Thread: Overly Complex Coding and Trading Systems.

This thread has been made to discuss Overly intrie Coding and Trading Systems with a single goal in mind. Performance.

Everything, from the tricked out custom software to the aid of spiritual beings can apparantly assist you to trade. But will it assist you trade EVERY TIME?

===============================



my trading platform, complex or simple, Needs to maximize...

Win %
Typical win/loss
Position Size

And minimize. .

Drawdowns
stress
discretion

and take into consideration anything that can affect it is tested probabilities (news etc.. .)

If It meets those goals, I have every traders fantasy. So I'm excited to learn everything that's possible in trading
=====================================

I encourage everyone with any tendency to contribute anything they like. I encourage you to contact me should you with to function together for each others' common betterment and the communities trading.

When I am pleased with the outcomes of the conversation, I will be producing a free ebook and other information on my free site. That is, (like it's been said by another), to help make YOU money. I suppose if you did not desire your work in the publiion, you would not publish it anyway...

Since I have nothing I have forward tested (my easy system works, thanks rob) , I open the floor to the remainder of my coworkers here at FF...

Originally Posted by ;

my trading platform, simple or complex, has to optimize...

Win %
Typical win/loss
Ranking Size

And minimize. .

Drawdowns
stress
discretion

So just how do you measure these things? Is a system with a 15% yield with a drawdown that is 1 percent greater than a system with a 100% yield with a 20 percent drawdown?

What do you do for in sample versus out of sample testing?

Just how much data do you use?

What program do you use for your testing?

How do you account for spread in your testing?

Can you optimise? What metrics do you use for your optimising?

There is a lot to think about before you even put your first indior on the graph.

Simon

Originally Posted by ;

This thread has been created to discuss Overly Complex Coding and Trading Systems

Oh. Sorry. No donation to this one...

1 Attachment(s)

Originally Posted by ;

So how can you measure those things? Is a system with a 15% yield with a 1 percent drawdown greater than a system with a 100% yield with a 20 percent drawdown?

What can you really do for in sample versus out of sample testing?

Just how much information can you use?

What appliion would you use for your own testing?

How can you account for disperse on your testing?

Do you optimise? What metrics would you use for your optimising?

There's much to consider before you even set your first index on the graph.

Simon

I hope you don't mind ManinBlack if I answer a few of these.

Hello DrRock

I have enjoyed your posts previously. So allow me to know if that makes sense to you.

Drawdown and performance, this would depend on how much somebody else has in their account. The more you have the more you're prepared to risk for greater yield. But with a small account appropriate cash management and diversifiion will surely be the key.

A system that has a massive drawdown and a major return could nevertheless be traded. If you handle your money correctly and you've got other systems running concurrently in a portfolio to decrease the impacts of the drawdown.
And let's not fool ourselves that largest drawdown will happen again. Murphy's law works.

Sampling data would also be contingent on the number of transactions your system takes I like to observe a couple of hundred transactions in the out of sampled data.

The Software that I will be using if I post coding and such is TradeStations but because the name of the discussion is complex coding well. I figure people would be able to translate it to whatever appliions. They are using.

Optimization, a fantastic tool if you use it correctly as far as the matrix that I use. If the machine will only be traded on a single money I like using return on account. If the system will be traded around multiple monies I like using profit element. Both of these matrix also tend to lessen the drawdown.

Obviously appropriate optimization is a major issue so I will start with some sophistied software

The picture below is a 3-D chart of an optimization run the system seemed mighty promising when you looked at the accounts after optimizing but a glance at the chart reveals different.Keep in mind a strong system would have a level place where parameters could be changed without much change to the internet profitor or end results.

Originally Posted by ;

The picture [edit: above end edit:]is a 3-D graph of an optimization run the system seemed powerful promising when you looked at the record after optimizing but a quick look at the graph reveals distinct.

Clearly my eyes are getting old. I needed to have a look to see it.

Could I ask what software you're using to generate the graph?

Thanks Claude. I am really hoping you get your teeth into this thread.

How does your garph operate? Claude?

Originally Posted by ;

How can your garph work? ?

Hello and Wallker

The graft is produced by means of a software called 3-D view by rina systems.
TradeStations puts out a record of all of the optimization runs. Then I make a text file from this and feed it to the computer software. It requires the million creates and runs the graph.

Tomorrow if I get a chance I'd like to speak a little about neural networks because it has seemingly become more mainstream and acceptable,particularly since some members here are utilizing vantage point.

Very interesting stuff claude, look forward to more

Dr. Rock, I know essentially nothing about that which we discussing here. Forex is all about traditional and im new to trading in general.

Claude... what would you recommend to a trader who wishes to learn the sort of things that you do? Is there books. Gurus?

For today, I will be googling a Few of Those topics

stone on

Cool thread MIB, I think I will share some of the stuff I've been doing. Perhaps this ought to be more about cutting edge and not as much about excessively complex. I dunno

I'm convinced the markets are wavelike, and if you use any kind of oscillator, you concur. One thing that I mentioned before is wave patterns. When the market is level or range jump, it is really building up energy in the form of a expanding frequency spectrum.

To start, export a graph's information from MetaTrader. Just have a graph open, the perform filesave as, and it is going to save it as a .csv file. You may open this document in OpenOffice or Excel.

This document has a bunch of columns. Delete all but the final price to get 1 column of information that looks just like what a signal analysis tool is looking for.

I attempted this with some shareware which you may get at SigView.com. Once you install and operate SigView, it is possible to open a text document under the File menu. It will ask you to get a sample speed. I enjoy 1000 but it does not matter.

