Partners in Automation Software.

I am looking at the possibility of having some automation software professionally developed. I have some experience with automation models but would now like to take this to the next step.

The opportunity is aimed at those not skilled in programming skills but have a basic idea of how to create systems.

The software could be used for:

• Generating signals – if you have certain criteria you can use the software to automatically generate signals.
• Automated trading – If you have a working and clearly defined strategy you can use it automate your trading 100%
• Scanning – use it to scan for opportunities, strong/weak stocks etc
• Money management – apply more sophisticated MM and risk management strategies to your trading.

It should be flexible so you can use it for your own individual needs.

I have looked at some of the costs and it has held me back for a while. If there are two or three traders who have thought about automation but have had the same obstacle then maybe we can talk about sharing the cost of getting this developed .

I would be interested to discuss further with serious traders.

I am a systems developer. Perhaps I am too skilled in programming and IT systems for you. I am interested in trading strategies that is profitable. So I am not looking for exploration, or shooting in the dark. I am not a trader, except in my own exploration. So I can't contribute towards a successful strategy. But I can guarantee a successful automated system for a pre-existing successful strategy.

I am willing to exchange ideas and help you on your automation quest, for free and right here on this thread. No financial interaction nor obligations of any type.

But i might help, even if it were by merely keeping your thread alive. That's right, since this is the only thread alive in this deserted section of the forum, thanks to there being 3 duplicate forums -- one for general mechanical trading systems, one for free systems, one for commercial systems. By the way, we should ask management to create also one section for "free systems by australian traders", and some more sections for free systems by traders from other continents. That way each mechanical trader could have his forum own section and comfortably talk to himself.

Regarding the topic you are discussing, I am good, but with simple stuff. I have automated 120 systems on one excel spreadsheet. Totally automated. I back-test stuff on tradestation. That's all I use.

Any questions you have, it would be interesting to hear them.

Travis thank you for your kind offer to help. Always looking to improve my skillsand understanding from someone more experienced.

You have had some excellent success with programming in excel. I have done some work in excel but my limitations as a programmer held me back . I couldnt get the order execution done though was getting good enough signals.

I think the a major help would be to start clarifying the entire process of generating a signal, in terms of analysis, setup, entry etc. I normally follow and try to define in hese steps but maybe still something missing.

What aspects do I need to consider, in terms of the flow, when programming automated strategies, in the order so they connect at the right times. This has been challenging so far.

Thanks

Well, it's pretty simple. I will summarize the process step by step. Then we can proceed further if you wish and if you have done the previous steps:

1) open an account with IB
2) install their TWS and their API software
3) start customizing their sample excel file, "twsdde.xls"

I know everything you have to do to make it work properly, but I can't give you my file nor do all the work for you.

Thanks you. I will get on it . Does IB provide historical data also or a facility where we can store a set amount of bars?

I don't know what you mean by "facility". But yes, IB does provide historical data for up to a couple of years back, for the futures I trade, and I suppose it provides data also for all the other tradable assets and securities.

Regarding the data, however... for backtesting on tradestation, I buy and recommend buying data from disktrading.com. Their data is as good as IB's, and you can get all the futures data you'll ever need for under 70 dollars or so. It goes back longer than 10 years usually (depending on the symbol).

Hi Mr Soros,

Your idea is great. Even then, there are a bunch of platforms that do what you want to do only (I agree) most of them are not as good as they should.

Forget client side automation (excel or anything else in that road). That is a loser game in the end of the track.

No serious strategy can be run in a fully automated way using a client side technology. Sooner or later things go wrong (in the technology side) and everything get f*cked up.

Best regards,

Horace

horace said:

Hi Mr Soros,

Your idea is great. Even then, there are a bunch of platforms that do what you want to do only (I agree) most of them are not as good as they should.

Forget client side automation (excel or anything else in that road). That is a loser game in the end of the track.

No serious strategy can be run in a fully automated way using a client side technology. Sooner or later things go wrong (in the technology side) and everything get f*cked up.

Best regards,

Horace

Click to expand...

Thanks for your comments Horace. Could you elaborate a little please. Why is the clinet side a loser game? What kind of things go wrong on the client technology side. It is a different perspective from the normal that strategies go wrong and cannot be automated.

I am under the impression that most of the technological issues can be dealt with .

Thanks

Mr Soros said:

Thanks for your comments Horace. Could you elaborate a little please. Why is the clinet side a loser game? What kind of things go wrong on the client technology side. It is a different perspective from the normal that strategies go wrong and cannot be automated.

I am under the impression that most of the technological issues can be dealt with .

Thanks

Click to expand...

Hi Mr Soros,

Well, it is indeed true that most of the technical issues can be dealt with, but it is very complicated and expensive.

Basically you would need to double your whole infrastructure and also add some components that figure out which part has gone wrong and use the correct backup.

For example: Imagine your app runs on a pc at your home office. Just to name a few (but not all) you have the following parts in te game: app, pc os, pc hardware, office network, DSL connection line & DSL connection service.

Let alone the rest of the infrastructure that starts on the internet and ends on the exchange using your broker. That is not part of this game for this discussion.

Now imagine all is ok and your app is short on whatever trade. Also, your pending closing orders are on your app because the pricing is not known upfront (like on trailing stops for example). Ok, now with an open position, your router simply hungs or looses the wan connection. Your app is pretty clever and detects that right away (lets say 1/2 minutes later). Somehow it sends a signal to whatever part of the puzzle to change to the backup connection scheme. Fortunately this second line is up and running and it only takes 2 more minutes to reconnect. Hopefully your app is inteligent enough to detect some bars are missing and somehow it manages to recover those bars and reconstruct the price history. Lets say this take 2 more minutes.

