Close-in companion stars are expected to adversely influence the formation and orbital stability of circumstellar (S-type) planets by tidally truncating protoplanetary discs1–4, impeding mutual accretion of planetesimals5–8 and narrowing dynamically stable regions9. This explains the observed dearth of S-type planets in tight binary star systems10–13. ν Octantis, whose stellar components have a mean separation of 2.6 au, has long been suspected of hosting a circum-primary planet in a retrograde and exceptionally wide orbit that resides midway between the stars14–20. Strong theoretical grounds against its formation and the absence of observational precedents, however, have challenged the reality of the planet. Here we present new radial velocity measurements that consolidate the planet hypothesis. Stable fits to all radial velocity data require the planetary orbit to be retrograde and practically coplanar. We also report the critical discovery from adaptive optics imaging that the companion star is a white dwarf. Our exploration of credible primordial binary orbital settings shows that the minimum separation between the stars was 1.3 au initially, which overlaps the current planetary orbit and makes any scenarios in which the circum-primary planetary orbit formed coevally with the young stars hardly conceivable. The retrograde planet must have originated from a circumbinary orbit or a second-generation protoplanetary disc, showing the role of binary stellar evolution in the formation and evolution of planetary systems.