Now you'll see your graph, and it should look the same as a line graph of the information you exported from MetaTrader. From there, you can manipulate the information in some really cool ways.

I enjoy considering:

The frequency range, which reveals the peak frequencies that contribute to the sample. This helps expose some of the minor oscillations in the information.

The Time FFT, which does the above little slices over time. This can demone the growth and contraction of the spectrum. As a rule of thumb, when the spectrum is wide, there is a great deal of pent-up energy in the system.

Filters. You can filter out the overall trend, in addition to a good deal of the random noise. If done correctly (and I can't begin to explain to you the way to do this correctly), it ought to be possible to get a sense of how many of the important frequencies are pointing down or up.

So if you are into this type of stuff, give it a try. Perhaps you'll place some patterns that help explain some mysterious market move. If you're looking for a system that works, this ain't it

Trading systems vary in complexity from the simple to the inscrutable. Joe Krutsinger

A long time has passed since the first neural version by McCulloch and Pitts (1943). Now, however, with so many dwelling having a pc. These artificial neural networks. May be used by nearly everybody.

They're are nonlinear versions, loosely based on the way the human mind is structured.

I will attempt to skip a great deal of details because I don't need to over complie things. Just the Fundamentals.

Our simple artificial neural network will have tow inputs these inputs could be anything. They will initially enter two neurons every small bit of information will be providing a weightkind of like a level of significance from their they will go to three other neurons, which can be interconnected that information will be processed . This information will then be delivered to the output neuron the info will be processed and also more weights will be added. Then all these weights these levels of significance. Will be added up.

Based on this advice. We can either purchase or sell, the British pound as this will be running in a brute force optimizer. It will return that it's just lost or made x amount of dollars, so the neural network will have learnt that the solution that it just came up was garbage or good and out there. Another alternative will be tried by it.

The process will repeat till it comes up with the best solution to this problem.

This neural system is one of the most straightforward. It is only a foreword learning, meaning that the information only goes one way from input to output and then it tries again, to see if it can come up with a better alternative.

A slightly more complied neural system uses back propagation meaning information goes ahead and backwards the nerves would attempt to come up with the solution. The error rate could be calculated. The bigger the error, the nearer it is to solving the problem. This information would be passed back to the neurons. They understand that they're headed in the wrong direction or the correct one. ((We might look at this one as well))

Weight could be positive or negative. So that it doesn't take much time. We just use -1 ,0 or 1 because we've got nine areas where weight could be added and three unique weights could be set in at these points. This gives us 19, 683 distinct combinations. For a much more elegant solution. We could let it go from -10 to 10, etc, but that would be quite time tested in TradeStations.

Keep this as simple as possible. The only 2 inputs, we'll give it's a simple moving average of three instead of itself. Three periods ago.

The second input will function as simple moving average of three compared to itself five periods ago.

Recall these inputs may be anything, examples: the final price of gold. The RSI, ADX a different currency, and so on.

The only difference here is that as a system developer. We don't need to explain to it purchase if the moving average is greater than 1 period ago and lower five periods back at marginally higher than two periods ago. On and so Forth. This will all be figured out from the neural system. Along with the level of significance for every one will likewise be figured out.

Function for TradeStations name: Htangent

Input: x (NumericSimple), input signal to serve

NTerms (NumericSimple); # terms in series

Var: pi (3.1415926536),

Sum (0),

ii (0);

Sum = 0.;

For ii = 0 to NTerms Begin

Sum = Sum 1. /(Electricity(((0.5 0.5)*pi), 2) Electricity(x, 2));

End;

Htangent = Sum * 2 * x;

__________________________________________________ ___________________________

Signal for TradeStations name: Neural net simple

Inputs: synapse1 (0), synapse2 (0),synapse3(0),synapseA1 (0),

synapseA2 (0),synapseA3 (0),synapseB1(0),synapseB2 (0),

synapseB3 (0);

Var: inputneuron1 (0),

inputneuron2 (0),

Hiddenneuron1 (0),

Hiddenneuron2 (0),

Hiddenneuron3 (0),

neuronOut (0);



inputs

if (moderate(C,3) - moderate(C,3)[2]) gt;0 Subsequently inputneuron1= 1 inputneuron1 = -1;

if (moderate(C,3) - moderate(C,3)[5]) gt;0 Subsequently inputneuron2= 1 inputneuron2 = -1;



neural net

Hiddenneuron1 = Htangent(synapse1*inputneuron1 synapseA1*inputneuron2, 50);

Hiddenneuron2 = Htangent(synapse2*inputneuron1 synapseA2*inputneuron2, 50);

Hiddenneuron3 = Htangent(synapse3*inputneuron1 synapseA3*inputneuron2, 50);

neuronOut = Htangent(synapseB1*Hiddenneuron1 synapseB2*Hiddenneuron2 synapseB3*Hiddenneuron3, 50);



purchase or sell

If neuronOut gt;= 0.5 then

Buy next bar at High end;

If neuronOut lt;= -0.5 afterward

Sell next bar at reduced end;

__________________________________________________ _______

Neural internet are no longer costly you can have them running in Excel for approximately 70 to 150 bucks only enter all your information in Excel sheet and let it go to city. I just thought it be fantastic if you could see the coding process which lays behind.

Maybe next time we could take a look at the more complied neural net, where the error rate is calculated along with back propagation. And instead of it figuring out how to use something in real-time, we'll get it to predict the closing price for the next few days.