Deep Position Analysis with Fritz

    Sometimes you want a really deep analysis of a particular position and there are several ways of doing it. Years ago the program Aquarium introduce IDeA (Interactive Deep Analysis), a feature that was all the rage. 

    There was nothing wrong with the program and their customer support was outstanding, but I didn’t like the feature because the learning curve was too steep and I didn’t have the patience! 

    Fritz has something similar, but the disadvantage is that once you stop the analysis the program does not remember it and the next time you access the saved analysis you have to start over. IDeA picks up the project where it left off. 

    In a fit of boredom, I wanted to see see how Fritz' Deep Position Analysis, which I never use, compares to Infinite Analysis, so I did a DPA on a critical position from an old correspondence game of mine and gave the program 30 minutes, but the process took only about 15 minutes. 

    According to the manual the DPA function is very good for getting deep and detailed analysis of a critical position and it is especially useful for correspondence players and, I suppose, opening research. 

    You can determine how deep and broad the tree should be. Using DPA allows you to set time to analyze or the depth to which it should go. You set the number of Branches. i.e. how many alternative moves are given. You set the length of analyzed variations. And, you determines when a variation will be discarded. e.g. the value of 100 (one Pawn) means that moves that are a Pawn worse than the best move will not be investigated. 

    At the end of the DPA all the variations that have been generated are sorted and the strongest move becomes the main line. Here were the results.   

    .

[Event "Correspondence, CCLA"] [Site "?"] [Date "1960.??.??"] [Round "?"] [White "Tartajubow"] [Black "Opponent"] [Result "1-0"] [Annotator "Stockfish 17.1"] [SetUp "1"] [FEN "3r1rk1/p3bpp1/1pqp1n1B/4p3/3QPPPP/2N5/PPPR4/2K4R w - - 0 17"] [PlyCount "8"] [EventDate "2005.??.??"] 17. Qe3 {In the game black played 17...Nxg4?} d5 (17... gxh6 18. g5 (18. f5 d5 19. Qxh6 dxe4 20. Rg2 e3 $11 {[%eval 0,33]}) 18... exf4 (18... Nh5 19. Nd5 Rde8 20. f5 Nf4 $16 {[%eval 142,25]}) (18... Nxe4 19. Nd5 (19. Nxe4 d5 20. Nf6+ Kh8 $18 {[%eval 180,23]}) 19... Qa4 20. Nxe7+ Kh7 $18 {[%eval 188,23]}) 19. Qxf4 ( 19. Qf3 Nh7 20. Nd5 Rde8 $15 {[%eval -33,20]}) (19. Qg1 Kh8 20. gxf6 Bxf6 $15 { [%eval -42,18]}) 19... Nh7 20. Nd5 Rde8 $16 {[%eval 71,29]}) 18. fxe5 (18. exd5 Nxg4 (18... Nxd5 19. Qxe5 Nf6 20. Rhd1 Rxd2 $11 {[%eval 29,25]}) 19. dxc6 (19. Qxe5 Nxe5 20. dxc6 Rxd2 $15 {[%eval -31,19]}) (19. Bg5 Nxe3 20. dxc6 f6 $17 { [%eval -115,17]}) (19. Rg2 Qxh6 20. Rxg4 f5 $15 {[%eval -68,17]}) 19... Nxe3 20. Rg1 Rxd2 $11 {[%eval 12,27]}) 18... Nxg4 19. Nxd5 Rxd5 20. exd5 Qxh6 $11 { [%eval 19,21]} 1-0

Now, compare that to 30 minutes using Infinite Analysis (set to show the top three lines). Notice there is a major diffeance at move 18.  

