Automated Trading Q & A

autotradingalgos

autotradingalgos

I thought it made sense to open a thread where I can try to answer those burning questions about Automated Trading / Financial Math. Particularly suited to those discretionary traders who are wondering where their money goes to :-). Feel free to post questions. Good trading to all
 
Hello, do you do use portfolio optimization?
Do you have any tips which way to look in financial math when looking for entry strategy?
Do you use technical analysis and if yes what kind of charts/timeframes do you use?
Thank you
 
Aha - great questions. Optimization? NEVER. Not for strategy formulation, or tuning a group of strategies into a portfolio. All it does is make the in-sample testing less relevant for the future. Curve fitting is a dangerous path and totally unecessary if you have enough "ok" systems in your portfolio.

Entry strategy- ok, well I look for basic ideas that make sense. If you buy on monday morning and sell on Tuesday morning (for example) what are the returns and odds associated with that strategy since at least 10 years across a number of contracts. Markets are run by people (and people program algos)... repeatable behaviour is what you are looking for.

I use technical indicators as filters for my trades and a basis for my stops. They are not the primary reason that I enter a trade. They are there for filtering purposes only. Regards and hope this helps
 
IAS_LLC said:
Isn't looking for the entries with a higher expected profit a form of optimization ?
More...

Thanks for the question. In the case of the strategy I trade I have always used the same entry point and have never looked at changing that to make potentially more profit. That would simply impose curve fitting on what is already a robust system. Regards and good trading
 
autotradingalgos said:
Thanks for the question. In the case of the strategy I trade I have always used the same entry point and have never looked at changing that to make potentially more profit. That would simply impose curve fitting on what is already a robust system. Regards and good trading
More...

Could you please explain how using statisticaly significant statistics is a form of curve fitting? I'm not talking about mining a data set and fitting a curve to maximize profits. I'm talking about only taking on trades which have E[] > 0
 
Hi IAS_LLC In your previous post I understood your

"Isn't looking for the entries with a higher expected profit a form of optimization"

to mean a modification of an existing entry looking for more profit. No you are absolutely right. Using statistics that produce a decent expectancy of profit in different markets conditions has no curve fitting aspects to it. When you start modifying things too much that is when it starts to creep in. That is a real temptation. Regards
 
Hi
I have recently started playing MT4 and wrote a simple trading algorithm as pracitice. Then I used MT4's Strategy Testing function to 'optimise' the parameters, trying to squeeze as much profit as possible from the algorithm. Then I read a few articles about curving fitting and why people shouldn't to do it.

I think what I have been testing is actually curving fitting and as you mentioned in the post, it is very temptating as the results look very good.

My question is: what should we take away from these back testings, if not a set of 'optimised' input parameters for the algorithm? How should we analyse the back testing results? Any recommendation of booking or reading materials on these matters are more than welcome.

Thanks in advance.
 
You must log in or register to reply here.
Back
Top