calculating correlation with data offset

I was calculating correlation on the NKY and SPX for a particular period and the results was 0.26, I was matching the datapoints with the same date, for example SPX close on 05 Jan with NKY close on 05 Jan. Then I thought when not match SPX close on 05 Jan with NKY close on 06 Jan, as anyway the two markets are not open at the same time so anyway you are always matching two date points with a gap. The result was a correlation of 0.4…

Could the 0.4 be a more accurate measure than the 0.26 ? How do you interpret the two measure? Any article to help with this.


Your correlation in either case is highly spurious, as both have about a 12 hour offset. There’s no different interpretation; both are going to have a high degree of random error. You should either choose assets that close at the same time (for instance NKY based ETF that trades in US hours, or even the NKY futures on CME), or change your data to use larger periods (for instance, weekly return correlation, instead of daily).

Daily returns have a low signal-to-noise ratio (high noise-to-signal).

Weekly returns or monthly returns have a much higher signal-to-noise ratio.

thanks guys for your responses, i will read up more based on your comments