Review request (somewhere SDK)

I had this correspondence game.

I made a review
to write my toughts and to check them with KataGo (I use LizGoban as well as OGS AI).
The overall picture is that both players made many slow and weak moves, continuously making gifts to the opponent (so kind of us! :D).

I found many mistakes in my play:

  • I chose a wrong approach in BL corner.
  • I left incomplete that joseki (I don’t know josekis).
  • I missed an opportunity of threatening to kill the 3-3 invasion in TR corner with a ko.
  • I tried to build in the middle but wasn’t consistent and I built very little.
  • I made a lot of slow and small moves.

But my bigger doubt is about the whole picture: until move 88 the winrate graph is a rollercoaster, but (thanks to my opponent) always comes back to about 50% and the score chart is completely flat, so I’d say that the game was quite even.

Could we say that I lost this game because of endgame?

Obviously, if I hadn’t made so many mistakes before, my winrate would have been better and I would have been ahead and later mistakes wouldn’t have been so important for the outcome.
So I’m asking myself (and asking you): should I improve my fuseki and middle game before worrying about endgame or the opposite, should I improve my endgame to have it’s strenght comparable with the others?

(BTW… this is weird:
this wasn’t definitely a game between 2k and 1k)
((I am also surprised that the new ranks were applied to previous game chat… I thought the chat were written at the moment and never updated anymore))


Chat ranks are (or were, at least in the early stages of the rank update) more optimistic than game ranks.

If the rankbump is the function R, then chat ranks are R(player) but game ranks are R(R(opponents)). The former generates a more flattering rank.

eg. my chat rank is strong 3k, but my game rank is only (strong) 4k.

I don’t think the answer to that question can be derived from this AI review alone.

For the most part of my own go career, I had the feeling that my endgame was relatively weak, compared to other players of my level. But my fighting was relatively strong.

Having this perception about my own strengths and weaknesses, I generally tried to fight my way into a lead before the endgame, so that I could afford to lose points in the endgame. Or preferably, win the game by resignation in the middle game to avoid the endgame altogether. So I tried to use my strengths to compensate for my weaknesses. To a certain extent, this worked to keep a mostly stable rank, but I wasn’t able to improve from this point. More fighting didn’t really work to get better results.

Then I started taking lessons from InSeong Hwang 8d EGF. His advice to get improving again, was to work on improving my weaknesses, not on improving my strengths. So I should play more calmly in the early game. Focus on taking territory, rather than trying to bully my opponent into submission in the opening and middle game. Improve my self control, my reading and my endgame. Become a more patient player. Learn to carry a close game through the endgame to a narrow win. Only using my fighting skills when I’m losing.

So how do you feel about your play in general? What are your strong points and weak points?


On (19), you should consider that R13 or R14 doesn’t just enclose the corner; it also cramps White a little and looks at a later exposure of aji around S8. Also note that a lot – perhaps most – of the time, it’s better to approach the 3-3 from the side rather than cap it.

White was negligent in not taking D6 on (24) and you were right to immediately exchange in that area. Whether D3 was a reasonable move is a bit above my level.

As you realised, (27) should push White into your wall, not away from it.

I agree with KataGo that (31) D6 was very important, definitely verging on mandatory.

I thought the (32) corner invasion was pretty well understood at new-ranks 4k level, but there you go.

(36) R18 is, afaik, not the main sequence for White to reach a ko. That is R16 R15 S18 Q19 R19. I think the game move dies to S16.


Are you talking about in-game chat?
Since I always play correspondence, I used to rely on in-game chat to see rank changes during the game: the “glhf-rank” versus the “gg-rank”. :grin:
And it worked!

Then I lost interest in rank changes and disabled ranks display, so I don’t know what happened more recently.