 

[Event "Correspondence, CCLA"] [Site "?"] [Date "1960.??.??"] [Round "?"] [White "Tartajubow"] [Black "Opponent"] [Result "1-0"] [Annotator "Stockfish 17.1"] [SetUp "1"] [FEN "3r1rk1/p3bpp1/1pqp1n1B/4p3/3QPPPP/2N5/PPPR4/2K4R w - - 0 17"] [PlyCount "77"] [EventDate "2005.??.??"] 17. Qe3 {In the game black played 17...Nxg4?} {} d5 ({ Variation 1:} 17... gxh6 18. g5 exf4 19. Qxf4 Nh7 20. Nd5 Rde8 21. gxh6 Kh8 22. Rg2 Qd7 23. Rhg1 Qe6 24. Nc7 Qf6 25. Qe3 Qxh4 26. Nxe8 Rxe8 27. Qf3 Qf6 28. Qg3 Qg6 29. Qxg6 fxg6 30. Rxg6 Bf6 31. Rf1 Be5 32. Rf7 a5 33. c3 a4 34. Ra7 b5 35. Kc2 Rc8 36. Kd1 Nf6 37. Rgg7 Rb8 38. Ke2 Nxe4 39. Rh7+ Kg8 40. Rhb7 Rxb7 41. Rxb7 Nc5 42. Ra7 Bf4 43. h7+ Kh8 44. Kf3 Be5 45. Kg4 Bg7 46. Ra5 Kxh7 47. Kf5 Kh6 48. Rxb5 a3 49. b4 Bxc3 50. bxc5 dxc5 51. Rb6+ Kh7 {[%eval 55,45]}) ({ Variation 2:} 17... exf4 18. Bxf4 d5 19. exd5 Nxd5 20. Qf3 Nb4 21. Qxc6 Nxc6 22. Nd5 Bc5 23. Rhd1 Rde8 24. c3 Re4 25. g5 a5 26. Rf1 Nd8 27. Bg3 b5 28. Kc2 Ne6 29. Nf4 a4 30. Rd7 Nxf4 31. Bxf4 Re2+ 32. Bd2 g6 33. Re1 Rg2 34. Rd5 Rc8 35. Re4 a3 36. bxa3 Bxa3 37. Rxb5 {[%eval 159,44]}) 18. exd5 Nxg4 19. dxc6 Nxe3 20. c7 Rxd2 21. Kxd2 Nf5 22. fxe5 gxh6 23. Rg1+ Ng7 24. Rg4 Rc8 25. Nd5 Bf8 26. h5 Kh8 27. Ra4 a5 28. Rf4 Kg8 29. Rc4 Ne6 30. a4 Kh7 31. Rc6 Bg7 32. Nxb6 Rxc7 33. Rxc7 Nxc7 34. Nd7 Ne6 35. Kd3 Nf8 36. Nxf8+ Bxf8 37. c3 Kg8 38. Kd4 Be7 39. Kc4 Kf8 40. Kb5 Bd8 41. b4 Ke7 42. Kc6 Ke6 43. b5 Kxe5 44. b6 Bxb6 45. Kxb6 f5 46. c4 f4 47. c5 f3 48. c6 f2 49. c7 f1=Q 50. c8=Q Qb1+ 51. Kxa5 Qe1+ 52. Ka6 Qe2+ 53. Kb7 Kf4 54. Qc5 Qe4+ 55. Kc7 {[%eval 15,45]} 1-0

    The bottom line (I think) is that there is little reason to bother with the Deep Position Analysis feature and simply use Infinite Analysis.

Capablanca Dodges a Bullet

    The 1931 New York Invitational was sponsored by the Manhattan Chess Club. Capablanca won with little difficulty although he did have a lost ending (surprise!) against Arthur Dake. Unfortunately for Dake, he started rushing his moves and refused to adjourn and eventually lost. 

    Capablanca also dodged a bullet in his game against Frank Marshall. The game appears in Harry Golombek’s book Capablanca’s Hundred Best Games of Chess that was published in 1946. I consider Golombek to have been a good writer, but he missed the mark in this game when he praised Capablanca's tactic at move 22 when, in fact, it should have lost. You can hardly blame Golombek for missing the refutation though! After all, both Marshall and Capablanca missed it, too! 

