Technical indicators combined with machine learning to design trading strategies

AI & Machine learning
1
Hello my good folks,
Am also relatively new in the algo world, but a bit okay with python.
I have discovered a strategy where technical indicators are combined or used with machine learning as features to design trading strategies.
I found this very fascinating and decided to pursue it..but unfortunately the author could not conclude the project in a manner that is explanatory enough, also the author can no longer be traced for any comments.
But i want to narrate the strategy whether someone is familiar and can help.
 
Five technical indicators have been correlated against the closing price of a stock and various correlation figures and charts generated.
 
One of the indicators eg. MACD was found to be the most correlated to the closing price.
 
The author went further to apply the same five technical indicators in machine learning strategy where he used them as multivariant features (independent variables) and the closing price as the dependent variable. He then modeled and fit algorithms to determine which models best predict the  prices.
He is no longer interested in the best indicator but rather the best machine learning model eg. Logistic regresion, forest trees, SVM etc.
He could also not explain how this strategy can br applied in practical manner.
 
Please if you find these to make some sense to you kindly help me understand the strategy and eventually backtest the result in backtrader. am also willing to provide links to the original project whenever necesary.
 
Regards.
John            
2

Hi John,

Are you referring to some particular Quantra course where the above-mentioned strategy was discussed?



Or is it something you read elsewhere and want to discuss how to implement the same?



Either case, please share the source/course so we may assist you further. 



Thanks.

3

Hi Gaurav, thanks for the quick response. please see below the link an referring to:

4

Hi John,

The blog is an external resource and I can't vouch for its veracity.



Having said that, the post seems to give you an indication of how to use different methods to formulate a strategy (or an ML model) rather than give you a working strategy (or model) which makes economic sense.



Please also keep in mind that if you use ML for financial time series, whatever technical indicators we use, their financial significance boils down to just numbers for the ML estimator. It is up to us to choose 'good' indicators to be used as features for the model training, and I think that's what the link you shared is about.



Hope this helps!

5

@Gaurav, Thanks very much for the response.

I did further search and happy to find " Introduction to Machine Learning for Trading " in the QuantInsti lessons thats has given more insights.

Regards.

John.

 

6

Happy to know that John! #GoAlgo !