Monday 31 August 2020

Hastings - A Potted History

THE first Hastings international chess congress was held in 1895 - a time when chess congresses of any kind were rare.
The second did not occur until 1919, when a tournament was held to mark victory in World War One.
This prompted the start of an annual event held over the New Year, starting in 1920-21 and supplemented by special summer congresses in 1922 and 1995.
There have been 95 New Year congresses - five were cancelled in the 1940s due to Adolf Hitler.
(You could blame 'war' or 'wartime regulations' for the cancellations, but everyone knew who was really to blame.)
Fast forward to 2020 and we were looking forward to the 96th New Year congress - the 100th overall Hastings international chess congress.
Then came covid-19, but many people thought Hastings would go ahead, like Biel, Prague and other prestigious international events, albeit in a modified format.
But such thinking failed to take into account that Hastings has got itself into a situation where it is largely beholden to a single sponsor.
That sponsor is Caplin Systems, a financial software firm run by John Ashworth, a keen but not very informed chess player (his ECF grade is 81).
A decision on how this year's event would be organised was due to be taken by the organising committee on August 22.
But before the committee met Ashworth gave an interview to The Chess Pit in which he claimed the vast majority of Hastings participants are in the "higher-risk category."
Furthermore he said there was a need to be "exceptionally cautious."
Now when your chief sponsor says "Jump!", you don't ask " Why?" - you ask "How high?"
And so it came to pass that, for the second time in its history, Hastings has been cancelled.
You could blame covid-19 or government regulations for the latest cancellation - you could, but I don't.

Champion Of Champions (part three)

HERE is how the tournament bracket looks after two matches.

Round of 16               Quarter-Finals
1. Steinitz
v---------------------------Carlsen
16. Carlsen (+102.5)

8.Tal (+8.5)
v----------------------------Tal
9.Petrosian

6. Botvinnik
v--------------------------
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

Round of 16
Match Three: Botvinnik v Fischer
Game One
Mikhail Botvinnik - Bobby Fischer
King's Indian Defence
1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6
Fischer also played the provocative 4...0-0, but usually preferred with the text.
5.f3
The Sämisch Variation was overwhelmingly Botvinnik's favourite response to the King's Indian.
5...e5
Played by Fischer 12 times in ChessBase's 2020 Mega database compared with 11 for 5...0-0.
6.Nge2 0-0 7.Bg5 c6 8.Qd2 Qa5
A position that does not occur in Botvinnik's games
White scores very well from here in Mega20, a fact reflected in an averaged evaluation from Stockfish11 and Komodo11.01 of +71.5 for Botvinnik, giving Fischer a tough task in game two.

Game Two
Bobby Fischer - Mikhail Botvinnik
French Winawer
1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.e5 c5 5.a3 Bxc3+ 6.bxc3
I do not think many chess fans will be surprised to see this position appear on the board in a 'game' between Fischer and Botvinnik.
6...Ne7 7.a4!?
Fischer did not believe in the attacking, but committal, 7.Qg4.
7...Nbc6 8.Nf3 Bd7 9.Bd3
We have reached a position that does not occur in Botvinnik's games
Fischer famously had trouble against the French Defence, despite publicly calling the Winawer unsound., so perhaps it is not surprising the engines give him an averaged score here of just +5.5.
That means Botvinnik wins the match with a score of +66, giving the older generation two wins out of three against their more-modern brethren.
Here is the updated tournament bracket.

Round of 16               Quarter-Finals
1. Steinitz
v---------------------------Carlsen
16. Carlsen (+102,5)

8.Tal (+8.5)
v----------------------------Tal
9.Petrosian

6. Botvinnik (+66)
v----------------------------Botvinnik
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

Sunday 30 August 2020

Hastings Update

EXPECT an announcement mid-September on whether there will be some sort of replacement for the 2020-21 congress, for example a smaller congress in the summer. There may also be a tentative announcement re hopes to resume normal service and hold a 2021-22 congress.

