Information-Driven Bars for Sampling Market Order Flow

Summary

The document concerns imbalance bars and run bars, sampling methods discussed in Advances in Financial Machine Learning. The author is implementing them and asks for further literature, explaining that the material consulted so far emphasizes volume sampling, VPIN, tick rules, and market microstructure models but offers little implementation detail on these bar types.

The post provides context for why information-driven sampling is of interest: fixed time bars may not reflect how much trading activity or directional order flow has occurred. However, it does not explain the algorithms, give implementation rules, present empirical tests, or answer the literature question. It is therefore useful mainly as an introduction to the topic and a pointer to the gap between general discussion of order-flow features and practical guidance for constructing imbalance and run bars.

Key ideas

• Imbalance bars and run bars are information-driven alternatives to fixed-time sampling.
• The post places these methods alongside volume sampling, VPIN, tick rules, and market microstructure models.
• It identifies implementation details and supporting literature as unresolved questions.
• The document itself supplies no algorithm, performance evidence, or definitive reference list.

Tags

# Information Driven Bars (Advances in Financial Machine Learning)


# Information Driven Bars (Advances in Financial Machine Learning)












My team and I are busy coding up a python implementation of the information driven bars (imbalance and run bars) mentioned in Chapter 2 of the text book Advances in Financial Machine Learning.

There really isn't a lot of information published on this technique. I have read the papers in the bib, mainly:



- Flow toxicity and liquidity in a high frequency world.

The two papers really highlight the importance of volume sampling and how market micro-structure features like VPIN can be used as an important features but neither provide deeper insight into the imbalance or run bars.

I then turned to chapter 19 which has a very nice explanation on the tick rule and the various micro-structure models and their generations.

However I still don't have a firm grasp on the implementation or details for the information driven bars.

The following blog post has helped us with the implementation Maks Ivanov.

Main question: Is there a piece of literature that I have missed? Where can we learn more about this technique. Is it mentioned somewhere in another journal or a slide show?

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.