This is due to way the number is calculated. The definition of noClickRate is the following:
The no-click-rate is the percentage of tracked searches yielding results that have a query that never generates any clicks.
A "search" in this definition indicates a "query term" used in search requests, like "orange", "apple" and "banana" in this documentation.
Let's say you have analytics of searches and clicks for 2 days.
Day 1: 1 search on "orange", no click
Day 2: 1 search on "orange ", 1 click.
If you look at day 1, you'll have 1 no click count.
If you look at day 2, you'll have 0 no click count.
If you look at the range [day1, day2] you'll have 0 no click count.
The no click count metric is aggregated per search term (query), so the larger the time window you are looking at, the more likely a given query will end up having clicks.
Because of this nature, the bigger the time window, the smaller the noClickRate value.