Champion Of Champions (part two)

HERE is how the tournament bracket looks after one match.

Round of 16               Quarter-Finals
1. Steinitz
v----------------------------Carlsen
16. Carlsen (+102.5)

8.Tal
v--------------------------
9.Petrosian

6. Botvinnik
v--------------------------
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

Round of 16
Match Two: Tal v Petrosian
Game One
Mikhail Tal - Tigran V Petrosian
French Tarrasch
1.e4 e6
This narrowly beats 1...c5 by 164 appearances to 160 in ChessBase's 2020 Mega database as Petrosian's favourite reply to 1.e4.
2.d4 d5 3.Nd2!?
Another narrow decision - Tal preferred this over 3.Nc3 by 49 games to 46.
3...c5 4.Ngf3 Nf6 5.exd5 exd5
Petrosian has four games with the text and four with 5...Nxd5. He scored 50% with both moves, but the former wins on tiebreak as none of the 5...Nxd5 games were when he had a Fide rating.
6.Bb5+ Bd7 7.Bxd7+ Nbxd7 8.0-0 Be7 9.dxc5 Nxc5 10.Nb3 Nce4
Finally we have reached a position that does not occur in Tal's games
Stockfish11 and Komodo11.01 agree this position is better for White, their evaluations averaging as a score for Tal of +44.5

Game Two
Tigran V Petrosian - Mikhail Tal
QGD Semi-Tarrasch
1.d4 Nf6 2.c4 e6 3.Nf3
Petrosian narrowly preferred the text to 3.Nc3 by 118 games to 106.
3...d5 4.Nc3 c5!?
An aggressive response, much in Tal's style.
5.e3!?
The main move is 5.cxd5, which Petrosian played less often but with which he scored much better.
5...Nc6 6.cxd5 Nxd5 7.Bc4 cxd4 8.exd4 Be7 9.0-0 0-0 10.Re1 Nxc3
This position does not appear in Petrosian's games
The engines prefer White, but their averaged evaluations give Petrosian a score of +36, which is not enough to overcome Tal's +44.5 from game one.
Here is the updated bracket after two matches.

Round of 16               Quarter-Finals
1. Steinitz
v----------------------------Carlsen
16. Carlsen (+102.5)

8.Tal (+8.5)
v----------------------------Tal
9.Petrosian

6. Botvinnik
v--------------------------
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

Carlsen against Tal should make for an interesting first quarter-final.

Saturday 29 August 2020

Champion Of Champions

I WAS musing the other day on the fact there have been 16 classical world chess champions - a figure ideal for organising a knockout tournament.
If the champions were seeded by chronological order, this is what the bracket would look like.

Round of 16
1. Steinitz
v-------------------------
16. Carlsen

8.Tal
v--------------------------
9.Petrosian

6. Botvinnik
v--------------------------
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

There are some mouthwatering match-ups, including the opening specialists Kasparov and Alekhine going head-to-head and the clash between champion-of-the-West Fischer and communist hero Botvinnik.
But how to decide the matches?
Just for fun, I am going to organise two-game mini-matches based on the players' games in ChessBase's 2020 Mega database.
In a 'game' each champion will make the move in any given position that he has played most often in Mega20.
When a position is reached which does not appear in a champion's repertoire, the game will be adjudicated by my main analysis engines Stockfish11 and Komodo11.01.
Whichever champion scores most over a two-game match will advance to the next round.
As I say, this is just for fun, but what else is there to do in lockdown?

Round of 16
Match One: Steinitz v Carlsen
Game One
Wilhelm Steinitz - Magnus Carlsen
Evans' Gambit
1.e4
Steinitz played 1.e4 in almost 80% of his games with the white pieces.
1...e5
Carlsen's reply is not so predictable, but the text has been more popular with him than 1...c5.
2.Nf3
Perhaps surprisingly this move is not the runaway winner. It appears 85 times in Steinitz's games, compared with 78 for 2.Nc3 and 65 for 2.f4.
2...Nc6 3.Bc4
Steinitz played this more than twice as often as 3.Bb5.
3...Bc5 4.b4!?
We have what is widely regarded as an archetypal 19th-century opening.
4...Bb6
This has been Carlsen's response both times he faced the Evans.
5.b5
Game One ends here as we have reached a position that does not appear in Carlsen's games
Both engines evaluate this position as better for Black. Averaging their two evaluations, once the evaluations become stable, gives Carlsen a score of +52 (in other words Black is 52 hundredths of a pawn better) - a great start with the black pieces.

