Old Dog with New Tricks

I encourage you to develop your version rfking. You never really know what you'll come up with until you try.

My "MACD EA" is kind of in the way, but based on the same fundamental optimising 7days on M5; EU & GU, based on this MACD EA: http://www.forexfactory.com/showpost...8&postcount=62

and working along side with:Automated Optimization of a Trading Robot in Real Trading


myfxbook: http://www.myfxbook.com/portfolio/fu...me3wksea/32204


Thanks RR for your diligent hard work, meant alot to me!!!

First, thank you so much RR to share these ea's with us. As a novice coder my jaw just drops looking at the code and at the results you get.

Without meaning to derail the thread, it seems that you are very concerned with legal issues. Have you been sued before by someone who lost money using an ea or does that happen in general? Reason I ask is that I see lots of people here on FF developing and sharing their eas and they usually seem less concerned. I certainly plan to share anything good that I may come up with in the future (don't hold your breath anyone) so if it's an issue it's good to be aware.

Thanks again for your hard work!

The short answer is, yes.

The long answer:

I have been sued three times in EA-related cases, all three times I was sued for negligence. There was a time when I put in a default lot size of 1.0 in an EA. Someone who basically put what was left of their life savings into one of my EAs blew the account, and blamed me for it. He lost something like $5,000, but in the lawsuit, he claimed tens of millions of dollars in "mental anguish" and other BS claims. I won and the judge awarded me double my legal fees. It's been almost two years now I have yet to be paid by this guy. After this case, I changed the default lot sizes of my EAs to 0.0 to prevent any more stupidity on the part of my end-users.

Court case number two was a little more interesting. A 20-something trust fund baby got his hands on one of my EAs and used it without understanding what an EA is or how it works. He lost about $50,000 and daddy comes running after me saying I targeted his little boy. We both went to court, and ultimately, they settled with me before a verdict could be handed down.

Case number three involved a wannabe fund manager. The guy didn't know how to trade, and thought he could take a few shortcuts by imitating how I trade. It didn't work for him. He lost a significant chunk of client money--not sure how he got it in the first place--and he filed a lawsuit claiming that I 'neglected' to say that EAs are not perfect. Judge threw the case out in summary judgement.

There is a fourth case which is still pending with the FTC. A gentleman hired a programmer who in turn blatantly stole one of my EAs and sold it as USDBot. All his offers to settle were rejected, and we go to trial next quarter. I have every intention of seizing every dollar he made selling the code, plus any punitive damages, and making sure every person who bought from him gets a full refund.

I'm very careful around legal issues, because let's face it, if I lose a lawsuit, I could potentially lose everything. I've taken steps to try and protect myself against such problems, but at the same time I don't really want to test them. Lawsuits are expensive, time consuming, and ugly, I'd rather be trading. Look for a book called the "Stella Awards" it talks about stupid, funny, and downright ridiculous lawsuits. In the US, people sue each other for no reason at times, and what really scares me is they win. One example: An elderly woman sues three teenagers after they scared her when they baked and delivered cookies to her house. She refused to drop the case in spite of letters of apology sent from the teenagers. It was quite clear that she wanted money for her 'irreparable mental anguish' and she won the case. People like this are what scare me a lot.

With regard to intellectual property, my programming and trading experiences have taken me to a position where I now consult with small cap funds (< 10million) who either want to start an algo-trading business, or want to improve on their existing one. I've been paid as much as $10,000 an hour to lead development for an EA which is responsible for half the fund's earnings this year (3 million). I can proudly say that every client that has worked with me and followed my instructions has made back double what they paid me minimum.

This HiLo Self Optimizing EA is the first in a series of 8 EAs which collectively manage 17 million. I suppose you can argue that I'm a fund-manager by proxy. The value of these EAs is definitely high, and I need to make sure I have some sort of recourse against anyone who steals my code. I keep advocating on these forums that I want more open-source development for these types of EAs, and thus far it seems rfking2 was the only one to answer the call. I was hoping for some of the giants like SteveHopwood, Hanover, Hayseed, FerruFX, CodeMeister, et al to join together and come up with some sort of library to enable my newest EA concept which uses NeuroEvolution with Augmenting Topologies (NEAT). In short, NEAT is a program which writes a program I tell NEAT what the holy grail does/is, and NEAT tells me what the holy grail looks like. I figured I might as well be the change that I want to happen, and that's why I released this older EA.

That was you, huh? I saw Mark Trenton changed his homepage listing the USDBot as "Sold out", and then heard that he had pending litigation. Well, good for you. I think this problem is more rampant than most realize. EA's are sold all the time on the ebay with no verification of the author.


I knew this day would come. I just thought it would be Cyberdine Systems that would be the forerunner in NeuroEvolution technology.

If a big dude in a black motorcycle jacket and sunglasses knocks on your door....well, your on your own then!

F*ck me.... I'm surprised you still bother after all that grief.

Hi RR,
Thanks for all your great work. Take a look at Support Vector Machines also. It seems they are the preferred machine learning method of late.

