Best automated trading system for futures

[oildaytrader]

Banks YES. EA System sellers NO. Unless somebody proves me otherwise....

Why should any body bother proving to anybody or somebody on the net?

Try it.

http://www.trade2win.com/boards/gen...08-learning-profitable-automated-trading.html

 

[oildaytrader]

Well then i'll never believe and maybe you may know something that
i dont. Yippeeeee....

We are not going to tell you what you don't know!

 

[SJ-60]

Shane,

If you cant be arsed looking on here PM me and i'll send you a few PDF's and a few links from here and FF that you can make money from.

Ged

Hi GED,
I am looking for some info which you offering to Shane I am new to trading and trying to develop a srtrong bussines if you can help with some usefull info I would appreciate that .
My email address is : [email protected]

Thanking you in anticipation of your info ,
Regards
SJ-60

 

[TickCOM]

There are two main types of automated trading. Systems which benefits from market differences in m secs, and strategies using a higher timeframe.

1) For Hfreq trading you need the tick data before the market, you need to pay to get the tick 100ms before the others, plus the ability to execute orders at a range of brokers. The system requires a set-up no individual trader can match.

2) Higher timeframe trading, typically pairs trading, if the timeframe is above lets say 60 minutes charts, no need for automated trading. Thus automated trading for individuals is for daytrading typically 1 - 5 minutes charts. To develop a strategy, backtest this, run and administrate such is quite a job.

Ask if anybody here is using unattende automated trading, and if they have succes - ask them how much work it takes to develop - and how much work it takes to administrate and run.

 

[Xeno]

2) Higher timeframe trading, typically pairs trading, if the timeframe is above lets say 60 minutes charts, no need for automated trading. Thus automated trading for individuals is for daytrading 1 - 5 minutes charts. To develop a strategy, backtest this, run and administrate such is quite a job.

Ask if anybody here is using unattende automated trading, and if they have succes - ask them how much work it takes to develop - and how much work it takes to administrate and run.

Higher time frame isn't necessarily pairs. And you missed the case where time frame is 60 or above, but tick or minute is used to fine tune entry - that's also useful to automate.

Development takes many months, but then gets much faster when you have lots of useful code written and plenty of experience.

 

[TickCOM]

Higher time frame isn't necessarily pairs. And you missed the case where time frame is 60 or above, but tick or minute is used to fine tune entry - that's also useful to automate.

Development takes many months, but then gets much faster when you have lots of useful code written and plenty of experience.

The point was - if the timeframe is higher than 60 minutes charts - why have a complex system running when you can check the charts 2-3 times a day. Please also note the word "typicaly".

 

[Xeno]

The point was - if the timeframe is bigger than 60 minutes charts - why have a complex system running when you can check the charts 2-3 times a day. Please also note the word "typicaly".

Yes, and I'm saying that high timeframe automated trading is not typically pairs.

If you have a complex system then checking the charts 2-3 times a day is unlikely to be a short glance. That is a good reason to automate. It may also be that your system is collecting non-chart data in the many hours up to its entry. There are loads of reasons in fact...

 

[TickCOM]

Yes, and I'm saying that high timeframe automated trading is not typically pairs.
.

I think many investors ( and the fund managers who did adm their funds ) who lost 80% of their funds in 2008 will disagree on that. ( Because the diff kept increasing and the pair didnt return to normal till in 2009).

 

[Xeno]

I think many investors ( and the fund managers who did adm their funds ) who lost 80% of their funds in 2008 will disagree on that. ( Because the diff kept increasing and the pair didnt return to normal till in 2009).

You've proved that pairs trading exists. Still doesn't make it typical, or change the fact that there are plenty of automated systems outside the two areas you proposed, or make automated systems less useful for high timeframes.

 

[retirerich]

Do you miss the opportunity to pick your favorite stock at the right price as you were not able to track the movement? With automated trading you don’t need to spend hours tracking the market or the price movement of your favorite security, it is a great tool for a beginner like me.

I was suggested about tradelikepros by one of clleague, and more than happy with the results that I have achieved so far!