Game Two
Magnus Carlsen - Wilhelm Steinitz
Spanish Cozio Deferred
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6!?
No Steinitz Defence - the great Czecho-Austro-American (Steinitz was born in Prague, then part of the Habsburgs' Austrian empire, but later became a US citizen) played Morphy's 3...a6 more frequently than he played his own defence, 3...d6.
4.Ba4 Nge7
A Steinitz favourite, beating out the Steinitz Defence Deferred: 4...d6. I guess 4...Nge7 is a Cozio Deferred.
Another short game, but again we have reached a position that does not occur in Carlsen's games
The engines again much prefer Carlsen's position, giving him an averaged score of +50.5, which adds up to a match score of +102.5 - a resounding victory.
So after one match the tournament bracket looks like this.

Round of 16                   Quarter-Finals
1. Steinitz
v-------------------------------Carlsen
16. Carlsen (+102.5)

8.Tal
v--------------------------
9.Petrosian

6. Botvinnik
v--------------------------
11. Fischer

13. Kasparov
v--------------------------
4. Alekhine

3. Capablanca
v--------------------------
14. Kramnik

5. Euwe
v--------------------------
12. Karpov

7. Smyslov
v--------------------------
10. Spassky

15. Anand
v--------------------------
2. Lasker

Friday 28 August 2020

Hastings Cancelled

THE annual Hastings congress, due to start on December 28, has been cancelled.
Congress chairman Marc Bryant states: "The 2020-21 Caplin Hastings international chess congress will regrettably not take place in its usual form.
"Covid-19 regulations make this impossible in relation to social distancing and number of competitors that could be accommodated.
"However Caplin Systems and the organisers of the congress will run an alternative small event. Details of this will be announced in the near future."
More info here: http://www.hastingschess.com/committee-meeting-congress-news/

Best By Test (part eight)

GOING back to one of the tables in the first part of this series, it showed White's most successful move in modern chess, from the viewpoint of rating performance, is 1.d4.

Table Three: Opening Moves By Rating Performance
Move..........Rating Performance
1.d4.....................+26
1.e4.....................+20
1.c4.....................+14
1.Nf3...................+11
1.b4.....................+6
1.f4......................+4
1.b3.....................+2
1.Nc3...................-17
1.g3.....................-23

This table is based on 238,275 games played last year that appear in ChessBase's 2020 Mega database. Moves occurring less than 0.1% of the time have been excluded.
A score of +26 means players using that move performed on average at 26 points above their elo rating.
The first place of 1.d4 is by a convincing margin - you have to go down to the difference in performance between 1.b3 and 1.Nc3 to find a larger gap.
Here I want to look at what is behind 1.d4's success statistically.

Table 28: Black Replies To 1.d4 By Popularity
Move..........Number Of Times Played
1...Nf6..................44,201
1...d5....................23,149
1...e6......................3,517
1...f5.......................2,298
1...d6......................2,215
1...g6......................2,051
1...c5......................1,028
1...c6.........................256
1...b6.........................153
1...Nc6.......................142

In table 28, from which moves occurring less than 0.1% of the time have been excluded, the popularity of 1...Nf6 easily exceeds all other moves combined.

Table 29: Black Replies To 1.d4 By Percentage Score
Move......................Score
1...c5.......................51%
1...g6.......................50%
1...d6.......................49%
1...Nf6.....................48%
1...f5........................47%
1...e6........................45%
1...d5........................42%
1...b6........................40%
1...Nc6.....................39%
1...c6........................34%

None of the popular moves does well when it comes to scoring percentage, but now we can look at rating performance.

