Why High-Frequency Cleaning May Use a Single Exchange

Summary

The note explains why a realized-kernel data-cleaning procedure may retain trades and quotes from one exchange, such as the NYSE, rather than combining records from multiple venues. The stated aim is to limit distortions caused by reporting delays: timestamps across exchanges may reflect different observation and transmission paths, so a price update at one venue can appear before another venue records it. Such timing differences can complicate high-frequency comparisons and realized-volatility estimation.

Using one exchange makes timing more consistent for the analysis, but it leaves out activity elsewhere. The note emphasizes this tradeoff: single-venue data are easier to handle, while combining exchanges can offer a fuller market picture and greater coverage of trading volume. It offers a rationale rather than a quantitative test of the effects, and does not specify synchronization methods or how to reconcile timestamps across venues. The appropriate choice therefore depends on whether cleaner timing or broader market representation matters more for the research question.

Key ideas

• Single-exchange filtering can reduce timing inconsistencies in high-frequency trade and quote data.
• Different reporting paths can make timestamps across venues difficult to compare directly.
• Using one venue simplifies analysis but excludes trading activity on other exchanges.
• Aggregated exchange data provide broader coverage while requiring careful treatment of timing delays.

Tags

# Cleaning of high-frequency data


# Cleaning of high-frequency data












In the paper "Realized kernels in practice: trades and quotes" by O. E.Bandorff-Nielsen etc. cf.

https://onlinelibrary.wiley.com/doi/full/10.1111/j.1368-423X.2008.00275.x

in the section dedicated to data cleaning the authors suggest:

> Retain entries originating from a single exchange (NYSE in our application). Delete other entries.

It is related to both Trades and Quotes data.

Why shouldn't we consider the data from other exchanges?

## Answer by chrisaycock (score 3, accepted)

https://quant.stackexchange.com/a/54552

The authors explain the reason later:

> It is used to reduce the impact of time‐delays in the reporting of trades and quote updates.

Because of the speed of light, the timestamps of data from external exchanges will have a different meaning from the timestamps of NYSE's own data. For example, a price update on NASDAQ will occur a few milliseconds before NYSE observes it.

The authors go on to consider modeling on one exchange vs all exchanges. Since NYSE's volume alone is less than all other exchanges' volume combined, it makes sense to consider the aggregate. Researchers will have to weigh the ease of one exchange vs the fuller picture that comes from all exchanges.

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.