There should be a way to optimize my algorism, as my code hasn't been designed to be efficient wise. But I would take a lot of time optimizing the code, which requires big chunk of time and mind-soaking, which I am not ready at this moment. Instead I distributed the calucation almost evenly to 3 seprate server using 3 distinct network. The results come together hopefully in the very early morning, so I could have enough time making decisions before the stock opening.

Proud to announce that the third svr is on a outdated android phone, still they are great for linux in terms of CPU and RAM used. Maybe I should turn more outdated devices into this low cost, low consumption svrs.

I think till now the prediction of the index is accurate, it is just that I do badly on choosing individual stocks. As such, at this time, I probably would have a better chance play the index related ETF. They are much steady and more close trend to my kdj and macd index. But I still need to look into that...

So 50 percent on ETF and 40 percent on ZN's recommendation and 10 on my machine results. Got stick to this and make records.

Until the 10 percent really counts.