In 1914, world champion Emanuel Lasker was trailing future world champion J.R. Capablanca in a tournament, meaning that he needed to win, not just draw, against Capablanca.
He played the exchange variation of the Ruy Lopez, trading his best attacking pieces, his queen and light squared bishop, early in this classic game. For these reasons, the variation was considered "drawish." Except that Lasker won.
Here's some background that may explain why:
Lasker was the best endgame player in the world. Capablanca was inferior (at the time) in this regard. (Both of these opinions are Capablanca's.)
Capablanca was already a ferocious middle game player (and seen as such).
At the time, Lasker was 46 years old and Capablanca was 26. Lasker was more "mature."
In "Chess Fundamentals," Capablanca noted that immediately after the queen exchange, "The reader should note that if all the pieces were exchanged, White would be practically a pawn ahead and therefore have a won game." In fact, Capablanca was "rattled" by the prospect of a lost endgame and therefore misplayed what passed for a middle game.
Even though it was regarded as "unconventional," did Lasker's strategy make sense from the point of view of his endgame advantage?