Minute "bar" support

[edtechre]

Minute "bar" support

[edtechre]

Thank you for your feedback, @spinus!

Regarding your first point, bars in PyBroker do not necessarily represent days. Instead, they represent the granularity of the data used for backtesting. For example, if you are using per-minute data, then bars will be per-minute as well.

As for your second point, I'm not quite sure how that would work. You can increase the granularity of your data to whatever you like, but the framework assumes that bars are discrete time steps that have already completed.

[spinus]

@edtechre thank you for very fast response.

1. That makes sense. Just documentation was misleading in few places as it refers to bars as days and I didn't find any general statement about that. Maybe worth clarifying.
2. Right, but often you cannot get second or minute level data, so your backtesting can be misleading when later algorithm is switched to live trading and won't be tested on sensitive "bar building" events which can trigger events that on daily level data that wouldn't be trigger. One idea for that, would be adding generation of sub-level bars.

[edtechre]

Thank you @spinus.

1. I will look over the docs again and see if I can make this more clear.
2. You should conduct backtesting using the same timeframe as your intended live trading. For example, if you plan to trade intraday, then you should use historical intraday bars for backtesting. Generating sub-level bars would not be possible if there is no available sub-level information. For instance, if you only have daily data, it is not feasible to increase the data granularity to intraday.