Against Stockfish: Game #01: Start of a New Journey

MCTS Training: Correspondence Game #01

Using the Monte Carlo Tree Search to hold or beat the unbeatable computer:

Game #01 Analysis:

Lessons from this game:

> Broaden your list of candidate moves (...b5?!)

> Every reasonable move demands at least a little bit of calculation (Qc2)

> Calculate the move until you reach a static position (...Nxa1?)

Monte Carlo Tree Search
In this series of games, I intend to use a similar method/algorithm to what AlphaZero used to defeat Stockfish 9 in 2017. The structure of the Algorithm is something like this:

> First, we take a list of candidate moves and see which moves are most preferred by our intuition.

> Then we make a play-out (a.k.a roll-out) by applying the same principle from our opponent's side and mark the evaluation when a static position is reached.

> Then we take the average of the so-found leaf positions to assign a value to the calculated candidate move.

> Then we repeat the process for all the candidate moves and compare the move values to come to a final conclusion.

This is how I am going to accomplish my goal. So wish me luck and share your thoughts in the forum. I am always open to feedback and criticism.

External links
YouTube: https://www.youtube.com/@felew699 (No videos yet. I am still working on it)

Against Stockfish: https://lichess.org/study/YTek9jaV

Grammar Editor: https://app.grammarly.com/