Are you relying on manual testing then to get these results? May through October is quite a long time to MANUALLY backtest this, it would be 6 months and if you trade on the M30 timeframe, it would take ages to test. Maybe you should tweak that a bit, how about you choose a month that was trendy and one that was choppy? Two different market types will yield best results so you know it will work in both types of markets. Also how it trades during news, concentrate on days where the market was going crazy because of the news.
Just my small input to help you receive better results