The discussion focuses on the utility of correlation as an indicator between time series, particularly in the context of currency pairs like EUR/USD and GBP/USD. Participants highlight that while correlation can be strong over short periods, it may weaken rapidly, suggesting that relying solely on average correlation may not be sufficient. The concept of Cross Correlation is introduced as a more robust measure, with its implementation available in R through the function ccf.
PREREQUISITESTraders, quantitative analysts, and data scientists interested in time series analysis and correlation metrics in financial markets.
Tosh5457 said:Is correlation between 2 time-series a useful indicator? In currency pairs for example, sometimes the correlation between 2 pairs (e.g. EUR/USD and GBP/USD) for the past x days is strong, but it can weaken very fast. Should I just take the average over time and go with it, or is there a better indicator to measure a relation between 2 time-series?