I cover all popular machine learning frameworks are going mainstream. I am anticipating that Tensor Flow will be the dominant player because of Google. We also talk about the future vision of integrating Redis and Motive Wave trading platform as well. Other programming languages other Python were covered including C++ and Java
I am quite surprised R is being promoted as a language ofr machine learning and data research. According to the latest Tiobe Programming Language Index, R is placed at 15 while other popular programming languages are in the top 10. The most notable is MATLAB where you would think it would wither to die. MATLAB has a lot of relevance still but this free R course might be of interested to you.
Here are some of the article highlights:
R burst into the scene with its powerful support for stats (linear and nonlinear modeling, classical statistical tests, time-series analysis, classification, clustering, …) and graphical techniques, but in recent year, it has gained a lot of popularity among Data Scientist and Data minors.
Along with Python, R is the second most popular language to learn the new world of Data Science, Machine Learning, Deep Learning, and Artificial intelligence.
Facebook will be pushing out a new version of PyTorch It seems to be the second most popular machine learning framework out there. It include Python machine learning native code.
Torch Script works one of two ways. New code can be written using the Torch Script language, which by design can compile readily to native code. It’s also possible to take existing Python code, decorate it with the @torch.jit.trace decorator, and have it just-in-time compiled to native code. However, this is not as effective as using Torch Script….
Another commonly used package, Cython, allows Python to be turned incrementally into C by way of custom syntax declarations. Cython can work with the whole range of Python and C types alike, as well as all of Python’s syntax, but Torch Script is restricted to operations on PyTorch tensors, integers, and floating-point numbers. And Torch Script can’t use constructions like exceptions.
Maybe I should I just stick with Tensor Flow 2 coming out in Dec or Jan? It seems to be gaining more popularity vs PyTorch. It should be just as easy to work with for us newbies.
I question the future
I would also like to predict this PyTorch scripting language may be tough to see in the long term. That is is of course with ‘let’s abandon everything’ Microsoft hat on.
Math is key sometimes to math. Guess what? I find the simpler, the better. If you know your difference between an simple moving average or a weighted average, you should be ok. You don’t need the complicated math as far as I can see. The complicated quant can be important for processes like machine learning, it can take you much longer to find the result.
Hi Bryan, I have read https://www.youtube.com/watch?v=A07CBHXq75A I was wondering, are there any codes on github where one could execute an indicator from MT4 or Tradingview via Java, or instead of executing orders, receiving buy sell signals via sms from an MT4 or tradingview indicator, via java, on an sms? How would you construct such indicator?
My answer is in the video but in a nutshell: always use the market data of where you trade or submit orders.
Everything described is in this future.
Also, in this video I addressed the changing situation of brokers due to their Python API support. One example is how Dukascopy seems to become irrelevant since more people want machine learning in their trading. It seems they seek Python for this. You can find all these answers towards the end of the video.
Java is not the way for algo trading and machine learning
if you are into Java and machine learning, you might want to hear what Saraj Raval who does a popular Youtube channel, he says that Java is an old language which he dislikes. He also compares it to when you learn first programming languages in your computer science class. I sort of agree considering where we are in the industry.
As an example, you could look the Alexa.com stats for Dkkascopy which has dropped from 30k to 45k roughly over the last year. Why the drop? I believe it is because the only API available for DUkascopy is Java with JForex. They may want to add Python at some point.
Speaking of Dukascopy and Python, you could check out Oanda’s ranking which is at 3.9k roughly since they added Python a few years ago. Also, they are much more popular than Interactive Brokers (12.2 ranking on Alexa) which has an added wonky Python support but still have C++ support as well.
C++ is the future vs Java for sure when it comes to machine learning and algo trading.
Why #Python over R for #algo#trading and #machinelearning. I explain in it in this video
hi bryan, pls can you guide me here: can i purchase an algo from you and pay for guidance on configuring it or purchase the config? whilst at the same time doing the R-course? i already know tech analysis – keen to get automated – been trading for a while know..
\Another claim of fame. This is good for mankind but there are the other nefarious. I just talked to a content editor of a large book publisher who told me most banking/hedge fund managers are not real believers in machine learning stuff for the company. They will however put together small teams to research to appease their clients who expect them to explore these topics. This is the result of a hyped industry. Also, remember these same managers are the ones who are ultimately signing the paychecks of these researchers.
Anyhow, this is the general concept until some team comes up with an entire approach to machine learning working for the automated space. As you know many of us can be wrong. I made some videos on this below.
Don’t forget this Monday Aug 27, I will be doing my webinar on this topic. I just been asked to ass Easy Language for Tradestation as well.