No you can't. Splunk never reads a file backwards. If it starts at the end, it simply skips the beginning and reads forward.
If you're trying to do this, I would suggest you simply create a new batch/sinkhole input in inputs.conf, pointing at a different location, or use the the
oneshot command to simply index each file on demand. Note also that you can't add arbitrary index-time fields to your data using the
***SPLUNK*** method anyway (only specific Splunk keys like
index), and would need to use additional
TRANSFORMS to make it work. You might have similar problems using any other method, however.