My friendly game @ Asia Square Tower 1 on 5 Oct 2016

Comments in the games are what I feel. Comments in the main text after the game is what the computer assesses.

Thanks for taking time to see this, and feel free to comment / criticise the moves for improvement/learning purpose.

Comments based on checking with computer analysis:

• 7.Qb3 can be countered with 7...dxc4 8.Bxc4 Nc6, threatening ...Nxd4 or ...Na4.
• The idea of 8...a6 and 9...b5 is playable, but may not be necessary. Another possible way to play is just 8...Be6 (if 9.Ng5 Bf5).
• Computer's evaluation keep going up and down over the move 11...Qb6!? It is reasonable to say white has a small edge (between +0.1 to +0.2) but computer doesn't have an immediate KO refutation as well.
• 15...Rfe8 is indeed an inaccuracy which 16.Qb3! forces 16...Rfe8 (evaluated at +0.5 to 0.6 or so). 14...c5 seeking to open lines and simplify the position, ...h6 simply kicking back the Knight, or even g6 to address the strong Bd3 hold over h7 square are computer's suggested moves.
• 15.Nce4 Qd5!? Even the computer assessment is uncertain if 15...Qd5 is a good move. Evaluation has been swinging between +0.1 to +0.4 The most critical being 16.Bc2 (threatening Bb3 pin) Nb6 and the position becomes wild.
• 17...Nxf6 is more forceful than 17...Bxf6 18.Bf5 (or insert 18.Bh7+ then 19Bf5).
• 19...Nxe3 missed the much more straight-forward 19...Nb4! 20.Bb1 Nxa2 21.Bxa2 Bxe4
•  20...Bxe4 21.Bxe4 Bg5 22.Bf5 is fine for White. Even position.
• After 25.Rc5?! Black can try to exploit the 2 Rooks vs 1 Rook on the e file and attack the d pawn. However, the winning moves and idea 25...Bh4+ 26.g3 Bf6 is still not fully understandable to me
• After 33.d6? Black missed the absolutely winning move 33...gxf3! (If 34.dxe7 fxe2 35.exd8=Q e1=Q+) with an excellent position due to White's overly exposed White King. In fact, computer sees a forced mate in 9 moves incl.35...e1=Q+. Black just stopped calculating when he thinks the loss of Rooks + promotion for White made no sense.