Weekly and Monthly Trend Following Strategies

[geektrader 10]

Hello,

 

I would like to hear from experienced system developers about common pitfalls to be aware of during development of strategies on high time frame like weekly and monthly.

 

My concerns are stemming from the fact that since the number of data points reduce drastically, and number of trades generated being even smaller, do the backtesting and optimization loose their statistical significance?

 

For example, simple MA crossover strategies on Weekly time frame on most futures data I have checked have given good results on 20 years of data. But they fail in WFO runs..

 

My dilemma is trying to estimate whether strategy is still worth exploring or not. One one had 20 years seems good amount period to cover different regimes while optimization and I get nice looking equity curves within my tolerance levels of drawdowns and distribution of returns.

 

However, if I do walk forward optimization (in Tradestation) over 10 time periods during same 20 years, the results vary a lot.

 

I am suspecting that during walk forward optimization, I don't have statistically significant number of data points (in terms of price as well as number of trades) and hence the strategy fails overall in such run.

 

I hope I am making sense here. Looking forward to hearing from you all.

[Obsidian 10]

Hello,

 

I would like to hear from experienced system developers about common pitfalls to be aware of during development of strategies on high time frame like weekly and monthly.

 

My concerns are stemming from the fact that since the number of data points reduce drastically, and number of trades generated being even smaller, do the backtesting and optimization loose their statistical significance?

 

For example, simple MA crossover strategies on Weekly time frame on most futures data I have checked have given good results on 20 years of data. But they fail in WFO runs..

 

My dilemma is trying to estimate whether strategy is still worth exploring or not. One one had 20 years seems good amount period to cover different regimes while optimization and I get nice looking equity curves within my tolerance levels of drawdowns and distribution of returns.

 

However, if I do walk forward optimization (in Tradestation) over 10 time periods during same 20 years, the results vary a lot.

 

I am suspecting that during walk forward optimization, I don't have statistically significant number of data points (in terms of price as well as number of trades) and hence the strategy fails overall in such run.

 

I hope I am making sense here. Looking forward to hearing from you all.

 

just asking out of curiosity, what is your priority? do you look for more profit or less drawdown?

[geektrader 10]

Honest answer I look at both and more..

 

distribution of return of lifetime of simulation, annualized returns, average/max drawdowns..

 

you can't ignore any of these IMO.

 

 

just asking out of curiosity, what is your priority? do you look for more profit or less drawdown?