Practical Uses of Autoregressive Conditional Duration Models

Summary

The document introduces autoregressive conditional duration models in the context of high-frequency financial data. The questioner has implemented an ACD model in R using trade-only observations and asks how modeling elapsed time between trades, or volume durations, might be useful in practice. The reply connects the topic to applications in market making and liquidity analysis, where irregular trading activity and time between trades may be relevant to understanding liquidity.

The response recommends a book on modeling irregularly spaced financial data as a source for real-world ACD applications. It does not describe a specific estimation method, empirical result, or trading strategy, and the suggested liquidity use is only a motivation rather than a demonstrated application. The document therefore gives a research direction for studying duration data, while leaving the practical modeling choices and evidence to further investigation.

Key ideas

• ACD models describe the time intervals between financial events such as trades.
• Trade durations and volume durations can be studied using high-frequency transaction data.
• The reply identifies liquidity analysis at a market maker as one possible practical application.
• A reference on irregularly spaced financial data is suggested for examples of real-world uses.
• The discussion proposes a research direction but reports no tested model or empirical findings.

Tags

# Application of ACD models


# Application of ACD models












I have been playing around with autoregressive conditional duration (ACD) models and I have a nicely working R based implementation using real high frequency data (trades only data).

However, what's the point with modelling duration between trades? Or even volume durations?

Any idea how this stuff could be used in practice? I would like to make a Master's thesis out of this.

## Answer by user2303 (score 3)

https://quant.stackexchange.com/a/3279

If you haven't already, I would check out Nikolaus Hautsch's book: Modelling Irregularly Spaced Financial Data [1]. I have just started reading it, but it looks like he devotes a good portion of the book to going through real world applications of ACD models.

I work at a market maker and spend a bit of my time analyzing our liquidity, so I thought it could be worth looking at ACD models. Would you mind sharing your R code, btw?

[1] http://www.amazon.com/Modelling-Irregularly-Spaced-Financial-Data/dp/3540211349

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.