how far back a trading strategy should be profitable to be valid ?

New comment
 
Hi, i got a few strategies developed and coded by me, i test them to be profitable for the last 7 years (2018-2024 included). i now have a strategy in my hands that is decently profitable 2016-2024 but is not profitable 2012-2015, is that a reason to not take it into consideration? In my mind is the fact that back then the trading conditions and the way the market moves were very different . what is your opinion? 

Thanks!
 
metinique:
Hi, i got a few strategies developed and coded by me, i test them to be profitable for the last 7 years (2018-2024 included). i now have a strategy in my hands that is decently profitable 2016-2024 but is not profitable 2012-2015, is that a reason to not take it into consideration? In my mind is the fact that back then the trading conditions and the way the market moves were very different . what is your opinion? 

Thanks!

Id say you could borrow from the neural net school of thought and switch the requirement like this :

-> For a strategy optimized from 1 year ago till the 1st of the current month (reoptimization each month) is that strategy profitable in the current month?

Repeat this starting from the month you can have a full years worth of data onwards.

So you'd be optimizing the optimization method essentially

How would that work on the tester : 

-You set a years worth of data optimization enabled backtest.

You take the best performing set and test it on the month AFTER the end of the optimization period.

-Repeat until "it looks like it gets it" most of the months tested

 
metinique:
i now have a strategy in my hands that is decently profitable 2016-2024 but is not profitable 2012-2015, is that a reason to not take it into consideration? In my mind is the fact that back then the trading conditions and the way the market moves were very different . what is your opinion?
This is certainly not a reason to abandon the strategy.
 
I would do as Lorntzos has suggested and that should give you a cloer look at both the trades at the period when the strategy isnt great, AND the statistics and metrics in the backtester will give you better idea what happened during that period to make the strategy not so good. Then, you can alter, modify, optimise the strategy so that it is more flexible and able to adapt to a similar period in the future.
 
Unless it’s a strategy with dynamically adjustable parameters, don’t expect it to adapt to the market conditions forever.

 
In my opinion the best approach is:

- Create a strategy on a previous date interval (for example 2019-2021) - this is where you create your strategy
- Validate the strategy and make optmizations if necessary on another period (2022-2023) - this is where you dont change your rules, but can make some parameter adjustments
- Confirm results on last period (2024) - this is where you actually test

If the results on 2024 are similar to 2022-2023, you should have a solid strategy.
 
metinique:
Hi, i got a few strategies developed and coded by me, i test them to be profitable for the last 7 years (2018-2024 included). i now have a strategy in my hands that is decently profitable 2016-2024 but is not profitable 2012-2015, is that a reason to not take it into consideration? In my mind is the fact that back then the trading conditions and the way the market moves were very different . what is your opinion? 

Thanks!
It is good from 2016 onward and I focus as market evolve cuz old data not always relevant. But do you think 2012-2015 still reflect present conditions?
 
Matthew88819 #:
It is good from 2016 onward and I focus as market evolve cuz old data not always relevant. But do you think 2012-2015 still reflect present conditions?

of course not , 2020 changed almost everything when it comes to safe heavens , indicies and even currencies . The volatility is different and also the tendency of some instruments, but nobody says we can't go back to that regime of trading that was present in the past.

 
Rafael Grecco #:
In my opinion the best approach is:

- Create a strategy on a previous date interval (for example 2019-2021) - this is where you create your strategy
- Validate the strategy and make optmizations if necessary on another period (2022-2023) - this is where you dont change your rules, but can make some parameter adjustments
- Confirm results on last period (2024) - this is where you actually test

If the results on 2024 are similar to 2022-2023, you should have a solid strategy.

great idea , thank you!

 
Rafael Grecco #:
In my opinion the best approach is:

- Create a strategy on a previous date interval (for example 2019-2021) - this is where you create your strategy
- Validate the strategy and make optmizations if necessary on another period (2022-2023) - this is where you dont change your rules, but can make some parameter adjustments
- Confirm results on last period (2024) - this is where you actually test

If the results on 2024 are similar to 2022-2023, you should have a solid strategy.
I'll give it a try, thank you.
 
  • I never used optimization mainly because it assumes the underlying logic is always there and you just need to adjust parameters for each period of time. While this is not the case necessarily. 
  • However as long as the strategy is designed to be general(for example a simple momentum detection rule and a simple trailing stop option) you can expect this to apply to any period of time... But even in this case you have to know when it is safe to apply the strategy by means of some other knowledge like fundamental analysis... 
  • Again if you really know enough that there is a momentum in market you usually go for holding the asset or take quite long term position is FX pairs... 
  • It looks like trading is nonsense and the EAs are for selling in Market and not using.
123
New comment