Your programming ideas?

[superfuego]

Hey there,

If you could program, what would you write?

Imagine: You have near instant access to real time data. You can very quickly do any kind of analysis on that data. You can near instantly place an order based on the most recent data or based on analysis of that data.

What algorithms, methods, or strategies would you want to put into action?

I'm a physicist, not a trader, and understand finance in very simple terms. If you could explain the fundamental reasoning behind the methods it would help me to understand.

Nelson

 

[pedro01]

Hey there,

If you could program, what would you write?

Imagine: You have near instant access to real time data. You can very quickly do any kind of analysis on that data. You can near instantly place an order based on the most recent data or based on analysis of that data.

What algorithms, methods, or strategies would you want to put into action?

I'm a physicist, not a trader, and understand finance in very simple terms. If you could explain the fundamental reasoning behind the methods it would help me to understand.

Nelson

Write something to tell you which of the 8,000+ instruments out there that are setting up an opportunity.

Then look at that instrument & make the final decision to place a trade or not MANUALLY.

Or you could go down the road of your fellow physicist Ehler (as I did) and realise none of the cycle indicators are predictive as per the works of J M Hurst who invented cycle analysis. Extending the cycle out to the future is not that hard but none of the cylce indicators I have seen output a prediction of the next time & price of the next cycle turn.

Until you have a predictive indicator, you will just be following the crowd IMO.

 

[dcraig1]

I reckon Pedro's pretty much right about this. Despite denials, many of the so called "systems" are based on the assumption that there is a hidden signal in a price time series and if you can discover it, then you are on the way to untold riches. This assumption is probably wrong.

IMHO, he is also right about scanning/screening. If you look hard enough and with the right tools, you should be able to find an instrument that is virtually "locked into" a very good trade setup. How to go about this is another matter entirely. You will find virtually no serious discussion of this on this board, but plenty of agonizing over "pin bars' and such like. This tells me something.

I'm not a physicist, but I do have a BSc in physics and have forgotten most of the physics I ever knew, which wasn't much in the first place. I don't consider this to be a loss as far as trading is concerned. Software development skills are however, very useful.

I also think that people who have ambitions of developing true high frequency trading systems in their garage are being unrealistic. The competition is fierce, much, much better financed and it all probability smarter. You can't win the latency war.

 

[dcraig1]

Replying a bit more directly to the OP, here is one idea

Most (it not all) of the free stock screeners on the net only support screening ie AND together a bunch of relations (technical and/or fundamental) and out pops a list of stocks. I've written my own screener (in Java) that both sorts and screens on both technical and fundamentals. You can do things like:

* exclude stocks with price < 5
* exclude stocks from worst performing industries/sectors
* from that result set find 200 stocks with highest sharpe ratio over 100 bars
* from that result set find 100 stocks whose earnings exceeded analyst's forecast by the greatest amount on the most recent earnings date.
* from that result set get 50 stocks with lowest price to sales ratio
* from that result set get 25 stocks with lowest price to free cash flow ratio
* from that result set get 10 stocks that have lowest RSI

You get the idea. It's a kind of "query language" for finding possible trade candidates, even if there is no language - just a GUI. The example only partially does it justice, but it does show that there is considerable scope for finding a good set of instruments that can be then further examined for possible trade setups.

The fundamental stuff may not be very significant for daytrading, but the technical stuff is. In my implementation, the technical stuff can be applied "real time" as well.

Be aware, that it's still quite hard to find an automated market beating "edge". Once you have found your instruments, you still have to trade them.

 

[superfuego]

I've spent some time working on automated market beating, but I don't know if that road will work out or not. People with experience tend to think it won't, and I generally trust them.

However I love the markets, and trading, and am curious what other ways I can use the skills I have to trade. I was a profitable day trader for a time just by looking at charts, I didn't really know what I was doing in terms of any formal language.

I like the idea of a market screener, that's excellent.

Nelson

 

[pedro01]

At the moment, I am actually developing a system with a friend that looks for specific price action across a scanner.

As well as on line scanners like FinViz, you can code scanners with things like TS radar screen and TS Scanner to look for very specific things such as S&R, price action etc. - anything you can imagine.

For instance, on Mr Charts thread, I took the price action he was looking for and programmed it on my scanner so that those compliant filtered to the top - giving me a few trading opportunities out of more than 500 stocks..


http://www.trade2win.com/boards/us-stocks/38036-how-make-money-trading-markets-11.html#post818908