Automated trading systems

Quote from dcraig:

Yes, but the opposite is also true. What if you encounter problems or limitations with TS (or any other product) that are a serious issue for you and THEY won't fix them in a time frame that's acceptable. You really are stuck then. A decent programmer should be able to fix any problem given the source code though it may require considerable effort. If you write it yourself, by definition you have the source code.

I'm not advocating learning Java or any other programming language but there are some advantages. The huge disadvantage is the time involved.

I havn't used TS so I can't comment but I have tried ESignal efs and I thought it was quite frankly, bloody awful. From that point on I use Java only. But I am a programmer, so it was not a difficult decision. For those that havn't programmed before, learning it will be a lot of work that is quite possibly not justified.

As for coding a few indicators, they are likely to be least of ones problems. All the common stuff like RSI, Stoch, MACD, Linear Regression etc are not difficult at all.
More...
Hi craig,

You sum it up nicely. Not wanting to exclude the (theoretical?) possibility of people making decent money of their TS or alike tools, I often wonder about it.

I kind of went through your experience and would compare developing trading strategies without programming litteracy as crawling in a 3 dimensional world on a flat 2D plane.

Of course 'picking up programming' is going to remain a pipe-dream for most who will try to get there quickly. It's like learning to play the violin or piano. Many try. Few bring it to a minimum level of virtuosity. Few can even imagine the work required to become a TRUE virtuoso.


Be good,
nono
 
Quote from dcraig:


I'm not advocating learning Java or any other programming language but there are some advantages. The huge disadvantage is the time involved.
I havn't used TS so I can't comment but I have tried ESignal efs and I thought it was quite frankly, bloody awful. From that point on I use Java only. But I am a programmer, so it was not a difficult decision. For those that havn't programmed before, learning it will be a lot of work that is quite possibly not justified.
--------------------------------------------------------------------------------
Ok, this question i had in mind always, if there are many C++,
java programmer out there that can write better, faster, more
flexible program than Tradestation, wealthlab,.......then why
they don't convert the source into some non-hardcore programing
application for every trader to use, and pocket $250 a month
for every user they get. TS plateform has 21,000 user today.
Please Give me a logical reason?
--------------------------------------------------------------------------------
As for coding a few indicators, they are likely to be least of ones problems. All the common stuff like RSI, Stoch, MACD, Linear Regression etc are not difficult at all.
--------------------------------------------------------------------------------
I think you didn't understand my question, re-read my previous post again?
More...
 
Quote from OptionScreen:

Quote from dcraig:


--------------------------------------------------------------------------------
Ok, this question i had in mind always, if there are many C++,
java programmer out there that can write better, faster, more
flexible program than Tradestation, wealthlab,.......then why
they don't convert the source into some non-hardcore programing
application for every trader to use, and pocket $250 a month
for every user they get. TS plateform has 21,000 user today.
Please Give me a logical reason?

I think you didn't understand my question, re-read my previous post again?
More...

Reproducing something like TS, WL, etc requires a hell of a time.

It took me 5+ years fulltime. A lot of learning is involved in finding the path that suits you best. Simply put, 'once you get there', you can do almost anything. You can't say that of TS, WL, etc. Let me quickly add that this may not be sufficient to make money. :)

nono
PS: it also makes me laugh with 'quants' (whatever this means), tinkering with or without C++.
More...
 
Quote from OptionScreen:

Quote from dcraig:


I'm not advocating learning Java or any other programming language but there are some advantages. The huge disadvantage is the time involved.
I havn't used TS so I can't comment but I have tried ESignal efs and I thought it was quite frankly, bloody awful. From that point on I use Java only. But I am a programmer, so it was not a difficult decision. For those that havn't programmed before, learning it will be a lot of work that is quite possibly not justified.
--------------------------------------------------------------------------------
Ok, this question i had in mind always, if there are many C++,
java programmer out there that can write better, faster, more
flexible program than Tradestation, wealthlab,.......then why
they don't convert the source into some non-hardcore programing
application for every trader to use, and pocket $250 a month
for every user they get. TS plateform has 21,000 user today.
Please Give me a logical reason?
--------------------------------------------------------------------------------
As for coding a few indicators, they are likely to be least of ones problems. All the common stuff like RSI, Stoch, MACD, Linear Regression etc are not difficult at all.
--------------------------------------------------------------------------------
I think you didn't understand my question, re-read my previous post again?
More...

I'm afraid running a software business is just a little bit more difficult than you imagine. Being the proud owner of a few tens of thousands of lines of code that mostly works is only a little piece of the story.

I understood your question. The point is that people developing their own stuff aren't trying to replicate TS or ESignal or anything else for that matter. They are developing something to do exactly what they want. If they have any sense, they are developing reusable libraries that can be used in future work as well.

