Peter
First steps are now available in a new module called roboquant-ml Still at prototyping phase to get the API stable.
Added better feature engineering to the ml module. Right now using Smile for the ML algo's like gradient boosting decision trees.
Which version are you using of roboquant? The historic data is perhaps related to issue #69, but that is only fixed in the 2.2.0-SNAPSHOT release.
I just did a quick test, and no problems retrieving historic data: ```Koltin val feed = IBKRHistoricFeed() // This assumes you have a valid market subscription for US stocks val...
Also tried the IBKR notebook and is also working. See screenshot below for the first few cells in the notebook (using Jupyter-Lab btw). 
Possible version differences? For IBKR I run 10.19, both the installed TWS API and IB GW. For the TWS API, this is also the stable version.
I remember there is some IBKR account setting that you want to share you data-subscription with the paper-trade account.
This can happen if you use high frequency data (like live quotes) and are not allowed too many API calls (I believe for Alpaca that also depends on the type...
Is there already any ETA for kotlin-jupyter-api being Kotlin 2.0 compatible? Would be great to be able to use the benefits of Kotlin 2.0.
Roboquant itself already support these TimeInForce policies but indeed doesn't map them yet to Alpaca orders. So should be easy to add them also for Alpaca.