How to review your own games

I decided to start a thread for people to share their tips on how to go about reviewing your own games in a productive way.

I think it’s worth separating with and without AI in responses.

Sometimes I struggle to make good use of AI because I am interested in local mistakes, but it’s often hard to steer the AI away from tenuki and 3-3 invasions!

Do you find critical turning points in the game and try playing out all the variations to see where you went right/wrong? Sometimes there are so many possibilities it’s hard to see what you could reasonably have read out in the game

Right now I try and just come away with one, maybe two key mistakes to remember.

What do you do? How about getting the most out of AI?

2 Likes

Its useful to understand how local moves sequence may look like, but its not necessarily good idea to actually do it in a game, tenuki may be better.

1 Like
  • Joseki mistakes: that’s the easiest, take note of the move shown by AI and try it next time.
  • AI is like a teacher, you learn more if you ask questions. Like “I didn’t know if move A or B was better, I was afraid of C but wasn’t sure, can you confirm?”
  • If the AI suggests a move that you would have never thought about, it may be a sign that you have to change your way of thinking. Try to understand the idea of the correct move, and why you missed it. Are you too defensive? Too aggressive? Did you fail to see a weakness in your group, or in the opponent’s group? Did you underestimate the strength of a group? Did you overestimate the importance of a stone? The AI showed that it was not necessary to connect two of your groups, can you see why? Or it told you that you should have connected two of your groups, can you see why? Etc.
  • Some AI moves are difficult to understand, don’t try to understand all of them. In fact, understanding 2-3 ideas from a game review is already good.
  • Looking at AI moves doesn’t replace reading practice. Knowing that a move is good isn’t enough if you can’t read followup variations.
4 Likes

I try to notice 2 things with AI:

  1. Large drops in winning percentages (mine or my opponent’s).
  2. Trying to save dead groups. Tsumego is my worst skill, so when I notice in AI review that I thought that something was going to work when it wasn’t, I try to take note.

Without AI, I try to notice when my opponents, especially stronger opponents, tenuki. If I thought I was threatening something and I clearly wasn’t, it’s something to take note.

Also, malc/ personal notes in chat help to remember to review specific moves.

1 Like

if you are few points behind it may display you as having 0%. Then if you lose 100 points, % graph will show nothing because it can’t go lower.
I completely stopped to look at %, its meaningless in most situations.
Score graph is better.

2 Likes

I did mean score tbh, I’ve switched to score since you and a few others had pointed to me how percentages don’t help. :slight_smile:

2 Likes

A stronger player review is very useful and he doesn’t have to be very stronger. A not so far from your level player reach to point easier to correct mistakes because you are not that far from his level, and still miss important things to correct.

Otherwise a very strong human player may be better at distinguishing priorities in the things you have to improve.

Myself I like short review pointing key moments in which the game lost tracks. Even better if I didn’t notice it and even better if I thought it was good when it was really bad idea.

1 Like

Just play out the tenuki or the invasion on the board, and see if the AI comes back to the local position a few moves later and shows you the next move.

Also, force the AI to play the local move immediately, and see if it thinks a local reply is best, or if the tenuki is actually much bigger for both sides.

2 Likes