There are many reasons one might want to do this. One is performance. You can't convince me that EL or efs or WealthLab will perform anything like decently designed Java or C++ code. Another may be fault tolerance and automatic failover. They could be many others..

As I said I'm not necessarily advocating learning Java, C++. If TS does what you want fine, but it is not the last word on trading systems development environment and neither is anything else for that matter - 21K users or not.
More...
 
Quote from OptionScreen:

Still you need to creat system variables to use them in
your function, like in easylanguage where you have
Vars, inputs, Truefalse/text/Numeric simple, series, ref,
ref arrays, while, for loops,...

How you will creat them? Because Easylanguage don't give
source code for these reserved words?
More...

???? loops?? arrays?? booleans? VARIABLES??that is part of coding in any language.
The "TS specific" stuff is very simple to figure out by looking at a EL manual and seeing what it means. Porting EL stuff is pretty much trivial. The time is spent on building the framework itself in which you plan on running those strategies :) Even that isn't difficult, just a lot more time consuming.

In response to the person who suggested it would take years to code up indicatiors: not close. You can do several in a single sitting if you have the formula's.

re: if one can build it, how come they don't sell it?
For starters, an automated system doesn't really need any of the bells and whistles of a "normal" trading platform. No charting, fancy gui's, etc. All it needs is a way to turn it on and off. Where as I'm happy to copy and paste back testing stats into a seperate MonteCarlo app I wrote, and dump the results into a csv file to view in excel, one could hardly market such a thing.
The point here is that what's required to make your own blackbox, even one that can run multiple strategies at once using a common framework, is not remotely the same as what one needs to compete against TS, etc. They are not the same sort of programs/products.
Most guys here with custom coded "boxes/bots/etc" could probably crank them out easily for others if they really wanted. I know a basic crossover type strategy would take me less then 10 minutes to build and have ready to run live against IB (unless it uses an indicator I haven't coded yet). The coding itself of this stuff is not difficult for a programmer - it's pretty easy compared to what we do during the day. The tough part about building a "white box" (a black box really means you don't know the logic behind it) is finding the edge you plan on automating :)
 
Quote from dcraig:

I'm afraid running a software business is just a little bit more difficult than you imagine. Being the proud owner of a few tens of thousands of lines of code that mostly works is only a little piece of the story.

I understood your question. The point is that people developing their own stuff aren't trying to replicate TS or ESignal or anything else for that matter. They are developing something to do exactly what they want. If they have any sense, they are developing reusable libraries that can be used in future work as well.

There are many reasons one might want to do this. One is performance. You can't convince me that EL or efs or WealthLab will perform anything like decently designed Java or C++ code. Another may be fault tolerance and automatic failover. They could be many others..

As I said I'm not necessarily advocating learning Java, C++. If TS does what you want fine, but it is not the last word on trading systems development environment and neither is anything else for that matter - 21K users or not.
More...

well stated...
 
Quote from OptionScreen:

Quote from dcraig:

Ok, this question i had in mind always, if there are many C++,
java programmer out there that can write better, faster, more
flexible program than Tradestation, wealthlab,.......then why
they don't convert the source into some non-hardcore programing
application for every trader to use, and pocket $250 a month
for every user they get.
More...
Hi dcraig!

Writing a custom application for myself is much different than developing a commercial product. (I have considered using a contract programmer in India to clean up my apps and turn them into commercial products... Might be interesting. Probably not, depends on the coder.)

There is no way to protect anything developed by myself that was really unique that would differentiate my application from anyone else out there. (A patent along the path taken by TT is a possibility. But an expensive long shot.) This means if I developed demand for my product, a larger company could copy and take my market, happens every day.

The Tradestation & eSignal market segment has so many users because of their data feed more than the strength of their application. I have an interface to IB for my application, but has anyone made a lot of money from an IB app? SierraCharts? NinjaTrader? Yeah some, but not the scale that will change your lifestyle.

One of my apps is a very nice RiskMgmt application that imports daily and monthly reports from IB and my Tradestation account. It has all of the screens described in "The Trading Game" with more stats and MonteCarlo screens. Great for a discretionary trader as well as a pure systems trader. Could I set up a web site and sell it? Probably. Other apps that are not as good are selling in the $150 range. How many customers for this type of application? Maybe 500. Ok, not pocket change, but after expenses, promotion costs, opportunity costs, etc. what is the net? It gets smaller fast. As soon as you start selling it and taking money for it people start asking for more features, bug fixes, advice, etc., etc., etc. The road to hell.

I cannot think of anything worse that selling a trading application and having 10 newbie phone calls every day because they lost money using my application doing something stupid in the market. ("Your call is important to us. Please leave a message at the tone....") Remember, no sofware application can be completely foolproof, as fools are so ingenious!

Software is a business, trading is a business. Both are difficult; easy solutions that are profitable are rare. Both markets are very efficient.

