Very much in agreement with the above. I use NinjaTrader 7, on which I have developed automated systems which trade 24/5 with very minimal manual operations. I find C# the very best programming language hands down, it's been instrumental for me in developing the operations support layer of my systems (including a number of workarounds for NT7 limitations). WFA is no panacea, a waste of time imho - I use statistical analysis to research & qualify trade setups (null hypothesis testing, etc). Currently all of the stat.ana done in Excel, as well as MC sim for which I use the Yasaii add-on.
The real question should be 1) Do you want to figure things out and explore concepts and ideas and have lots of fun designing trading mechanisms OR 2) Do you want to generate trades that turn a reasonable profit. If 1) above then my god, yes, do lots of 3D graphs, get 5 monitors, program 27 moving average and RSI based systems, read lots of investment news stories, learn to program in Tradestation and 6 other languages, think, wonder, test 4 new indicators per week, when those don't work invent your own indicators and give them cute names, .....57 other things.....then take a break ...then start over again. If 2) above get some free software that obliviates or automates the smattering of useful elements in 1) above. The labels associated with these software tools include predictive analytics, deep learning, modeling, support vector machines, machine learning. (Orange, RapidMiner, etc) Think Deep Philosophical Thoughts about the nature of changes, cause and effect, the arrow of time, study Plato and Werner Heisenberg, decide what C.G. Jung's collective unconsciousness has to do with markets that reverse direction for no obvious news event, learn why the Proof of Bell's Theorem is probably the most significant event in history. Postulate and populate and n-dimensional data matrix built to on price action transformed by your Deep Thoughts. Feed the Matrix into the software and let it run....it might take hours or days for the learning to happen. When the machine is done examine the Validation set carefully for statistical characteris that indicate stability. Restructure the Test, Training and Validation set historical time slices and repeat the learning process. If good do walk forward dummy trades. If good test the entire process on a different market. If good then go trade actual money. Note: In 2) you will have a Black Box. You will not know nor can human consciousness grasp why it works...too many things, millions of relationship between thousands of components. If psychologically you need to know WHY then stick with 1) above.
With Deep Learning enjoyed the huge success today, it's a shame it cannot be exploited in trading practice. In image object recognition, identification and classification, deep learning is now superior to human's ability to perform these tasks. At a very minimum, trader should not need to study candle stick anymore. DL can pick the type of chart pattern faster and broader than any trader can do. What's more interesting is the trading signal generation, if it is not based on chart pattern, how to apply the powerful DL to the problem domain? In other words, we need to transform the trading signal problem to the problem where DL shines, which is classification.