Simplistic analysis aside, I'd consider winning your campaigns to be a hard requirement for "best" general and this excludes Napoleon, Lee, Rommel, etc.
The qualities of a general will be demonstrated both in wins and losses, and whether he's victorious in the end tells more about the military power of a country or king, not about the quality of the general.
Right. Indeed, efforts have been made to quantify this that take into account the contribution of other factors (e.g., T.N. Dupuy’s Quantified Judgement Method of Analysis which check later as developed into the Tactical Numerical Deterministic Model.) And Napoleon seems to have come out pretty outstanding, as I recall, in analysis using the QJMA.