Backward Optimization

[Deleted]  
Theoretically speaking is it logical to say use a recent time period for in sample optimization and then use a backward time period for out of sample testing. For example optimizing a system on 4H data for 2012 and then use 2011 H4 data for out of sample testing. Will the results of the system be credible if it is profitable from 2011 thru to 2012?