Thanks for your story.
I feel incredibly weak and disoriented, but in the end results aren’t that bad! :grin:
I feel myself as my EGD rank says: not SDK yet!
And reaching SDK is my only goal, so I should be quite happy with my new 3.1k :grinning_face_with_smiling_eyes:

More seriously, I think I have a grasp on game mechanics but still struggle a lot to recognize good direction of play.I’d like to make more efficient moves, better recognizing weaknesses, maybe even exploiting aji a little.

One weak point I feel is reading: actually I don’t read at all. I just play instinctively.
This brings not anticipating opponent’s moves. Many times my opponents make moves that I didn’t expect. Some are blunders, but some aren’t! :grin:
Some are actually good moves that just I didn’t see.
That’s also the reason why I struggle doing tsumego: I just can’t see all possible moves.
Probably I should work on that.

Really? S8? This is science fiction for me! :grinning_face_with_smiling_eyes:

If I recall well, katago was agreeing on that, but I’ll check again.

Really? :flushed:
I must check that too!


Really? S8? This is science fiction for me!

Waltheri is the best place to research these variations, since the surrounding stones make them quite context-dependent.

I don’t really know that site. I went there long time ago and couldn’t find a way to take some good from it. Pro games are so far from my level!

How could I search this specific pattern?
I just tried to reproduce my game but no pro player played all of our first 18 moves. :grin:
I was quite surprised that someone in the '60s played 11 of them. :grinning_face_with_smiling_eyes:

1 Like

The are two eays to search for local positions.

  1. Select the Around stones option, which creates a “bubble” of three points’ empty space around the searched position.

  2. Click and drag to draw a rectangular selection around those stones you want to search generically, and then select Search W or Search B.

Around stones is more restrictive than rectangular selection, so it’s usually best to use that method first. If neither produce any results, check OGS Joseki; if there’s nothing on there, you can always make a question thread on the forum.

1 Like

Based on this game alone, I would characterise your style as overly cautious and passive.

But especially with black, I think you should try to play more ambitious and active moves in the opening, trying to build a framework, instead of letting white split the board in small areas and rely on komi to give him an edge. The position after move 18 may be even from an AI perspective, but from a human perspective, white is doing well.

Take move 5 for example. Did you consider playing the Chinese opening (R9)? Pros mostly stopped playing it after AI, but it is actually not bad according to AI. Especially with white playing 3-3 on move 4, I think the Chinese opening is perfectly fine. Or if you want to play slightly more modern you could play Q6 or Q5 on move 5.

Some other moves that feel passive to me:
Move 19 (D6 would be more active)
Move 27 (C2 would be more active)
Move 37 (S16 would be more active)
Move 43 (if we ignore that fact that D6 is a must, Q10 would be more active locally)
Move 45 (R11 or O10 would be more active)
Move 75 (connecting at D6 creates cutting points in white’s shape)

Also in the endgame, you tend to play somewhat passive moves, following your opponent. You keep playing gote moves that make too few points, such as moves 81, 85, 87, 89. On move 80, the game was still even, but by move 104, your opponent is leading.

Move 107 damages your lower right corner, causing black to fall further behind. You don’t really seem to be aware of the danger (your opponent could even have created a ko to kill your whole group on move 124). Perhaps this points out an issue that you mentioned yourself. Your intuition about danger and safety is not very accurate. You tend to play unneccessary defensive moves and you tend to miss neccessary defensive moves.

How to improve from here? I think your weakness is fighting and tactics. A natural way of avoiding that weakness is avoiding fights, as you seem to do. But when you keep doing that, you won’t improve. So I would advise you to play more actively and aggressively and fight. Making frameworks in the opening (such as the Chinese opening) is a natural way to make games with lots of fighting.
When you adopt this new style, you will probably fail miserably initially and your rating will go down. But in the long run, I think this is the way forward for you.

1 Like

I think S8 is a bit inaccurate in this shape. Black R12 would expose a follow up at R8, not S8 (see this related post-joseki).

If white R10 were at Q10, then black would have a follow up at S8 (see this post-joseki or this related post-joseki).

1 Like