tweaking of parameters. Have you taken any measures to ensure that your backtests are not over-optimised?

Sam.

 

[pboyles]

The backtest is all well and good but do you have any forward test results? The reason I ask is because, like I alluded to in my previous post, it's pretty easy to get good backtests. Pretty much any system can be curve fit or over optimised with a little bit of tweaking of parameters. Have you taken any measures to ensure that your backtests are not over-optimised?

Sam.

The easiest way is just to keep you losing positions open hence the 92% drawdown with no dips in the equity curve.

 

[megamuel]

pb, I wasn't referring to that backtest. Obviously that one is cr@p but he admitted that. He said it was just one he was working on. Nobody would actually try to sell an EA with a historical drawndown of 92%.... That would just be retarded. I was asking him if he could share results of an EA he was actually trading or selling, to which he responded asking me to check the one out on his website. I did, and sure, the backtest looks good. But my point is, getting a good backtest is easy. Getting good forward test results is not so easy. So really we need forward test results (ideally something like myfxbook) or information on how the vendor in question has avoided curve fitting and over-optimisation. Now, I know how this can be done with certain techniques, but I want to see if the vendor knows.... By his/her lack of response I'm guessing it's a no but I will remain patient.

Sam.

 

[megamuel]

Example, looks at this backtest from one of my systems:

Looks pretty good doesn't it. Nearly £11k profit in 10 years trading 0.1 lots. Drawdown just 11.13%. For sure if I set a website up with that backtest on the homepage I'd be able to sell that. But the truth is, that is completely curve fit to that sample and over optimised. It doesn't work before that sample and it won't work after that sample. So, my question is, are you selling something similar? Or have you taken steps to avoid curve fitting and over-optimisation?

Sam.

 

[999cjb]

Sorry megamuel, I have to sleep at some time!

I produced a spreadsheet of almost one year's live running on one of my accounts for a member when he wanted to compare my results with his own.

Not sure how I can get that to you & I am not keen on attracting more missiles by just posting it here unless you ask me to..

Just let me know how you want it and I'll find it for you.

I can tell you I am personally running eleven instances of that EA, 9 for EURUSD and 2 for AUDUSD. This is a long term EA and needs to "bed in" for a couple of trades. After that, results for all 9 EURUSD trades are remarkably similar, regardless of broker.

I can't say it has been consistently profitable because it lost in the 2007-8 year. But you will see from the spreadsheets & stats on the site that this loss was easily made up in the following year. Due to the Euro difficulties and the general lack of any consistent trend pattern this summer, the EA is only just in profit by a few pips. But, in the circumstances, its better than a loss.

The EA runs on crosses of long term trends and can seem to do some pretty stupid things sometimes, but it does work. If it didn't, I wouldn't run it myself or allow anyone else to either.

We have tried to improve the performance but whatever we do, in the long term it has always, so far, performed best as it is.

 

[megamuel]

Sleep is for girls! Spreadsheets are no good really for obvious reasons. It's ok though, you don't need to prove anything to me, I'm quite happy plodding along trying to develop my own systems, I am not buying anything. I just thought I'd try and help you defend yourself from the angry mob but I must say I am a little disappointed with your response, despite being seemingly honest. Anyway, C'est la vie.

Sam.

 

[999cjb]

Sleep is for girls! Spreadsheets are no good really for obvious reasons. It's ok though, you don't need to prove anything to me, I'm quite happy plodding along trying to develop my own systems, I am not buying anything. I just thought I'd try and help you defend yourself from the angry mob but I must say I am a little disappointed with your response, despite being seemingly honest. Anyway, C'est la vie.

Sam.

Perhaps the information below will be of assistance to those reading this thread.

francisfinley has made a number of posts here rubbishing me generally. He has now done the same on another forum where I am somewhat better known. Here is his post on the tread entitled:-

Re: Do you like average commercial EA sales pages?

What like yours Chris?

Easy Forex - Home

Why is it on forexfbi? http://www.forexfbi.com/forex-robots/forex-ezefx-ea-review-ezefx-com-scam/

So you see he is continuing with the same theme he has been pushing here.

Now here are some responses:-

francis u from forexfbi?

forexfbi is bull****, all the EA's it recommends are **** and never work for anyone but them. i remeber it was supporting EA's that were proven to change results and repost them once the SL was removed. it is a scammers site and i would never trust it biased towards crap EA's that have the best affliate commission.

Personally, i think that forexfbi page is very unfair towards Chris. He also can't really answer your question without looking like he is just being defensive. I have known Chris a long time and he is as genuine as it comes. He may or may not have a great EA, i'm not sure as i haven't looked at his in too much depth, although i have looked at it going back some time now, but i can promise you that he is a "genuine" EA developer/trader who is in this for trading and not for pushing out EA's for profit.

It might be worth noting the following about their review...

1. ezefx have no affiliate page (that i know of). = no incentive for sites to recommend for referral, and also sites are unlikely to dedicate resources to investigating the EA properly.

2. Their review is full of 'probably' and other statements which look to me like a lax approach to investigating. They don't even know what timeframe or style of strategy it uses, their columns left as "n/a". I wonder why they even published a review at all.

3. $100k profit on $10k over 11 years is apparently bad? To me, it is simply a sign that the backtest may be 'genuine' and not overoptimised for the purpose of showing slick curves to generate sales. Seeing this comment has actually made me want to look at ezefx again, not put me off in the slightest, i feel it is a positive. Looking at the shape of the equity curve actually has me pretty impressed as there is absolutely zero obvious signs of overoptimisation or problems associated with large stoploss system (very large equity drops, etc). I can't see the drawdown over those 11 years but it looks small.

4. I agree with his point about no refunds though. At least Chris is honest and clear in his no refund policy though. Many paypal accepting vendors simply refuse the refund knowing that paypal don't really care, but promise refunds in their literature. At $25 a month the cost is very low compared to similar products and would allow anyone to test it out very cheaply.

Finally, Chris is a longstanding member of our forum here and has made some interesting contributions to the community. I don't always see eye to eye with him on everything and we've had our fair share of ummmm "disagreements" but the fact he is so active here and you can see all his posts should help anyone research a bit about who he is, what he believes and the level of his experience in trading. I have no real reason to defend Chris here other as like i said we had our share of fallouts on various things, but i did want to just mention a few points to balance things a little - and would encourage you to research for yourself as i mentioned.

I don't feel I need to say anything else about this except to tell you that we are continuing to develop our EAs, they do work and we are doing all we can to make them work better in the future. If and when we are successful in this, any past members who have lost out from some of our earlier EAs will be offered free membership for varying periods depending on when they joined.