Table 30: Black Replies To 1.d4 By Rating Performance (tie broken by frequency of appearance)
Move.................Rating Performance
1...Nf6......................-21
1...d5........................-21
1...e6........................-32
1...c6........................-34
1...f5........................-40
1...c5........................-44
1...d6........................-45
1...g6........................-51
1...b6........................-59
1...Nc6......................-117

Table 30 has a much more conventional look to it, although it is quite a coincidence that the two most-popular moves should give an identical rating performance.

Thursday 27 August 2020

Best By Test (part seven)

WHY does 1...e5 do so badly as a reply to 1.e4, both percentage-wise and in rating performance?
We saw in part two of this series that 1...e5 scores just 44%, which is below every other move, apart from 1...b6, that occurs at least 0.1 per cent of the time.
Its rating performance in those games is -28, well behind 1...e6, 1...d5, 1...c5 and 1...c6.
All statistics are based on 238,275 games played last year that appear in ChessBase's 2020 Mega database.
In the popularity table below I have excluded moves occurring less than 0.1% of the time.

Table 22: Popularity Of White Moves After 1.e4 e5
Move.......Number Of Times Played
2.Nf3.....................26,326
2.Bc4.......................1,192
2.Nc3..........................975
2.f4.............................519
2.d4............................405
2.d3..............................71
2.c4..............................63

The move 2.Nf3 is overwhelmingly most popular, occurring almost 90% of the time.

Table 23: White Second Moves By Percentage Score (ties broken by frequency of appearance)
Move.......................Score
2.d4..........................58%
2.Nf3........................56%
2.f4...........................56%
2.Bc4........................55%
2.Nc3........................55%
2.c4...........................37%
2.d3...........................32%

Table 23 is very tight at the top, with just three percentage points covering the most-successful five moves, the only surprise, perhaps, being 2.d4 coming out on top.

Table 24: White Second Moves By Rating Performance (tie broken by frequency of appearance)
Move..................Rating Performance
2.d4............................+39
2.Nf3..........................+32
2.f4.............................+3
2.Bc4..........................=
2.Nc3.........................-12
2.c4............................-12
2.d3............................-72

The move 2.d4 confirms its success, although White's traditional path of 2.Nf3 also scores very well.
The performance of the latter is probably of most interest to blacks who answer 1.e4 with 1...e5 as 2.Nf3 is played almost 90% of the time.
In the table below I have excluded moves occurring less than 0.1% of the time.

Table 25: Popularity Of Black Moves After 1.e4 e5 2.Nf3
Move..................Number Of Times Played
2...Nc6.............................22,580
2...Nf6...............................2,726
2...d6....................................896
2...d5......................................79
2...f5.......................................66
2...Qe7...................................44
2...Bc5....................................31

No major surprises here, unless you count that of 2...Bc5 passing the 0.1% threshold.

Table 26: Black Second Moves By Percentage Score (tie broken by frequency of appearance)
Move...............................Score
2...Nc6.............................45%
2...Qe7.............................45%
2...f5.................................43%
2...Nf6..............................40%
2...d6................................38%
2...d5................................33%
2...Bc5..............................21%

The success of 2...Qe7 is certainly surprising, but is from a small sample size.

Table 27: Black Second Moves By Rating Performance
Move.......................Rating Performance
2...Bc5..............................+19
2...Nf6...............................-14
2...d5.................................-24
2...Qe7..............................-25
2...Nc6..............................-31
2...f5..................................-47
2...d6.................................-49

I certainly did not expect this, but it should very much be borne in mind that 2...Bc5's success is based on just 31 games.
Many of those games featured low-rated and unrated players, which can be seen in that only five games saw the reply that is clearly the best, ie 3.Nxe5.

