There appears to be a small bug relating to end-of-year processing or categorisation in the PVOutput website logic.
Checking through the many weekly/monthly/yearly statistics pages, I came across WillisAve 7.370kW which shows a ‘Dec19’ line later than the ‘Jan19’ line, with a record date of 31/12/18, and a ‘Total Generation’ equal to that of the single day 31/12/18.
Dec19 is still 11.8 months away, so somehow the 31 December 2018 day stats has been identified as a record for a TARDIS-accessable future December 2019 instead.