You’ve put it altogether.
Now let’s see if she flies.
And if so, how far?
Episode 47’s question is from DeMarcus
“Do I backtest my entire system the same way I backtested my confirmation indicator?”
DeMarcus from Kissimmee, FL
You can listen to the episode here, or as always, continue reading on
How It Was
In Episode 21 of the podcast we talked about how to backtest a single indicator for the purpose of making it your number one, your confirmation indicator, which the rest of your signal would be based from.
This did have some moving pieces to it, but was fairly crude in its application.
All you wanted to do was take one indicator and have a race to your first take profit or to your stop loss, whichever occurred first.
We’ve added a lot since. Things had gotten more complex, we went slow so it doesn’t seem complex now, but it’s apples and oranges to what we were doing 6 months ago.
Now that you have a completed system, you need to see if it’s worthy of taking up 6-12 months of your time as you forward test it out.
Believe it or not, the whole process for backtesting has now gotten quite a bit easier.
How It Is
The steps are fairly simple now.
1 – On your chart, select any currency pair
2 – Go back in time as far as you have chosen to backtest. If this is 3 years, at the time of posting, you would go back to late May, 2016.
3 – Take every signal your algorithm allows, set up your risk management, and play it out in real time.
4 – Record the results every time a trade is closed out
5 – Repeat with the other pairs
6 – Total your overall average yearly ROI by dividing the results by 3 in this particular case
** I got some pushback on the yearly ROI saying it’s not accurate because I’m not accounting for compound interest. Remember though, we just want the percentage here. Yes, your money will compound, but we’re testing for the overall success of your system, and nothing more. Compound interest is not a factor here.
This Will Not Be 100% Accurate!
Your actual real-time results will probably be worse, especially at first, and this is expected.
You are not accounting for having multiple signals at one time, and having to choose between them.
Trading in real time is completely different than backtesting. Even on demo, emotions are in play.
This downturn in results will likely happen, but hopefully with experience, and my trading psychology playlist, you will overcome this, and results will go back up.
You must factor in news avoidance here, because it’s very real and can skew your results pretty sharply if you simply skip this part.
We spoke about this in the Forex News blog, if you need instructions on what to avoid and how.
For The Sake Of Speed
This is not as laborious of a process as people first think it is, but it does take time to do this over 28 currency pairs.
To make things quicker, you may just go ahead and backtest your system on 5 currency pairs just to see if it’s generating plus money. If it’s not, you can likely abort and start assembling a new one.
The pairs I recommend you try this on are…
This gives you enough of an assortment that covers every currency of the 8 majors, and also gives you a sample size of pairs that move for different reasons.
If it won’t work here, chances are it won’t get dramatically better over the other 23 pairs either.
This is good for systems that may just require a tweak in the setting here or there to really make a difference in your results.
This is a crucial part of the equation.
If you get lazy or skip over steps here, you’re building a house on a foundation that is cracked at best, and unstable at worst.
The next 6-12 months of your life, and possible the next 30-40 years of your life depend on how meticulous you are in this stage of the game.
So at this point, anyone who cuts corners on this part is somebody I’m secretly rooting against.
I guess that’s no longer a secret anymore.