Wednesday 26 August 2020

Best By Test (part six)

ARGUABLY the biggest surprise from statistically analysing the 238,275 games played last year and appearing in ChessBase's 2020 Mega database is the success of 1...e6 in reply to 1.e4.
The move 1...e6 turns in a rating performance of just -10, ie the black players performed just 10 elo below their rating.
Here, as a reminder, is how the scores rank.

Table Six: Black Replies By Rating Performance (tie broken by frequency of appearance)
Move.........Rating Performance
1...e6.....................-10
1...d5.....................-18
1...c5.....................-19
1...c6.....................-19
1...e5.....................-28
1...d6.....................-41
1...Nf6...................-44
1...g6.....................-57
1...b6.....................-70
1...Nc6...................-71

As can be seen, 1...e6 is a clear winner, so it should be interesting to see how blacks are racking up this score.
In the table below I have left out moves that occur less than 0.1% of the time.

Table 16: White Moves After 1.e4 e6
Move.........Number Of Times Played
2.d4.........................12,081
2.d3..............................903
2.Nf3............................713
2.Qe2...........................246
2.Nc3............................215
2.c4...............................141
2.b3...............................140
2.f4..................................65
2.g3.................................21
2.e5.................................18

No shocks there, at least to me, although I am a bit surprised 2.b3 was played (slightly) less often than 2.c4.

Table 17: White Second Moves By Percentage Score (ties broken by frequency of appearance)
Move...........................Score
2.Qe2............................61%
2.c4...............................56%
2.d4...............................54%
2.d3...............................54%
2.b3...............................54%
2.g3...............................50%
2.f4................................43%
2.Nf3..............................41%
2.Nc3.............................41%
2.e5...............................39%

The move 2.Qe2 is a clear winner, and it is not as though the result comes from a particularly small sample size.
But as usual it is probably wise to pay more attention to rating performance.

Table 18: White Second Moves By Rating Performance
Move.......................Rating Performance
2.Qe2.............................+24
2.d4................................+18
2.b3................................+17
2.d3................................+11
2.Nc3.............................+5
2.c4................................-13
2.g3................................-42
2.Nf3..............................-47
2.f4.................................-96
2.e5................................-154

The move 2.Qe2 holds first place, and by a relatively clear margin, but there is not much to choose between the mainstream 2.d4 and 2.b3.
How should Black meet 2.Qe2, at least from a statistical viewpoint?
There are 246 games from last year in Mega20 with the position after 1.e4 e6 2.Qe2, as shown in the table below (I have ignored moves appearing fewer than 10 times).

Table 19: Black Moves After 1.e4 e6 2.Qe2
Move........Number Of Times Played
2...c5.......................130
2...Be7......................28
2...Nc6......................20
2...d5...........'.............19
2...e5.........................16

Could the popularity of 2...c5 be the key to 2.Qe2's success in that by playing it blacks are switching from a French set-up to a Sicilian one?

Table 20: Black Second Moves By Percentage Score
Move......................Score
2...Nc6....................50%
2...Be7....................45%
2...d5.......................42%
2...e5.......................41%
2...c5.......................40%

Popularity is no guarantee of success, it seems.

Table 21: Black Second Moves By Rating Performance
Move..............Rating Performance
2...e5.......................+37
2...Nc6.....................+6
2...Be7.....................=
2...d5.......................-15
2...c5.......................-18

Here is food for thought. The poor result of 2...c5 is confirmed, while the performance of 2...e5 is startling, albeit from a small number of games.

Tuesday 25 August 2020

Best By Test (part five)

ONE of the most intriguing of the Sicilian statistics is the success of the Alapin, ie 2.c3.
Its percentage score of 49% is nothing special, although it equals the score of 2.Nc3 and is only two percentage points behind 2.Nf3.
But 2.c3's rating performance of +40 is well clear of the field, in particular leading 2.Nf3 by 19 points and 2.Nc3 by 27 points.
Here I want to look at 2.c3's success, from a statistical viewpoint, in its 2,662 appearances in games played last year in ChessBase's 2020 Mega database (I will ignore moves occurring less than one percent of the time).