So, from end to end, you may have dedicated (if everything was ok) 8-10 minutes. If your are (un)lucky enough, your app triggered a stop loss in minute 2 and now (6 minutes later) the price is elsewhere and you have the position in your app unsynced with the real one in your account. That is, your app thinks you are flat but you are actually short with an open order waiting to be filled...

Summing up on caos, your app now opens a new long (since it "knows" your are flat). It sends the buy order just to close (on the real account) your open trade...

As you can imagine, this is a pretty ugly situation. I won't go on but could.

I have seen lots of problems like this happening with automated strategies even with people trading on virtual machines on hosted servers with UPS and 99.9% connectivity.

Believe me, there are zillions of problems that can occurr and most probably when they do, your account will get affected.

Even with server based strategies problems do arise form time to time (at least that is my experience -and I trade automated strategies for real each trading day-).

Hope this helps.

Horace

What is the difference between "virtual machines on hosted servers with UPS" and running the systems from home? A laptop at your home (no need for UPS, because of its battery) in my opinion is just as good as the server I am renting for 250 dollars a month. It's better actually, because that way no one else can see my strategies. The advantage of the server is that when I can go on vacation I can rely on it.

All the problems you mention can happen, but what counts is how often they happen, and the probability of things going wrong depend on how "high-frequency" and complex your automated system is. In my case the automated systems are simple, and I am running 120 systems on one excel workbook, and the systems open and close about 15 trades per day, and things only go wrong once every six months: when there's a power outage on the server that lasts longer than the UPS can supplement (30 minutes). If the server was a laptop we might never experience any problems, because its battery can last up to 8 hours (for example, the HP 6730B that I have here has an extra battery that can last up to 8 hours).

travis said:

What is the difference between "virtual machines on hosted servers with UPS" and running the systems from home?

Click to expand...

It's very little, that's what I say in my post travis.

Well, I don't know anymore if there's any disagreements, but I thought there was one on this: I think you can run things from home fine, and I thought you were saying you can't do it because there's too many problems.

Why don't you fix these problems, or hire someone who can ?

BeginnerJoe said:

Why don't you fix these problems, or hire someone who can ?

Click to expand...

Are you walking with a limp, Joe?

Pazienza said:

Are you walking with a limp, Joe?

Click to expand...

No. I am someone who can fix these kinds of stuff. So I am just curious why people wouldn't. If the profitability is impacted by failing systems, then fix them. I see no reason for client systems to fail as an inevitability. For instance MT4 is a typical client side execution platform. Does that fail often ? I played with it and it never failed. The data connection for charts is rock solid in MT4. There's no reason for any execution connections to be different.

BeginnerJoe said:

The data connection for charts is rock solid in MT4..

Click to expand...

In the real world things are slightly different.

the hare said:

In the real world things are slightly different.

Click to expand...

Ok, I use a real world real account with IGindex. Their web based execution program is a client side java program. It rarely failed to execute my orders. Is that real enough a world for you ?

I agree that some of them issues can hapen, but at the same time, as travis mentions , it is how often they happen. Nowadays the internet is isue is a smaller problem than it used to be and still happens on a home connection.

I would be looking to to have my trades run on a server to minimise the effects of this.

The issue with the open trade being recognised as flat one and a second trade being opened, i believe that can still be recognised before placing a trade. It could be possible to do this by:

1./ Checking the funds allocation - if the position was open when the server went down, then when connection resumes, the balance update would still show an allocation of funds to that instrument, so you could avoid the long.

Checking the current price against the entry price , this would be programmed for such circumstances.

Or possibly have the app send orders directly to the broker, as it detects the signal becoming weaker. This method may prove ineffective because the signal weakens often but the connection resumes. Depends how you play it from the, do you leave the orders with the brokers , or do you cancel them ?


I am not experienced in the programming side but am sure something along these lines can be programmed to work. Maybe you have tried and have better experience there.

I have seen log files that keep a record of whats been happening, not sure how effective they would be in keeping track of whats going wrong.

Here's my advice. If you can rely on a good connection, opt for running your own "server" at home, by using a laptop. A "server" is nothing but a computer, where you only run the systems. The many advantages of running it from home are these:

1) A laptop costs you 500 dollars (HP 6730B), it functions as a UPS, and lasts much longer (if you buy the extra battery, it lasts 8 hours or longer). If you rent a server, you spend at least 200 dollars per month, so you could buy a new laptop every 2 months. Cheaper and better. And if you ever stop, you're left with an extra laptop.

2) If you run things from home, you can do much more, check much more, see things more closely and see what is wrong, and don't have to contact anyone but yourself.

3) If you run things from home, you don't have to be afraid of people stealing your work -- which with automated trading is a major concern.

I would conclude that there are no doubts you should run things from home, provided you have a good internet connection where you are. On top of it, with all the money you'd be saving, you could use two internet connections, so that one is triggered when the other one is not working. TWS doesn't mind at all -- you don't have to restart it or anything. I've been running things from home for years and it was just fine. Now I run things from a server only because:

1) I trust the guy running the server (he's helped for years with the creation of trading systems anyway, so if he wants, he can steal all the systems).

2) the guy running the server (from his home, in the closet) is so good with TCP-IP and hackers and anti-virus software (he runs an ISP company) that the systems are safer at his place than at mine. On the other hand, having to VNC to it, they might be safer at my home, because I would not even need VNC. No, but then again, my collaborators have to connect, so I would still need VNC, so it is indeed safer at his place.

3) the server is a good backup for me, so if there's an earthquake where I am or a fire in the house, there's still a place where everything is safe.

I would never run a server if it weren't at his home. I would not trust any rented server, ever. If he didn't do this for me (of course I am paying), I would do it from home.