Replaying pro game records as learning method

Hi there. I would like to know what you think about repeated replaying of (uncommented) pro game records as a learning method when used in conjunction with puzzle solving and playing plus review of live games.


It can be a pleasant thing to do, like reading a novel. You may learn new patterns, and open your mind to other ways of playing. I don’t know how useful it is compared to other learning methods. It may depend on your way of practicing: how much you learn mostly depends on how much mental effort you put in the activity. Anyway it can’t hurt, so if you like to replay pro games, go ahead.


Dwyrin described that one thing he did was memorise pro games.


I can’t really answer as I’ve (still) not managed to do this but I can see the appeal and would really like to

I think it’s alright to replay games you like. Thought there’s a caveat to replaying modern professional games. While in the opening players will play perfectly due to memorization of ai variations, which they research extensively for a daily basis, in latter stages of the game they would start to make mistakes due to time pressure. While you could assume in historical games that every move is perfect(with exception to the opening) and rely on players making correct choices to learn from them.

Basically before ai professionals used to review historical games to improve. With ai now you can be reassured that each move is perfect, so value of historical games dropped in a sense that you have one more source of flawless go performance in face of go playing ai.

Sidenote: I’ve heard that watching professional games might be less useful than watching games played by players of similar rank(just few stones higher). The reason is you could understand more and learn more as a consequence this way, because top professionals’ thinking is too “abstract” compared to amateur level for example.
If you ddk> watch mid sdk play;
if you’re lower sdk player> watch how 4-1kyus play;
If you’re high sdk> watch low dans play;
low dan>high dans games;
high dan> low pros…

1 Like

You don’t need to watch perfect moves. All pro’s moves are good enough and interesting to learn from, unless you are high dan.

1 Like

I’m at a loss. Did you reply to my comment or is it’s just a coincidence that you chose to address same question I discussed, but wasn’t included in the main post?

You may add 2 steps

  • guess the move
  • guess why the pro play here instead

Actually I didn’t understand what you had in mind, since you said two contradictory things:

  • “caveat” about pros making mistakes
  • saying that watching low dan games (which are full of mistakes) is good enough for kyu players.

To reiterate my opinion, pro games are “safe” to watch in the sense that mistakes are almost all the time very small compared to our ordinary play. Even if their moves are not perfect, the fact that the pro thought about a move that didn’t cross your mind is already interesting and can be learned from.


I think replaying uncommented pro game records is not a bad learning method per se, but it is much more effective when you have additional commentary. Learning about the thought process behind the moves is very helpful, so you can apply similar ideas in your own games.

1 Like

I think it’s fun and useful to replay pro games, even somewhat mindlessly. I don’t do it so much these days, but earlier in my journey it definitely helped me develop my instinct for where a “normal” play would be. One caveat is that it took me a long time to lose my first 100 games, and I may very well have developed faster by playing myself than by replaying pro games. But if you like it, do it. It’s definitely beneficial.

Doing it with more focus and mental effort is good, too.


I just wanted to state that pros are imperfect, so if you think pro made a mistake it might be that he actually made mistake. In most cases when you would doubt pro, he or she would be correct and you wrong, but still you cannot exclude that possibility. I didn’t say that it’s good or bad, but I think it’s important to be conscious of that.

About watching games of similarly strong players that’s what I’ve heard. Don’t know how true or false is that. I would try to review games I would understand, if I have trouble seeing players thoughts I don’t think that it’s as useful as it can possibly be if I understand his thoughts process. I myself would take play style into account also. If you’re a moyo player you probably would understand Takemiya better, while for territory-hugging player his moves would make to sense. I would think that style preference is more important than strength(it’s possible to have player of same rank but who’s move choices you don’t understand at all because of how differently he approaches go).


I want to echo the “useful” part of this (not that I’ve done much/enough myself). In the last couple of years, I have done some reading about how learning works (not specific to Go), and it’s much more about training the unconscious than I had assumed before (much less about consciously thinking hard). One natural application to Go is that you can learn a lot by mindlessly walking in the steps of pros (or, presumably, of KataGo).

1 Like

But but but… What impress me is the creative reading beyond the mindlessly walking. (I agree that’s mostly concerning highly advanced players.).

1 Like

I’d compare watching pro games mindlessly with listening mindlessly to conversations in a foreign language. You get familiar with sounds, and even with some expressions, so this is definitely useful, but much less than trying actively to understand the conversation.


Ah, you said watching. I thought we were talking about replaying physically. The latter builds actual habits. More akin to repeating back a conversation in a foreign language word by word

(Edit: for learning, it’s important to be actively involved… not necessarily important to be conscious…)


I wasn’t sure what “replaying” means:

  • open an sgf file and click on the arrow to watch move after move
  • same as above, but replay it on a real board at the same time
  • use a game diagram (with 250 numbered stones on it) and replay the game on a board.

John Fairbairn says that T Mark Hall improved from 2d to 4d just from transcribing 800 game records (diagrams) of Go Seigen. The reason is that to do it efficiently, you really need to guess where the next move is, or at least in which part of the board it is, so you need to understand what’s going on, and be aware of the whole board and not just the local fight. I suppose you can also do that while clicking through an sgf file, the main point is that you have to understand the conversation, not just repeat it like a parrot (even if parrots are smart animals and may actually understand a little of what is being said).

I’m a bit confused. Apart from repeating a conversation, do you have other examples of learning methods in which you are actively involved and mindless at the same time?