Coding an Automated System

WTF Joe, it's only 10:15. I'll take the odd thing before RTH on occasion, but it's 5.15 am in New York and an hour earlier in Chicago.

I might go and have breakfast in a bit though.

On the subject of time (and in the purpose and mercy of de-railing a useless thread quite early on) allow me to recommend the time pieces of Dubey & Schaldenbrand for keeping a discreet nonchalent eye on what time it is in NY whilst breakfasting out. My Rolex bi-metal GMT often looks a litle brash, but this cracking little number is spot on, a beautiful little time piece.

I have the black face model you see here, I also have an un used white faced model in my collection which is earmarked for my eldest Son for his 18th, they originally retailed at $5000, you can pick them up for a ridiculous sum of circa $800 now..imho they are an absolute steal at $800, particularly if you get the exhibition back on them..that is all..

picture.JPG
 
Welcome to the 21st century BS, we have phones which show the time.

On topic, it really depends on how complicated the system is, then again, I've found the simpler things to be more profitable (excluding HFT obv).
 
Seriously, the next step should be learning to code.

The coding is the simple part, a chimpanzee could do it. Actually I dont believe its possible to design an automated strategy without being able to code. 10 years ago I used to believe it was as simple as finding someone who knew how to trade, define a set of rules and automate. It isnt, its a completely different game.

One of the big problems is that most of the people who offer programming services are pretty clueless about automated trading. If they knew what they where doing they'd be doing it rather than working for chump change. The only solution is to do it yourself. At a later stage you can bring in specialist help if required.

The only other advice I can give is dont start by learning a trading platform language such as MQL, start off with something simple until you understand the basic building blocks and concepts like variable scope, loops, conditional statements, functions etc. when you are comfortable you can add in the complexities of dealing with a real time data feed.

The reason I say that is that testing an application in real time that is responding to events based on an unknown real time data feed is a lot more complex than you might typically experiece in other types of development projects where you can engineer test cases and get instant feedback. For example, you might want to test that your ma cross to open a buy trade is working, but then the feckin market goes short ! and so the conditions you wanted to test didnt occur, and might not occur again for quite sometime, and when they do, perhaps you discover a bug, and the whole feckin process starts again. It is possible to generate test conditions and use the backteter, but its not something you should be doing unless you are comfortable with the programming language.

Thanks for the help Hare
 
It's likely he will need access to his brokers API via Excel COM if he creates a stand alone trading application, otherwise they may provide access to their platform via DDE.

No DMA for him, huh ? So which brokers offer retail API access ?
 
Why the mysteriousness. Isn't forums all about sharing and helping ? It's not like your strategy is embedded in the info.

Because you’re most definitely the expert Joe, I’ve seen it in every thread you post in. You know all about programming and development costs, automation, DMA, shares, futures, spread betting, FOREX and [insert topic]. A clever person like you doesn’t need my help for anything.
 
No DMA for him, huh ? So which brokers offer retail API access ?

loads of em. if it's forex then mb, fxcm, and db (before they closed), provide an api free of charge, some impose a charge with rebates for volume (oanda)

as a general rule, if they aint crappy mt4 bucketshops they'll usually provide some sort of api.
 
loads of em. if it's forex then mb, fxcm, and db (before they closed), provide an api free of charge, some impose a charge with rebates for volume (oanda)

as a general rule, if they aint crappy mt4 bucketshops they'll usually provide some sort of api.

Thanks. That's what I call a helpful answer, and no strategies revealed.
 
I have coded up a few EAs of manual systems i have tried. Normally the way it works is the manual system does OK. I make a profit but miss lots of good trades. Then i think i know MQL, why dont i just automate it? Thing is you forget how much discretion goes into even what you think is a mechanical system. Even if you just glance at a particular indicator once in a while when things look marginal it is really difficult to put something like that into rules.

If you do use very specific entry and exit criteria it should be possible to create an EA. It wont cost you a lot, in fact you could probably get someone to do it free (dont expect any support on it though). The best way though would be to learn to do it yourself. That way you would fully understand why it doesnt work the way you expect.

If you find yourself still unable to get it coded i will have a go for you. No charge as if it is profitable i would use it myself. You might be able to get other traders to do a similar arrangement.
 
If you find yourself still unable to get it coded i will have a go for you. No charge as if it is profitable i would use it myself. You might be able to get other traders to do a similar arrangement.

Good strategy. I tried it. No luck. So I decided to make my own money making strategy instead.

I estimate the strategies used by people here are not profitable enough for them to realise they can get more by intelligently and selectively sharing.
 
Good strategy. I tried it. No luck. So I decided to make my own money making strategy instead.

I estimate the strategies used by people here are not profitable enough for them to realise they can get more by intelligently and selectively sharing.

I trade renko these days so in general EAs are no use to me. Its pretty much impossible to define the sort of strict exit that an EA would need that works with renko. I have tried so many but always lose over time. Not becuase the strategy is bad or the entries are bad but because i will exit manually based on how i feel a trade is progressing, and that seems to work for me. The sort of critera an EA would use always give too much back.

It doesnt mean i am not interested in how other people trade. For me to code up an EA in mql its only a few hours of my time (i have built up a collection of code blocks and templates) and if i can make a bit of money from it great. If not then i have found another method that doesnt work. It doesnt really cost me anything.
 
Ok so me and a friend are thinking that one of our strategies would probably work ok if it was coded to work as a fully automated system.
At the moment we trade completely manual with all our set ups but would like to give automated trading a go.

Also if there was a way to alarm us when certain set ups start to appear even that would be good.

Who here does automated trading? And if you wrote it yourself, how did you go about it?

Thanks

http://www.prorealtime.com
Audio / SMS alerts for S&R or indicators etc.

http://www.timetotrade.eu/
SMS / Email alerts for S&R or indicators etc.
 
Last edited:
Ok so me and a friend are thinking that one of our strategies would probably work ok if it was coded to work as a fully automated system.
At the moment we trade completely manual with all our set ups but would like to give automated trading a go.

Also if there was a way to alarm us when certain set ups start to appear even that would be good.

Who here does automated trading? And if you wrote it yourself, how did you go about it?

Thanks

I started off by watching a webinar on "algorithmic trading and MATLAB" from the MathWorks.
 
Top