|
PureBytes Links
Trading Reference Links
|
Statistically speaking, there are conversion factors for unequal sample
sizes, probably not in Excel calculations.
Its best to keep equal time elements as well as in order of (time) series,
another factor that increases confidence levels.
Its easy to make statistics lie by placing outliers near end of time series,
so don't look for the best/worst possible results:
do a sensitivity analysis with data if any question.
Phil
> -----Original Message-----
> From: Ross Bond [mailto:Ross.Bond@xxxxxxxxxx]
> Sent: Saturday, September 18, 2004 7:59 PM
> To: Omega List
> Subject: Resend: Correlation Question
>
>
> Oops, I think this might not have got sent the first time:
>
> A statistical question if I may:
>
> When using Excel to perform a correlation test (using the CORREL
> function) with the two series of trade results in columns consisting
> of an unequal number of values, should I attempt to manually align the
> series by date order or not?
>
> The result varies substantially depending on whether or not I manually
> align them. It makes sense to me to align the series by date, but is
> that the correct method mathematically speaking?
>
> Thanks in advance!
>
> --
> Best regards,
> Ross mail to: Ross.Bond@xxxxxxxxxx
>
|