Well after one week, some intertesting results
In fact I run 2 tests,
Test 1 with the best parameters I had on the backtests results
Test 2 with the parameters I find better based on the analyze I did (Methodology on the CCI file )
And what you see is very interesting (Same algos of course, only the parameters change)
With the test 1, the results are not good because of a big drawdown -8 % after one week
But with the test 2, the results are much better (ok after only one week)
But remember that when you do a walk forward it takes the results like on test 1 !
So like I previously said, for me, I think more and more, that Walk Forward is actually not very useful because it chooses the results only based on Gain (Not the realty !) (You can choose other filters but unuseful too for me), and not on an indicator we wanted/created for example
So I think that if you have good results with WF, well it’s ok but if you have bad results it doesn’t mean your algo is bad
So now I will continue to let the algos runs and see when the crash, to estimate when it needed to repupdate the parameters with a new backtest
Bye