Optimization backtest and forward test correlation

 
Hi,

I have a strategy and I optimize it for a better input parameter. Now I optimize + backtest + forward test, supposedly the result of OptimizeA parameter was the best in backtest but on forward test OptimizeB parameter was the best. While backtest shows best parameter on the historical price and forward test shows us if that parameter has an indication of over-fitting (robustness), from those result which optimization parameter should I use?