Markus K.

Same behaviour: ``` root@1536b88504d4:/# pip install --upgrade git+https://github.com/microprediction/timemachines.git Collecting git+https://github.com/microprediction/timemachines.git Cloning https://github.com/microprediction/timemachines.git to /tmp/pip-req-build-35yoc7nx Running command git clone --filter=blob:none -q https://github.com/microprediction/timemachines.git /tmp/pip-req-build-35yoc7nx Resolved https://github.com/microprediction/timemachines.git to commit bab4ce3faa9e767ca7ef277847154c99aafaef69 Preparing metadata (setup.py)...

Incredible work. Will you port this to ANKI as plugin again, too?

I think the issue/solution lies in `return price_array, tech_array, turbulence_array` in the data_processor / run function. If you would pass the whole dataframe all necessary data would be available. Not...

After looking at other examples and (gym)environments I think best practice would be doing it inside the env: [https://github.com/AminHP/gym-anytrading/blob/master/gym_anytrading/envs/forex_env.py#L19](https://github.com/AminHP/gym-anytrading/blob/master/gym_anytrading/envs/forex_env.py#L19) Init the env with the df and then processing the data...

Example: ```python class DataProcessor: def __init__(self, data_source: str, **kwargs): self.df = pd.DataFrame() ... # all the functions just add / modify the dataframe def add_technical_indicator(self, tech_indicator_list: List[str]) -> None: """...

> Your suggestion is good. We consider to use the df['key'] to pass the data. We also consider using dict, d['key']: np.array to pass data, which is more efficient. In...

Awesome. BTW regarding the decision between the dataframe vs numpy. Using a dataframe as long as possible ensures compatiblity to other libraries. An example would be [tsfresh ](https://tsfresh.readthedocs.io/en/latest/text/list_of_features.html)which is an...

Sadly, my time is pretty limited currently due to university and other project(s). If I'm not mistaken, you can just use the Volatility Features code from the kaggle notebook I...

Might be an interesting feature to test: https://www.kaggle.com/wptouxx/g-research-cointegration-analysis Cointegration tests for all asset pairs and obtaining their p-values.

Awesome @aron-alarik - this would also fix #45 . I think it's better like you said to make it more modular by accepting the features via a dataframe. BTW what...