How to Use TA-Lib With Machine Learning For Trading Strategies

- How to run TA-Lib (Technical Analysis Library) in a Google Colab environment
- Create and plot technical indicators with candlesticks
- Run a specific set of technical indicators en-masse.
- Extract 30+ candlestick patterns and train a machine learning system
- Backtest and improve the candlestick pattern strategy

Shape, co-create and improve our content based on your individual needs fill out this short survey (it only takes 3-5 minutes)

Check out our playlists:
Algo Trading from Scratch

High Performance Backtesting and Trading with NautilusTrader

Feature Engineering For Financial Applications

[HansMcMurdy]

Did i miss where the links is to this codebase?
Also any advice on running this locally on windows because TA-lib hates windows...

[javadghasemi]

hi, tnx for the great video, how can I download all codes?

[timmydotlife]

It is by coincidence that I land on this channel. I know you from the chat with traders podcast. My head was kind of hurting afterwards :D, but definitely sparked an interest! Thank you for sharing!

[marcusl]

Hey, what an interesting video! Thank you for this amazing video!

[ihtishamali]

Hi, Thank for for the wonderful video. However, I did not understand why the two strategies were combined near the end. It seem that part of the combined strategy comes from benchmark returns and part from the retrun prediction of the model. Isn't the former something which we cannot know in advance and thus useless in predictive manner or did I misunderstand some point. Or is the intent just to show that if somehow we are able to counter the wrong predictions (borderline ensemble learning) we can improve the model performance?

[hbbexxter]

I pretty much followed the logic here, but unfortunately, the logistic regressor does not converge (43:14). I experimented with number of iterations, increasing the tolerance, using a different solver and taking fewer inputs, however, it does not work. I cannot see a particular preprocessing pipeline, but I also used the method.

Do you select a particular set of candle stick patterns, test for multicollinearity or consider any other normalization for the logistic regressor to converge?

Any help is highly appreciated.

[ihebbibani]

Hi Dr Starke, I have a problem with using machine learning models as we can use them only on train and test datasets and not on out of sample way . What to do then to use artificial intelligence for trading ?

[YoungMoneyInvestments]

Im seeing this video a year later. Is there anything you’ve found extremely helpful since publishing this video? Is there any insights you have now that would have been revelatory when you filmed this? Thanks for the vid.

[thesocialmaestro__]

How to mix ai with machine learning and give continuous data to the model.

[rajvaibhavyadav]

Sir, is it recommended to use TA-lib's ema function or shall I use some other library? Great video btw.

[anubhavbairoliya]

I want to use Chande Kroll Stop indicator. Can anyone please help me. It is urgent. Thanks

[KLENTONY]

Eu gostaria de entrar em com contato com vocês através de outra plataforma. Voces tem um canal no Telegram ou qualquer outro meio de contato para que eu possa esclarecer alguma dúvidas? tenho interesse em comprar o curso de programação Python.