Based on the preliminary result from yesterday, I've done further testing and optimisation today. Overall, there are 8 more products that could generate positive return which were not included in the portfolio yesterday, as they were not profitable.
The testing period, including in-sample and out-sample periods, has been kept the same so we can make direct comparison.
Personally, I can't decide which portfolio is better as it seems the first one has better statistics overall but I like the short drawdown length for the second portfolio. I'd still throw both of these strategies into demo accounts and see how they perform.