[Welcome to Sensei's Library!]

StartingPoints
ReferenceSection
About


Paths
StatisticalAnalysis

 

Enclosing from hoshi versus shimari
Path: StatisticalAnalysis   · Prev: StaircaseStatistics   · Next: NumberOfDifferentFuseki
    Keywords: Opening

I believe it is "well known" among amateurs that there are no particularly effective follow-up moves in the corner after one starts with a play at hoshi. This is what most joseki books tell us. Nevertheless playing a small knight or large knight extension from a hoshi stone seems to be becoming somewhat popular in recent professional play. I wanted to investigate whether this has led to favorable results. As usual I turned to GoGoD CD and Kombilo. The results were very surprising!

I looked at the period 1990 to date (July 2002) and compared three choices:

Playing a small knight's move enclosure

[Diagram]
Diag.: Small knight's shimari

(Actually this is a small lie, there were 55 cases of making the shimari by starting with the 5-3 point :-)

Playing a small knight's extension from a hoshi stone

[Diagram]
Diag.: Small knight's extension


Playing a large knight's extension from a hoshi stone

[Diagram]
Diag.: Large knight's extension


Note that the search space includes empty adjacent star points so this excludes situations where the player has already extended to the middle of the board (e.g. a san-ren-sei pattern).

All Games 1990 to date (for reference)

  • 6370 games
  • Black won 53.1%
  • White won 46.8%
  • Precision 1.3%

Small knight' shimari

  • 1472 games; 1243 shimari played by Black, 229 shimari played by White
  • Expected (average) winning percentage for this ratio of Black and White games = 52.1%
  • Player making shimari won 51.0% (precision 2.6%)
  • Performance versus expected result = -1.1 percentage points

Small knight's extension

  • 393 games; 176 extension played by Black, 217 extension played by White
  • Expected (average) winning percentage for this ratio of Black and White games = 49.6%
  • Player making extension won 54.5% (precision 5.0%)
  • Performance versus expected result = +4.9 percentage points

Large knight's extension

  • 355 games; 143 extension played by Black, 212 extension played by White
  • Expected (average) winning percentage for this ratio of Black and White games = 49.3%
  • Player making extension won 54.6% (precision 5.3%)
  • Performance versus expected result = +5.3 percentage points

The number of games involved did not provide a very high degree of precision. However, the results are certainly suggestive. Have you tried this in your own games lately? --DaveSigaty


adamzero: In my own games, I've noticed that the small knights extension off the 4-4 works particularly nicely. I like to play fairly profit-oriented game, making strong groups from which I can attack or invade without fear. Often my opponent builds moyo that I have to invade, but because this happens so frequently invading has become one of the better parts of my game. The small knights extension off the hoshi works well for me because it threatens to grab very large profit, inviting my opponent to challenge, and I can then attack strongly. I particularly like playing it when it leaves the open side of the hoshi stone facing my own thickness. I've had good results with this. I don't like the large knights extension because it seems to invite the san-san invasion too strongly, and I'm rarely satisfied with the result. And when I'm playing against it I immediately start trying to figure out the appropriate timing for the san-san invasion, and having played it I am usually pleased with my gains.



Path: StatisticalAnalysis   · Prev: StaircaseStatistics   · Next: NumberOfDifferentFuseki
This is a copy of the living page "Enclosing from hoshi versus shimari" at Sensei's Library.
(OC) 2003 the Authors, published under the OpenContent License V1.0.