Good trading!!
More...
 
Quote from jason_l:

???? loops?? arrays?? booleans? VARIABLES??that is part of coding in any language.
The "TS specific" stuff is very simple to figure out by looking at a EL manual and seeing what it means. Porting EL stuff is pretty much trivial. The time is spent on building the framework itself in which you plan on running those strategies :) Even that isn't difficult, just a lot more time consuming.

In response to the person who suggested it would take years to code up indicators: not close. You can do several in a single sitting if you have the formula's.

re: if one can build it, how come they don't sell it?
For starters, an automated system doesn't really need any of the bells and whistles of a "normal" trading platform. No charting, fancy gui's, etc. All it needs is a way to turn it on and off. Where as I'm happy to copy and paste back testing stats into a seperate MonteCarlo app I wrote, and dump the results into a csv file to view in excel, one could hardly market such a thing.
The point here is that what's required to make your own blackbox, even one that can run multiple strategies at once using a common framework, is not remotely the same as what one needs to compete against TS, etc. They are not the same sort of programs/products.
Most guys here with custom coded "boxes/bots/etc" could probably crank them out easily for others if they really wanted. I know a basic crossover type strategy would take me less then 10 minutes to build and have ready to run live against IB (unless it uses an indicator I haven't coded yet). The coding itself of this stuff is not difficult for a programmer - it's pretty easy compared to what we do during the day. The tough part about building a "white box" (a black box really means you don't know the logic behind it) is finding the edge you plan on automating :)
More...
Jason,

I don't like the 'black box' terminology. It's like 'quant' and other vapor words. In one way you are right about the above, but that's not how things work.
(1) Indicator building: if all you have to do is code up a formula, that's easy. A huge amount of time goes in the design of a framework to use these things intelligently.

(2)The same with whipping up some 'black box'-code (sorry) and hook it up to IB or so. Few if any know exactly what they need to put into that boxie for it to start pumping money. Typically people don't, and only dream about it, especially those used to losing money.

(3) The purpose of the exercise is to provide you with tools to cook up profitable 'tricks' - call them algorithms to put in your 'black boxie'. This is, at least in my case, why I had to build an analysis environment better than TS, WL, etc. If you manage to get any success in this area, sticking on a bit of code to start trading via IB is indeed very simple.

(4) I don't believe that making things look so easy is very realistic. Don't forget that in this game, you kind of play against rather very smart competitors. You got to be at least as smart as them to simply break even.

nono
:)
 
Quote from nononsense:

Jason,

I don't like the 'black box' terminology. It's like 'quant' and other vapor words. In one way you are right about the above, but that's not how things work.
(1) Indicator building: if all you have to do is code up a formula, that's easy. A huge amount of time goes in the design of a framework to use these things intelligently.

(2)The same with whipping up some 'black box'-code (sorry) and hook it up to IB or so if you know exactly what you need to start pumping money. Most people don't, especially those used to losing money.

(3) The purpose of the exercise is to provide you tools to cook up profitable 'tricks' - call them algorithms to put in your 'black boxie'. This is, at least in my case, why I had to build an analysis environment better than TS, WL, etc. If you manage to get any success in this area, sticking on a bit of code to start trading via IB is indeed very simple.

(4) I don't believe that making things look so easy is very realistic. Don't forget that in this game, you kind of play against rather very smart competitors. You got to be at least as smart as them to simply break even.

nono
:)
More...

nono: I agree with everything you say here. Maybe this was lost in my ramblings, but my point was that coding the software is not the issue nor the holy grail to profitable trading - figuring out WHAT to code, ie your edge, is the biggest issue. next up is coding the execution platform, etc.. last and least significant is coding the indicators and rules to your system(s). It requires no special talent what-so-ever for a real programmer to actually code this stuff vs anything else they code for 2000+ hrs a year - the special talent is knowing WHAT to code :) A rogrammer w/o a winnig edge is just a programmer, a trader with a winning edge and no programming skills is still a successfull trader :)

From some of the questions I see being asked, I wanted people to realize that there's no magic to the coding or automating ideas. The REAL hurdle is coming up with a profitable strategy(s). I wasn't attempting to minimalize the effort of developing a profitable automated solution as a whole, but rather trying to impress on people that coding the indicators, etc, is really a non-issue in big picture. One should not think "if I could only get my own 'black box' coded up, I'd make some real money!".. Otherwise, every programmer would be rich :)

BTW: I find this term irritating as well. To other people out there who don't understand why it bugs some folks, "black box" is really just an IT term describing any system in which you can't see into - ie, you don't know the inner workings/rules/etc. A lot of what people here refer to "black box" systems, are really not. One can not code thier own black box :) But, if after coding it they were to lease it to someone else w/o divulging how it works, then to the lesee it would be a black box.
 
You must log in or register to reply here.
Back
Top