SVM is more popular now because it's simpler to use. There are some inherent problems with SVM IMO

Thanks for the reply. That's something to keep in mind for everyone who releases a piece of code here on FF. Even people who aren't from the US can still get sued there. I'm glad that you've come out on top in these cases and I really hope you continue to do so.

I would like to help with coding but tbh I lack the skills to do anything that complicated, back when I studied computer science things like NN were beyond anything we could imagine and since then my path has strayed far from coding. Still, if you need simple stuff done don't ever hesitate to ask me. I cannot imagine that the more established coders you mentioned are not willing to help, perhaps it's just a matter of pm-ing some of them and ask?

Quick question here ... How long will it take to optimise before it will begin to make trades?

I have 1 million bars 1M on my screen for several hours now and no trades.

I have been using for 2 weeks now, and the result is marvelous. Added 10% to my account(demo).
It seems I only managed to open GBP/JPY and not able to open other pair even "legal" to trade that position. And it opens 2 position(2 GBP/JPY) instead of 1. The position No.2 is opened after hitting the lower low/higher high. I believe there should only be one position open.

I am using HiLo Trader 2.

Btw, I'm using money management, set to TRUE with the formula below.

Lots = MathFloor((AccountEquity()/10000000*Risk*StopLoss))*0.01;

It will use risk factor to determine the lot size.

In my case, I use Risk = 25, and it will give 2.5% max per trade.

Other changes I made included changing the success score to expected value.

double BOSuccessScore = (BOWinRate*TakeProfit-(100-BOWinRate)*StopLoss)/100;

double CTSuccessScore = (CTWinRate * TakeProfit - (100-CTWinRate)* StopLoss)/100;

And I only trade when expected value is higher than 200. Set MinSuccessScore = 200.

Takes my computer 10-15 minutes.

On my 2.66Ghz quad, it took about an hour per chart.

I shuffled the settings like a deck of cards and searched threw every line of code. Unfortunatly I could not replicate the problem nore find any bugs. If anybody else is having this problem please let me know.


Thanks Ron, I came close to uploading it as Rons HiLo forgetting that we have the same name LOL. I certainly do plan on continuing my development of this EA. I am not so sure if the trend up and down idea will work well with any hourly arrayed optimizer. But time will tell. The HiLo entry is the most basic of entries and works well when optimzed correctly. I would like to test what other entries could be well optimized. Typically I would use a seperate EA to manage my trades, but the TP optimizer is working perfectly. All in all I see an infinite amount of potential in this EA. Keep up the good work Ron.

Anybody spell checking me can eat it cause I wrote this from my cell phone

RR, a contributor to one of my threads brought this post to my attention.

Please do not take my lack of contribution here as lack of interest or support. Whilst I enjoy your description of me, I deny its accuracy. I am just a hack who bolts bits of mql4 code together to create trading robots. You have probably forgotten more about this stuff than I will ever learn.

I have been too busy as a mechanic to join in here. The speed with which I am coding robots is decreasing as some of them actually become successful traders, so maybe the opportunities to contribute may arise.

Just don't count on me contributing anything useful. I read the first few posts here when you started this thread, and recall not understanding much.

Cheers

I appreciate your dedication to the open source. I have been reading your old threads and it looks like you have come a long way since your divide by zero error. Day 1 of testing this EA, first I looked at the code carefully and put it straight to a live acct. (Anybody new... this is never recommended unless you know what you are doing, and are willing to throw your money in the dumpster) I played with the settings until I was happy. Changed the code so I could visualize the HiLo lines. Then put it on low, 1 micro per trade, 1 trade per pair, on my nine most profitable pairs with the old HiLo. In one day ended up with $83 profit with 33 winning trades and not a single loss. I do have 1 floating -63 pip and I am waiting to see how the traditional Sunday volatility goes for the rest of my open trades. (bad for old HiLow) But with hourly arrayed optimization I am fairly confident in the bot. I heard about USDbot, didn't know that was you, good luck man. Anything you need I got your back, I am going to start reading up on NEAT asap. If by some random chance I do come up with something good. Its all yours to sell to a hedge. I used to own a mortgage company so I want no part of the financial industry. I'm in this to learn and share the knowledge.

Have a good weekend.

Hi Ron,

Thanks a lot for sharing your new EA. It looks impressing.
Please tell me what's your SL, TP, TS and parameters to optimize?

I'm set HighestSuccessScore to true and use 1 000 000 bars to optimize and after 1hour it used about 40 000 ticks to optimize.
I have Pentium D 2x2,66GHz, 1GB RAM, Win7

Regards

Lower the bar count I guess. 1GB ram is pretty small IMO, the computer I have this EA running on has 32GB and is running 20 instances of this EA.

As a followup, consider the number of bars you're backtesting against in terms of the time. E.G. 1440 bars is one trading day. There are roughly 20 trading days in a month and 200 in a year.


1.5 million 1M bars is approximately 5 years of data to optimize against. Getting uses 50,000 bars which is approximately 6 weeks and has achieved pretty good results based on the screenshot I've seen.