Skip to content
All library documents

Estimating Price Lag Between Exchanges with Cross-Correlation

Article Quant Q&A · Author: RLaszlo
Read the original

Summary

The document asks how to estimate the time offset between two exchanges whose prices move similarly but are recorded or updated at different times. It describes one exchange as roughly two minutes behind in the observed data and seeks a way to shift one price series to align it with the other.

The proposed method is time-lagged cross-correlation: calculate correlation between the two series at candidate time offsets and identify the offset with the strongest fit. The answer points to an external discussion and an overview of synchrony measures, but supplies no formula, implementation steps, or numerical validation. The approach therefore serves as a useful starting point rather than a complete procedure. Results can depend on sampling frequency, missing observations, price changes that are not synchronized, and the window analyzed; a high correlation does not by itself establish that one venue causes or consistently leads the other.

Key ideas

  • Time-lagged cross-correlation can estimate the offset that best aligns two price series.
  • The example describes one exchange as appearing about two minutes behind another.
  • The document points to the method but does not provide implementation details or validation.
  • Estimated lag can vary with sampling, missing data, and the analysis window.

Tags

StatisticsMarket microstructure
Full text
# How to exactly calculate lag between 2 exchanges


# How to exactly calculate lag between 2 exchanges












Let's assume that there are two exchanges. One exchange is slow for various reasons.(for eg it is an open outcry versus electronic exchange) Even when there is no lag the prices will not match exactly but almost. From the data below I can see that exchange2 is around 2 minutes late compaired to exchange1, but how can I calculate the lag in excel or python? In other words by how much should I shift the graph of exchange1 to best fit the graph of exchange2? Is there a mathematical formula for it?

## Answer by RLaszlo (score 0)

https://quant.stackexchange.com/a/69613

Time Lagged Cross Correlation seems to do the job perfectly.

https://towardsdatascience.com/four-ways-to-quantify-synchrony-between-time-series-data-b99136c4a9c9

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.