Trading systems currently working

[belflan]

I had a couple of previously "bullet proof" strats go off the rail a bit but this just highlighted the need to system diversification.
.

Hi TWI,

Do you find it best to use a rolling test period rather anchored date test period for optimizing? So that trading systems evolve with changes in market conditions

Thanks
Belflan

 

[luckyd1976]

I think it is misconception to think that algo strats do not work in this environment. Renaissance are +50% and Winton +20% which is a godo indication that algo's have actually shone through. I have found the trend followers and very short term momentum models have worked best over the course of the year. I had a couple of previously "bullet proof" strats go off the rail a bit but this just highlighted the need to system diversification.
Trends your friends and go with flow are the most imoportant advice from me.

My trenders are doing only ok this year. For me my bulletproof strats got whipsawed from my stops. In years past the stops werent an issue but hey thats the joy of trading.

 

[TWI]

beflan,

Any optimisation I do tends to be on an initialy anchored date set and maybe 2 different sets for each product but I do not see this as essential. All I look for with optimisation is instability. It is best way I think. Just need to test for the instability between the non dynamic variables in the system and as such find what are the most mid stable parameters. I do try to make everything as adaptable to volatility as possible but that is more a sizing issue.
I know there are many different ways to approach optimisation and some people mistakenly believe any optimisation is curve fitting but actually all systematic method could be called curve fitting. I have not read everything on the subject and my approach is pretty straight forward but for the most part it works and I have been able to find profitablilty for many years as a result so stick with it.
I am not perfect by any means and so I have had what I thought to be very robust models fail. It was not due to optimisation however, it was rather the underlying method that simply became invalid. It is important to realie when this happens and when it does, it is always expensive but that is part of the game and why diversification is so essential.

 

[alan5616]

Hi,

I'm interested in hearing from anyone using a commercial system that is still working (in the current conditions) and has also worked in the bull market we were recently in.

Could you recommend anything worthwhile for a novice who doesn't want to go through the learning curve of system development and is happy to pay for a system so long as it works?

Cheers,
PB.

Unfortunately, the vendor, of the best system I have ever encountered, has now withdrawn it from the market. He sold 50 copies and that was all. I'm glad that I was one of the 50