 

[SpreadIntel]

Is this automated just for the outright or for spreads?

 

[cedricthefrog]

There are two main types of automated trading. Systems which benefits from market differences in m secs, and strategies using a higher timeframe.

1) For Hfreq trading you need the tick data before the market, you need to pay to get the tick 100ms before the others, plus the ability to execute orders at a range of brokers. The system requires a set-up no individual trader can match.

2) Higher timeframe trading, typically pairs trading, if the timeframe is above lets say 60 minutes charts, no need for automated trading. Thus automated trading for individuals is for daytrading typically 1 - 5 minutes charts. To develop a strategy, backtest this, run and administrate such is quite a job.

Ask if anybody here is using unattende automated trading, and if they have succes - ask them how much work it takes to develop - and how much work it takes to administrate and run.

I run automated strategies on the FX markets. Let me just say a few things;

TickCOM is right. Very few individuals have the technical ability, time, and money to set up a high frequency trading business. I know it is possible but the large firms have the advantage in that market. It is lucrative but not for the majority on this board.

In my experience, the optimal domains for the automated retail trader are in the 5 minute up to daily timeframes. I personal work from the 15 minute and up looking for small statistical edges that persist for months at a time. I have a few swing systems that use forms of traditional technical analysis but in my experience they are not big money makers.

It has taken about 3 years of hard slog to get a stable automated platform running at even then it is not hugely advanced. For me the whole process was enjoyable (if a little frustrating) but there were no shortcuts. Oh,..and dump metatrader backtester - that is one of the worst backtesting tools i've ever had the misfortune to use.

 

[Studog]

tradeonauto has something worth considering.
It is hard yakka developing these things. Most recently I am having success using 35 second bars with quite aggressive entries. The problem is that bar length for me is key and as Cedrik has alluded to, there are certain periods that can be very good and hitting that period brings the bucks. It doesnt mean that period of time will exist forever however so automating is fine as long as you ask urself questions when things turn a little pearshaped.

 

[Boletus]

Interesting discussion. I wonder if anyone here has tried writing (not buying!) an automated system that works with raw data rather than with data that's divided up into bars? I'm currently attempting do just that (not for futures but for spot forex); I acquired 10 years or so of data and I have written a framework including my own charts that backtests on this data using strategies programmed in C++. I intend to refine this system until I feel it can be refined no further. So far my best system made 500% in 6 years, but it had a whole year of being on a losing streak so is no good to me yet. I don't trust it not to lose all my cash. Plus my broker (Oanda) have a 600 USD a month fee for using their API, so my system would have to be able to pay this fee and then some.

But I'm encouraged by these results since at the moment I'm still learning the ins and outs of pattern recognition, chart display etc, and I can see that there are a ton more things I can do to get profitability up.

Any thoughts on how wild this goose chase is?

My two pence worth on the subject of buying automated systems: I know nothing about the subject but it's hard to see why someone would sell a profitable system, unless the deal was that the profits were very modest but very reliable, like 10% a year or something.

 

[Dommo]

Interesting discussion. I wonder if anyone here has tried writing (not buying!) an automated system that works with raw data rather than with data that's divided up into bars? I'm currently attempting do just that (not for futures but for spot forex); I acquired 10 years or so of data and I have written a framework including my own charts that backtests on this data using strategies programmed in C++. I intend to refine this system until I feel it can be refined no further. So far my best system made 500% in 6 years, but it had a whole year of being on a losing streak so is no good to me yet. I don't trust it not to lose all my cash. Plus my broker (Oanda) have a 600 USD a month fee for using their API, so my system would have to be able to pay this fee and then some.

But I'm encouraged by these results since at the moment I'm still learning the ins and outs of pattern recognition, chart display etc, and I can see that there are a ton more things I can do to get profitability up.

Any thoughts on how wild this goose chase is?

My two pence worth on the subject of buying automated systems: I know nothing about the subject but it's hard to see why someone would sell a profitable system, unless the deal was that the profits were very modest but very reliable, like 10% a year or something.

