Backtesting Software

[runmael]

Hi everyone, I would like to resurrect one question which was asked couple of years ago, I guess this situation changed a lot till today.

So, for all those which backtest their ideas before committing real money to them, can I ask what methods & software you use to perform this? I am interested in learning drawbacks of various available software and see if there is room for improvement.
Thanks :clap:

 

[dbphoenix]

It's never changed for me: bring up a chart and watch price move, either in real-time, delayed, or replay. Make extensive notes regarding activity, pace, extent, trending, ranging, where changes -- particularly breakouts and reversals -- occur. Consolidate the data and put together a tentative trading plan.

No special software needed.

Db

 

[malaguti]

Hi everyone, I would like to resurrect one question which was asked couple of years ago, I guess this situation changed a lot till today.

So, for all those which backtest their ideas before committing real money to them, can I ask what methods & software you use to perform this? I am interested in learning drawbacks of various available software and see if there is room for improvement.
Thanks :clap:

I've tried a few..
esignal, metastock, omnitrader, amibroker

i found esignal and metastock were only good at backtesting individually. (when i tried them that is, a few years back now). these were no good as you need to be able to test a theory across all different types, not just curve fitted to one.
amibroker and omnitrader are much better at these. my preference is omnitrader and still use it to backtest only

i haven't tried wealthlab or right-edge which are supposed to be the best at what they do. they all require coding ability though so be prepared to put work into them. all depends on what you want to do to be honest

 

[runmael]

What about OptionVue and Optionnet Explorer? I hear about them too, anybody used them, what are their drawbacks?

 

[Big_P]

I have a view on this ... And I have to agree with dbphoenix.

Scenario 1 - Punch all of your variables into a programme. Tweek them to your heart's content to bring the best returns. Couple of hours work and you have a fool-proof system to make money from trading, right?

Scenario 2 - Bring up a chart. See what's going on in real time. Take notes. Take more notes. Investigate. Be curious. Go back and watch the same thing again. Take notes. Tweek. Develop a trading plan.

Both approaches *in theory* would give the same results. However, consider this ...

You have a trading plan through scenario 1. You start trading live and you take a run of losses. Are you really going to trust your plan from the 5 minutes work that a backtesting programme has churned out?

Same thing with scenario 2. If you TRULY understand your plan, the percentages, the fact that a string of losses will occur knowing that in the longterm you have an edge... then that will give you courage to continue.

Only watching your chosen market until your eyes bleed will give you the confidence to trade your plan without any psychological barriers.


Just my 2p ...

 

[userque]

Hi everyone, I would like to resurrect one question which was asked couple of years ago, I guess this situation changed a lot till today.

So, for all those which backtest their ideas before committing real money to them, can I ask what methods & software you use to perform this? I am interested in learning drawbacks of various available software and see if there is room for improvement.
Thanks :clap:

I simulate how the system would have performed historically using Excel, Excel VBA, Excel Solver, and likely Python in the future.

 

[malaguti]

I have a view on this ... And I have to agree with dbphoenix.

Scenario 1 - Punch all of your variables into a programme. Tweek them to your heart's content to bring the best returns. Couple of hours work and you have a fool-proof system to make money from trading, right?

Scenario 2 - Bring up a chart. See what's going on in real time. Take notes. Take more notes. Investigate. Be curious. Go back and watch the same thing again. Take notes. Tweek. Develop a trading plan.

Both approaches *in theory* would give the same results. However, consider this ...

You have a trading plan through scenario 1. You start trading live and you take a run of losses. Are you really going to trust your plan from the 5 minutes work that a backtesting programme has churned out?

Same thing with scenario 2. If you TRULY understand your plan, the percentages, the fact that a string of losses will occur knowing that in the longterm you have an edge... then that will give you courage to continue.

Only watching your chosen market until your eyes bleed will give you the confidence to trade your plan without any psychological barriers.


Just my 2p ...

can't disagree, but there is a 3rd scenario (probably more i'm sure)..
you spend your days and days watching, taking notes, developing a plan just to find your system provides no edge whatsoever.
that backtest at least tells you "i may be onto something"
backtesting doesn't have to be your final system, it can just as easily give you something to work with..all that time saved
at least thats what i use backtesting for

like I say I agree with you. eyeballing just can't be beaten, but the time saved to quickly test a theory is in my opinion priceless

 

[options-george]

For manually backtesting a trading strategy - i.e. uploading real historical data for an instrument, and then rolling forward one candle at a time and opening and closing trades - I would recommend ForexTester2, costs $200 for a one-off and has excellent customer service (I have used them for 5+ years now).

This is much slower than other backtesting methods, but it does allow you to effectively paper-trade your strategy. And the paper-trading would be significantly faster than paper-trading on a real-time basis.

 

[Big_P]

can't disagree, but there is a 3rd scenario (probably more i'm sure)..
you spend your days and days watching, taking notes, developing a plan just to find your system provides no edge whatsoever.
that backtest at least tells you "i may be onto something"
backtesting doesn't have to be your final system, it can just as easily give you something to work with..all that time saved
at least thats what i use backtesting for

like I say I agree with you. eyeballing just can't be beaten, but the time saved to quickly test a theory is in my opinion priceless

A very solid point. And I don't disagree with you.

But, it's only by eyeballing a screen for hundreds/thousands of hours that you notice nuances in your chosen market which might give you the feeling "I'm onto something" ...

Backtesting at this point would be a great tool to re-inforce your findings.

But I'd argue that using backtesting to discover a new method would be completely ineffectual.

 

[runmael]

I see the point in all methods, probably the best approach is to combine them. I like the @Big_P quote about really understanding your strategy!

But how to see its performance so we could get a better understanding of it? I see here the space where backtesting might come.

What are some of the features you would like to see in your perfect backtesting tool?