I think this is not possible to judge unless we know the whole development process. You could have introduced a bias during the strategy design but we can't possibly know from the code and results only.
The development process was actually quite short. All I did was experiment with dual SMA crossover on some pairs with not much success and then tried to simply enter one trade per day with stoploss and trailing in the direction that depended on the relation between the two SMAs. And of course I optimized the parameters. This resulted in the code above.
It turned out that this worked in the past for USD/JPY but not really for the others.
Anyway, the equity curve does look quite ugly. The ulcer index is unbearably high and the win rate way too low. I think this is psychologially untradable.