Yes - I use a Windows/C api (for quotes, orders, a/c, positions) and wrote everything from scratch. I have some code in C++ but find C# a lot faster and easier for development. Unless millisecond timing is critical or you're doing something extremely CPU intensive the drop in code speed (and I don't think it's all that much) is perfectly acceptable. If you're on windows and know c++ you would find c# easy to get into and very productive, I think. For anyone else interested in writing charts "Practical C# Charts and Graphics" by Jack Xu will give a head start on the graphics side of things. Staring with ideas in this book I wrote line and candle-style charts. I don't use TA (or, indeed candlesticks, except as a view of price range) so that's the full extent of my graphics.

I started from scratch so as to avoid getting hit by any limitations in commercial scripting engines. There is of course a lot of work involved but with the basic infrastructure in place, you can start to do more interesting things. I wouldn't discount commercial/broker supplied systems - I'm sure that with a script you can rapidly develop and test ideas more quickly than writing out new strategies in c#, but you may hit limitations on what you can do. On that note, if anyone has recommendations for a script parsing framework that would be of interest.

I also looked recently at ESPER/NESPER which is an event processing engine that operates like a realtime in-memory database, with a SQL-like syntax. It looks very interesting although I still have a prejudice for procedural over declarative programming - I'd rather write "if(x>a)..." than "where (x>a)".

I am comfortable running the automated system unattended (continuously from Sunday night to Friday night) although of course I keep an eye on it.

 

[Boletus]

Do you find it profitable, Dommo? Or still borderline at the moment?

I'm happy with this way of working since I can code about as fast as I can think, but I wonder how easy it is to make profits this way?

Also any API recommendations?

Thanks!

 

[Dommo]

Do you find it profitable, Dommo? Or still borderline at the moment?

I'm happy with this way of working since I can code about as fast as I can think, but I wonder how easy it is to make profits this way?

Also any API recommendations?

Thanks!

It is profitable (if it weren't, it wouldn't get anywhere near a market!). I view the whole thing as a statistical business, although I'm not doing anything like HFT. I do not set stoplosses, although I have rules for dealing with drawdown. The automated trade app trades better than I do, even with the same rules, basically because it implements the rules in a cold way, and it is 'always on'.

I use GFT api. I think FXCM may also have released an api recently, and there are of course others. I couldn't say how GFT api compares to others, and I've never used FIX. There are maybe a couple of things I would like to see done slightly differently, but these are minor and it does all the basics that I need. You get the static api and documentation, but you need to spend some time figuring out the dynamic sequence of messages (I'm sure you know what I'm talking about). There are code examples. I have found GFT customer service to be excellent (and I have no affiliation except as a client).

 

[Jason Rogers]

I think FXCM may also have released an api recently, and there are of course others.

The list of available API's is here http://forexforums.dailyfx.com/?p=api-support . You mentioned C# coding in your previous post. Strategy Trader was recently released by FXCM, and all strategy and indicator coding for the platform is in C# . Details here http://www.fxcm.co.uk/strategy-trader.jsp#API

 

[fatty99]

Keep in mind that there are several advantages to automation.

The part that particularly interests me is the "watchdog" feature. Rather than babysit a chart for hours on end, you can employ your technical indicator requirements and feel confident you'll execute a position order without missing an opportunity. This is particularly useful to those traders who are employing strategies that may only feature a few opportunities a week. You can go to bed knowing that if the market continues to "setup" that you won't miss your entry (or face having to get up every 30 minutes to check it). This allows you trade off hours (on European and Japanese market times) where viable opportunities may exist.

The other huge advantage is removal of emotion. Like stated earlier, if you've developed a strategy, backtested it and verified that it's profitable, then stick with it. Sometimes that's easier said than done. Even with manual trading, I find the best strategies (for me personally) are those where I set a stop and limit and walk away. There are those "intuitive" traders.....that ain't me. "Never add to a loser. Never cut a winner short." Obviously knowing the difference means the difference between bankruptcy and billionaires. For those of us who aren't oracles, automation provides an emotionless option to help you "stick to your guns."