I recently read Marcos Lopez de Prado's book "Advances in Financial Machine Learning" where I was introduced to the concept of using volume/dollar bars instead of time bars. As far as I understand, whereas time bars have you taking a measurement every set time interval (ex. once per hour), using volume/dollar bars means you take a measurement of the asset every time a certain number of shares has been exchange (or, in the case of dollar bars, a certain number of dollars' worth of shares has been exchanged).
My question is this: Are volume/dollar bars designed solely for time-series analysis of a single variable? Is there any way I can use volume/dollar bars if I'm using multiple (and exogenous) predictors?
What I mean is that if I am correlating NASDAQ price with daily crude oil price (for example), if I use volume bars, does that mean I have to take measurements of both NASDAQ and crude oil at these irregular time intervals determined by NASDAQ's trading volume? This seems like a very difficult task, even when it comes to analyzing historical data.
Again, my question concerns how to align timestamps of different assets (with different volumes exchanged) when using volume bars in multivariate analysis.
My question was also asked here, but either I completely misunderstood the answer, or it didn't actually answer the question.