Originally posted by Thanky
Yes agree about
Dave's site, good intro videos and also some other content about back testing ideas from famous books and the like - good prompt to think about back testing discretionary ideas or examples we come across to see if they actually work in practice! Also that ideas / edges aren't constant over time. One of the reasons I want to run multiple systems.
In terms of ideas, I have a few, from discretionary observations I've made watching markets but in terms of first steps for myself, I would be more interested in just basic backtests of systems we find online or in books, and perhaps play around with a little to spark some discussion. If we can find examples like the Mean Reversion strategy from
Marwood's site that would be great for me.
I'm also really enjoying the podcast Sci
shared - so will look to pull ideas from authors there too.
I'm interested in testing systems that have some context around at the beginning so I appreciate some of the intricacies of where I may be over confident where I shouldn't be and the like.
Definitely happy for more advanced or greenfield test development too, but as I'll be learning the programme at the same time would like to make my life a little easier with a resource to bounce off for the first few. Happy to also do that in my own time but I suspect even collaborating on more basic systems will likely throw up a few discussions like the use of leverage that you ran through above.
Hi Thanky - Yes I agree with all the points you make. I just had another look at Marwood's MRS article and there's a lot of work to do even when testing just a simple system! That's probably why I've been avoiding doing this type of testing, although I know I have to do it prior to going live with a system. I think it will be an excellent exercise for both of us and anyone else who wants to learn how the professionals develop systems. In the past I have coded system rules that I find on-line or in books, I call them "Key Ideas", and then have a look at them on a selection of charts. I find myself rejecting a lot of ideas and then searching for better ones or trying to come up with what i consider better indicators to incorporate into a system. Generally i go around in circles for a while and then give up. When you are looking for a system with say a 10% edge and acceptable drawdown I believe/know that the only way to select or develop such a system is via rigorous testing. Simply looking at a few charts to see how the signals work isn't good enough.
The MRS rules Marwood presents are pretty simple and will be easy to code in Amibroker. I am aware of other MRS's that we can consider as well. I'll have a look around and see what I can dig up. I guess a good start will be to start answering and documenting some of the questions presented in the article.
1. What markets do you want to focuson?
2. What timeframes do you want to test?
3. What ideas will you test?
4. What kind of tests will you run?
5. What parameters will you test?
6. What metrics will you use to measuresuccess?
7. When will you give up on an idea?
We really need to answer the first 4 questions initially and then perform a "Sanity Check" on the system.
I know you have an initial challenge of setting up your software. If you do select Amibroker I am happy to help in any way I can. Let me know when you are ready to start system testing. In the mean time I'll have a look for other MRS's that we can consider.
p.s. we can share system documentation etc via email if you like - Wilf
(20min delay)