Table 13: Popularity Of Black Moves After 1.e4 c5 2.c3
Move.....Number Of Times Played
2...Nf6....................979
2...d5......................947
2...d6......................211
2...e6......................181
2...g6......................164
2...Nc6.....................93
2...e5........................53

I am not surprised at the modern popularity of 2...Nf6 and 2...d5, but I was not expecting 2...e6 to be less popular than 2...d6.

Table 14: Black Second Moves By Percentage Score (tie broken by frequency of appearance)
Move....................Score
2...Nf6..................56%
2...d5....................51%
2...e5....................49%
2...d6....................46%
2...e6....................43%
2...g6....................43%
2...Nc6.................32%

What is interesting here is that the moves' percentage scores, with the exception of 2...e5, are in the same order as the number of times they are played.

Table 15: Black Second Moves By Rating Performance
Move.............Rating Performance
2...d5....................-20
2...e5....................-34
2...Nc6.................-38
2...Nf6..................-40
2...e6....................-65
2...g6....................-65
2...d6....................-75

Table 15 shakes things up, in particular with the poor performances of the third-and-fourth most-popular moves, 2...d6 and 2...e6.
The move 2...e5 confirms the relative success of its percentage score, but this is from a small sample size.
It is striking how poorly 2...g6 does, both percentage-wise and in rating performance.
(To be continued)

Monday 24 August 2020

Best By Test (part four)

CONCENTRATING for the moment on 1.e4 c5 2.Nf3 it is interesting to see the popularity of Black's second-move choices in modern chess.
As usual I am statistically analysing from the 238,275 games played in 2019 that appear in ChessBase's 2020 Mega database, while ignoring those moves that appear less than 0.1% of the time.
There are 34,594 games that began 1.e4 c5 2.Nf3, so the cut-off point is 35 games.

Table 10: Popularity Of Black Moves After 1.e4 c5 2.Nf3
Move.....Number Of Times Played
2...d6....................13,952
2...Nc6...................9,649
2...e6......................8,246
2...g6......................1,806
2...a6.........................693
2...Nf6......................268
2...b6..........................42

The big three should surprise no one, and I expect most people will have expected them to appear in the order shown.

Table 11: Black Second Moves By Percentage Score (ties broken by frequency of appearance)
Move...................Score
2...b6....................58%
2...a6....................54%
2...e6....................50%
2...d6....................49%
2...Nc6.................49%
2...Nf6..................48%
2...g6....................46%

The success of the relatively unpopular moves, 2...a6 and especially 2...b6, is a major surprise, but as usual this may change when moves are ranked by rating performance.

Table 12: Black Second Moves By Rating Performance
Move..........Rating Performance
2...g6......................-9
2...b6......................-15
2...d6......................-16
2...e6......................-23
2...Nc6...................-25
2...Nf6....................-26
2...a6......................-43

So 2...b6 confirms its success, but 2...a6 collapses. The other big mover is 2...g6 - bottom percentage-wise but top in rating performance.
There is a big caveat with 2...b6,  however - the statistics are based on just 42 games.

Sunday 23 August 2020

Best By Test (part three)

BLACK'S overwhelmingly most-popular reply to 1.e4 is 1...c5 which, as we saw in part two, scores 50% with a rating performance of -19.
No fewer than 24 moves were tried by White on move two in the 238,275 games from 2019 that appear in ChessBase's 2020 Mega database.
As usual I will ignore those moves played less than 0.1% of the time. That leaves 14 moves to consider.

Table Seven: Popularity Of White Moves After 1.e4 c5
Move.....Number Of Times Played
2.Nf3...................34,594
2.Nc3....................5,408
2.c3.......................2,662
2.d4..........................972
2.d3..........................355
2.c4..........................335
2.b3..........................333
2.Bc4.......................309
2.Ne2.......................280
2.f4...........................226
2.g3..........................169
2.Be2........................135
2.b4..........................119
2.a3............................99

The most-popular move was always going to be 2.Nf3, but I am a little surprised it is so far ahead of all other moves combined.
Surprises for me are that 2.f4 and 2.g3 are so low on the list, and that 2.Na3 is not played enough to cross the 0.1% threshold.

