On a 3 shot group, the standard deviation will always be half of the extreme spread.
And I disagree with the other posters. If used in a load development ladder of .5 grain increments, a three shot group method will tell you what to re-test by showing you the nodes. If the SD stays good again on a different day, you have a winner for that temperature range 95% of the time.
In other words, a 5 shot group on one day is less meaningful than two 3 shot groups on two different days.
If it's not far, it's boring.