Advice on applying MOC systems to additional universe(s)

[Howard Lask]

I’ve been pretty busy with my day job for most of the year however I’ve been able to carve out some time for systems development more recently and will hopefully have some capacity into 2021.

I’ve recently added a short MOC system to my existing long MOC system – both trade the R1000 universe.

I’ve noticed reading through the forum that many are running R1000 and R2000 MOC systems alongside each other. I have a couple of questions/observations and would be great to get some advice on how others have approached this.

I’m going through the process of taking the R1000 systems and running an optimisation on number of positions based upon R2000 using existing parameters. Generally I am seeing that I need to increase the number of positions in order to contain Drawdowns and maintain MAR ratios. Nevertheless performance is generally inferior to the initially developed R1000 system.

Some decisions regarding process I am pondering:

[ol]
[li]develop and optimise the system for one of the universes initially and then apply to the other vs develop and optimise for R3000 and then implement for each universe vs some other approach that I have not considered![/li]
[li]Other than number of positions, use the same optimisation parameters for both universes vs optimise parameters for each universe independently?[/li]
[/ol]

Appreciate your suggestions and feedback….

[JulianCohen]

So you’ve opened a topic for a lot of discussion.

My two bobs worth:

First have a good look through Matthews (rather lengthy) posts on position size in R2000 and R3000 as that is the first consideration….whether your position size is small enough to make sure you are going to maximise your chances of getting fills in real life. This can be controlled via % of position size, number of entries and also with Price and Volume limiters.

Second point.
You can also consider creating a dynamic watch list in Norgate to access All the stocks in the NYSE and trade that…just to vary things up a bit

When I went through this process I ended up tweaking some parameters between the systems but I tried to keep it to price and Volume changes only. My thinking is that the least amount of optimisation the better. So R1000 I might run at $10-500 and R2000 at $10-$2000 as an example.

For me, I would like to see the system perform well enough with the parameters unchanged apart from Price and/or volume

Having said that sometimes a system just does better in one universe than another. So maybe it is worth taking the basics of the working system and changing something else apart from the parameters, in order to make it work. Maybe rank by a different method, use a MA of a different size, or include one if you haven’t already, or take it out if it is already in there.

So basically modify the system until you can find something that works, or you throw the computer out of the window….whichever comes first.

[Howard Lask]

Thanks Julian,

I have been following your suggestion to just vary price and volume – increasing price and volume levels does appear to improve the R2000 system and if I can make that work without adjusting other parameters then so much the better. I too would prefer to see the system perform with other parameters unchanged as that further validates it’s robustness.

Your suggestion to apply the system to NYSe stocks is interesting, I will have a look at that next…

[JulianCohen]

Let me know if you need any help. I either have instructions on how to do it somewhere or can make one up for you.

[Author]

Posts