Game review request (IGS 17K)

xoxox · Post by **xoxox** » Sun Oct 28, 2018 2:51 pm

Any kind of criticism, suggestions are welcome. Game is attached below.

Review-01.sgf: (2.22 KiB) Downloaded 578 times

jlt · Post by **jlt** » Sun Oct 28, 2018 3:34 pm

The first remark I would make is that you should apply the proverb Corners, then sides, then center. For instance, :b7:

is an armpit hit, this is usually bad. At early stages of the game, it's better to approach a corner (for instance F3 or C6).

Another example is move 123. You are playing on the fifth line, but to make territory you should play on the third line (or sometimes the fourth).

Also, look at the whole board. After :w32:

, White has a huge wall, and threatens to make a very large territory on the lower edge, so you should try to play towards that area (or better, you should have noticed that after :w20:

). Other example : :b53:

doesn't make many points. There are much larger areas on the board.

A few other remarks:

:b15:

is almost useless. Your group is already safe, there is no need to reinforce it. Instead, you can approach a corner, or pincer the weak O3 group.

:b17:

I don't see any reason not to use the solid connection Q3 (assuming you want to connect the stones).

:b27:

you capture one stone but the global result is bad after :w30:

because White's wall becomes very strong. (In fact, White could have omitted :w30:

and in that case you would have lost the initiative, which is also bad.) You should leave these kinds of small captures for the endgame.

(I can see other bad moves on both sides but won't comment everything.)

EdLee · Post by **EdLee** » Sun Oct 28, 2018 7:04 pm

EdLee · Post by **EdLee** » Sun Oct 28, 2018 9:43 pm

Hi xoxox,

Curious about the engine's opinion.

:b9:

Strange.

This one depends on your plan...

:b15:

...if this was your plan for :b13:

, then it's a bad plan ( very slow ).
If you didn't have a plan for :b13:

, which is very likely, then it makes sense you ended up with a bad combination of ( :b13:

+

= too slow ).

:b17:

Maybe you played this move because in the past, you had some bad experience with your opponents jumping into your 3-3 and destroyed your corner ?
If so, that's misplaced thinking here. Just connect solidly at Q3.
Study why Q3 is better than your :b17:

.

Very slow, almost a pass. Moreover, ( :b21:

+

) is good for :white:

.

Ditto: very slow, almost a pass.

Without looking at the rest of the game, we can already see from ( :b15:

,

) that you very often play very slow, very small moves. This is due to errors in your local and global evaluation ( assessment of the whole board situation ).

Can you see why ( :b15:

,

) are very slow and small ?

jlt · Post by **jlt** » Tue Oct 30, 2018 5:11 am

@EdLee: I don't have access to LeelaZero right now, but I just checked on Leela 0.11.0 what it thinks about move 7.

After move 7 is played, and after 60000 simulations, the position evaluation is about B+0.3.
If instead of R9 Black plays C6, the position is B+2.7
If instead of R9 Black plays F3, the position is B+2.2
If instead of R9 Black plays R8, the position is B+1.5

I don't have the patience to check other moves, as my computer is rather slow. In summary, move 7 is not a big mistake but is considered as a ~2 point loss compared to approaching the lower left corner with C6 or F3.

EdLee · Post by **EdLee** » Tue Oct 30, 2018 5:31 am

Hi jlt,

Thanks. 2% for a super-human engine is quite negligible for ddk or sdk. Thanks.

Tryss · Post by **Tryss** » Tue Oct 30, 2018 5:42 am

LZ think it's closer to a -4% mistake, with black winrate going from 57% to 53%. She prefer to play on the lower left (3-3 or usual approach at C6/F3)

But :b9:

is worse (-4.6%, with black winrate now under 50%)

jlt · Post by **jlt** » Tue Oct 30, 2018 6:23 am

EdLee: not 2% but 2 points.

xoxox · Post by **xoxox** » Tue Oct 30, 2018 7:29 am

jlt wrote: For instance, is an armpit hit, this is usually bad.

Yes, this is usually bad but in Chinese fuseki I have seen this move in books, video analysis etc. It might not be so bad here and probably depends upon follow up moves, which I don't think I played very accurately.

xoxox · Post by **xoxox** » Tue Oct 30, 2018 7:39 am

jlt wrote:Another example is move 123. You are playing on the fifth line, but to make territory you should play on the third line (or sometimes the fourth).

Yes, I realised that soon after making that move. I may be was thinking about influence or something like that, which I think is wrong idea in given position. One can only learn by playing wrong moves first.

xoxox · Post by **xoxox** » Tue Oct 30, 2018 7:53 am

jlt wrote:Also, look at the whole board. After , White has a huge wall, and threatens to make a very large territory on the lower edge, so you should try to play towards that area (or better, you should have noticed that after ). Other example : doesn't make many points. There are much larger areas on the board.

I think move :b31:

was total waste and I could have played better. Actually this sequence helped white build some influence with future chances of large moyo. I usually do really bad in joseki/fuseki stage. I still don't understand why :b53:

is a small move? Doesn't that secure territory more than 2 points?

jlt wrote:(I can see other bad moves on both sides but won't comment everything.)

Any suggestions about middle game play?

xoxox · Post by **xoxox** » Tue Oct 30, 2018 8:03 am

EdLee wrote: Curious about the engine's opinion.

Strange.

do LZ agree with my move? I am on a mobile device, have app with LZ+FB elf weights but haven't checked it as I am more interested in human moves than that of AI's for now.

:b9:

why it is strange? Is it because other bigger moves are available? Should I have played on another star point or an approach move in white corner?

xoxox · Post by **xoxox** » Tue Oct 30, 2018 8:14 am

EdLee wrote: This one depends on your plan...

...if this was your plan for , then it's a bad plan ( very slow ).
If you didn't have a plan for , which is very likely, then it makes sense you ended up with a bad combination of ( + = too slow ).

Maybe you played this move because in the past, you had some bad experience with your opponents jumping into your 3-3 and destroyed your corner ?
If so, that's misplaced thinking here. Just connect solidly at Q3.
Study why Q3 is better than your .

+

was part of the plan, now I realise that :b15:

was just a stupid (and unnecessary) response to :w14:

which could have been omitted completely. Even this plan was slow.

You are partially right about san-san invasion but I planned this because my opponent was also playing slow, or I might be wrong.

xoxox · Post by **xoxox** » Tue Oct 30, 2018 8:24 am

EdLee wrote:Can you see why ( , , ) are very slow and small?

Yes, what alternate moves you suggest in tune with chinese fuseki?

xoxox · Post by **xoxox** » Tue Oct 30, 2018 8:40 am

Instead of :b7:

LZ suggest early san-san invasion at C3 which I think is really a bad move for beginner like me. It might work for AZ or LZ but to a new learner of the game it might not be a 'necessary' move.

Life In 19x19

Game review request (IGS 17K)

Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)

Re: Game review request (IGS 17K)