It's an interesting idea but it's not really a fair test since your computer opponent blundered with 17.. Rcd8 just when things were getting interesting. Stockfish says that Clarabit had cut your evaluation lead from 1.37 after Kxf2 to 0.91 before it blundered.
It looks like Clarabit's rating wouldn't be 1411 here.
I think you should try that position against a better computer, maybe not Stockfish but something that would at least be your equal.
I've had a lot of fun using Lucas Chess's wide array of playing engines, levels, styles and forced initial first few opening moves to study openings I'm playing or considering and having to deal with a lot of different variations.
This is just a game vs an ELO 1411 engine, but it explores what to do when someone pulls off that old, common exchange at f2 or f7 of a castled position of B+N for R+P - often mistakenly considered "equal."
The game that starts out quiet - in fact it can be called, by transposition. Giuoco Pianissimo (Very Quiet Game), but Black livens things up by choosing to exchange his Bishop and Knight for my Rook and Pawn at f2.
The old Fred Reinfeld Numbers tell us it's an even trade: Q = 9 pawns, R = 5, B = 3, N = 3, P = 1.
But Larry Kaufman's award winning study of 80,000,000 positions in a 1999 Chess Life article gives us: Q = 9.75, R = 5, B = 3.25, N = 3.25, P = 1.
So the Black trading B + N for my R + P, everything else being equal, is a 1/2 pawn equivalent in White's favor. With two N's on the board and an extra piece, my plan was too keep things bottled up until Black's hands got too full:
Chess.com Max. Anal. Indexes:
Lucas Chess/Stockfish 8 (quick 12 ply) more realistic Indexes: