Interesting Endgame in a Simul

I had a winning advantage but couldn't keep it and the position was equal by move 20. However, after move 20... Ke8, I soon regained the upper hand and the advantage in the ensuing endgame was around +6 to +7. Yet, I couldn't find a way to break into his position. On move 53, I sacrificed the bishop and went into a knight vs pawn endgame which turned out to be winning. On analysis, Stockfish recommends 53. Ba5 with an advantage of +7. However, if you input the move (Ra7+) played in the game, Stockfish increases the evaluation to +14. I found it interesting that Stockfish misses this move on a regular lichess computer analysis.

The f5 pawn break was the clearest way to convert.
37. f5 gxf5+ 38. Kf4 heading to h5.
37. f5 exf5+ 38. Kd5 with domination.

Yes, it would have been much simpler and cleaner. During the game, I was too fixated on winning the b5 pawn.

Well played. You could have played differently, but you did all right. Good game.

Thanks!

39. f5! +-