[Event "New York"] [Site ""] [Date "1931.04.29"] [Round "9"] [White "Frank Marshall"] [Black "Jose Capablanca"] [Result "0-1"] [ECO "E16"] [Annotator "Stockfish 17.1"] [PlyCount "72"] [EventDate "1931.04.18"] {E16: Queen's Indian} 1. Nf3 Nf6 2. d4 e6 3. c4 b6 4. g3 Bb7 5. Bg2 Bb4+ 6. Bd2 Bxd2+ 7. Nbxd2 {This is not as good as 7.Qxd2 because the N belongs on c3 so white can answer the coming ...c5 with d5.} O-O 8. O-O c5 {The freeing move. It's instructive to see how Capablanca manages to utilize his advantage on the Q-side.} 9. dxc5 bxc5 10. Rc1 {While not bad, this is a routine move that serves no real purpose. Normal is 10.Qc2} Qc7 11. Nb3 {This is a rather poor move that leaves the N out of play. It would have been betder to admit his 7th move was not so good and reposition it with 11.Nb1 and 12.Nc3. Capablanca already has a slight advantage, but that's all he needs.} d6 12. Qd2 Nc6 13. Rfd1 Rfd8 {Black is slightly better.} 14. Nh4 {In this game at least, Marshall's handling of his Ns leaves a lit to be desired! Here there is no possibility of a K-side attack. Even engines are having a hard time suggesting anything positive. Who would want to play Stockfish’s suggestions of 14.Qe3 or in that is probably the best idea, repairing the Ns poor position with 14. Na1 and 15.Nc2?} a5 15. a4 {Another poor move; it weakens the Q-side and gives black's N a strong outpost on b4. White's best idea was still repositioning the N as mentioned witi Na1-c2} Rab8 16. Rc3 Ba8 17. h3 {This part of a K-side attack that never comes to fruition. At this poimt white is already strategically lost.} Nb4 18. Bxa8 Rxa8 19. Qf4 {Part of his hoped for attack on f7, but there was nothing more promising.} Qc6 20. Rf3 {The next part of his plam is to drive the N away wiht g4 and g5} Rd7 {This defends f7, but there is also a hidden point...it's part of a tactical trick that is not quite correct.} 21. g4 Qxa4 {Capablanca plays a nifty little tactical trick, but it dissipates his advantage. In his book of Capablanca’s best games Harry Golombek praises Capablanca's idea, but these days any patzer with an engine can show it’s not the best line. So, what should he have played?} (21... Ne8 22. g5 {Beating the air.} Rb7 {White has no useful continuation and he going to lose the a-Pawn.}) 22. Rxd6 Nbd5 {[%mdl 8192] Golombek called this "a neat little combination, winning a P by force." It's actually what should have been a losing move that's worthy of a couple of question markd.} (22... Rxd6 { results in equality after} 23. Qxd6 Qd7) 23. Qe5 {[%mdl 8192] It's Marshall's turn to play a move worthy of two question marks.} (23. Ra6 {An amazing move.} Rdd8 (23... Rxa6 {23...Rda7 3.29} 24. Qb8+ Rd8 25. Qxd8+ Qe8 26. Qxe8+ Nxe8 27. cxd5 {and white has won a piece.}) 24. Rxa8 Rxa8 25. cxd5 {White is up a piece, }) 23... Rxd6 $19 24. Qxd6 {Now things are back on track and Capablanca has managed to come out with what is still a winning position.} Ne4 25. Qe5 Qxc4 { Capablanca is a P up and his Q-side attack is still going strong. He finishes up neatly.} 26. Rd3 a4 27. f3 Nef6 28. Nd2 Qc1+ 29. Kf2 h6 30. f4 c4 {Black is clearly winning.} 31. Rd4 c3 {Creating a passed a-Pawn.} 32. bxc3 a3 {[%mdl 32] } 33. g5 a2 34. Nb3 Qxc3 35. gxf6 Qxb3 36. Rd1 Qxd1 {White resigned.} 0-1