Pro's and Con's of Backtesting - your input please

[charliebob]

Hi,

I am working on a report for my colleagues and subscribers on the topic of backtesting and system analysis, but there is debate on the value and validity of backtesting.

I would appreciate it if you would post your thoughts on backtesting, particularly suggestions, assumptions, considerations, benefits and possible liabilities.

Simply post your replies here for everyone to see. This will hopefully be an enlightening thread for all.

Thanks!

 

[Hoggums]

Hi,

I am working on a report for my colleagues and subscribers on the topic of backtesting and system analysis, but there is debate on the value and validity of backtesting.

I would appreciate it if you would post your thoughts on backtesting, particularly suggestions, assumptions, considerations, benefits and possible liabilities.

Simply post your replies here for everyone to see. This will hopefully be an enlightening thread for all.

Thanks!

I have been running a thread which is backtesting a few indicators.

Backtesting doesn't prove that your system will work on forward data - particularly if you've been through any kind of optimisation - however what it does prove is that if when you backtest your system and it doesn't make a profit then you do not have a viable trading strategy.

Personally - anyone who is trading an automatic - or semi automatic system - then backtesting is absolutely vital. However I think a lot of people neither have the resources or technical expertise to properly backtest a system. They assume because it's pip positive it's going to make them money - but they don't look at the drawdowns using money management. Then when this happens to them trading for read they get surprised and scared and pull the plug - something they might have been aware of if they'd done a full backtest in the first place.

Also - if you've done any tweaking to your system you should always test it on UNSEEN data (a forward test), again - if your system is successful - it doesn't mean it will be successful in the future - but if it is NOT successful then you need to look at your "Curve fitting"

Just my ideas...

 

[blackcab]

Hi,

I am working on a report for my colleagues and subscribers on the topic of backtesting and system analysis, but there is debate on the value and validity of backtesting.

I would appreciate it if you would post your thoughts on backtesting, particularly suggestions, assumptions, considerations, benefits and possible liabilities.

Simply post your replies here for everyone to see. This will hopefully be an enlightening thread for all.

Thanks!

Backtesting can't take account of how placing the simulated trade in reality would have affected the market. If markets are chaotic systems then this fact could render all backtesting utterly useless.