Table Eight: White Second Moves By Percentage Score (ties broken by frequency of appearance)
Move.............Score
2.Ne2.............55%
2.Be2.............53%
2.Nf3..............51%
2.Nc3.............49%
2.c3................49%
2.b3................49%
2.g3................49%
2.a3................48%
2.b4................45%
2.d4................44%
2.d3................44%
2.f4.................44%
2.c4................39%
2.Bc4..............27%

The success of 2.Ne2 and 2.Be2 is the big story here, although as I write this I have yet to compile the rating performances which, as usual may tell a very different tale.
The miserable score of 2.Bc4 also surprises me.

Table Nine: White Second Moves By Rating Performance (ties broken by frequency of appearance)
Move.............Rating Performance
2.c3..........................+40
2.f4...........................+31
2.Be2........................+25
2.Nf3........................+21
2.d4..........................+17
2.Nc3.......................+13
2.Ne2.......................+13
2.b3..........................+12
2.d3..........................+8
2.g3...........................+7
2.c4...........................-5
2.b4...........................-12
2.Bc4........................-28
2.a3...........................-63

Well, 2.Ne2 does not repeat its percentage success, but 2.Be2 has again done very well.
Perhaps the latter wrong-foots blacks, while 2.Ne2 is more likely to transpose to familiar Open or Closed Sicilian lines.
But the stand-out performances, at least to me, are the high plus-scores of 2.f4 and especially 2.c3.
(To be continued)

Saturday 22 August 2020

Best By Test (part two)

WE saw in table three (part one of Best By Test) how 1.e4 performs at a very creditable +20 rating points, easily beating third-placed 1.c4 by six points, but being an equal number of points behind 1.d4.
Here I want to look at why this is so.
Black has 20 legal replies to 1.e4, and 18 of those appear in the 238,275 games from 2019 in ChessBase's 2020 Mega database.
The two missing replies are 1...b5, which drops a pawn to 2.Bxb5, and ... can you guess?
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
*****
The other missing reply is 1...h5, although some other replies appear only a handful of times, including 1...Nh6 (once), 1...f6 (five times) and 1...g5 (eight times).
As before I will only deal here with moves that occur at least 0.1% of the time.

Table Four: Popularity Of Black Replies To 1.e4
Move.....Number Of Times Played
1...c5...................45,756
1...e5...................29,593
1...e6...................14,187
1...c6...................10,096
1...d6....................4,782
1...d5....................3,751
1...g6....................3,187
1...Nf6..................1,784
1...Nc6.....................640
1...b6........................463

No surprise to see 1...c5 topping the list - it is so dominant it outnumbers the next two most-popular moves, 1...e5 and 1...e6, combined.
I was, however, mildly surprised to see 1...d6 ahead of 1...d5 and to see how unpopular 1...Nf6 is.
Table five shows how well Black does percentage-wise, and there is a surprise winner - at least I doubt if many people will have guessed it - in the shape of 1...g6.

Table Five: Black Replies By Percentage Score (ties broken by frequency of appearance)
Move........Score
1...g6.........51%
1...c5.........50%
1...Nf6.......49%
1...c6..........48%
1...Nc6.......48%
1...e6..........47%
1...d6..........47%
1...d5..........45%
1...e5..........44%
1...b6..........42%

I also doubt if all that many people will be surprised to find 1...c5 doing so well, but the poor score of 1...e5 is, to me certainly, a shock.
But, as I warned in part one, it can be misleading to look at percentages in isolation.

Table Six: Black Replies By Rating Performance (ties broken by frequency of appearance)
Move.........Rating Performance
1...e6.....................-10
1...d5.....................-18
1...c5.....................-19
1...c6.....................-19
1...e5.....................-28
1...d6.....................-41
1...Nf6...................-44
1...g6.....................-57
1...b6.....................-70
1...Nc6..................-71

Table six shows 1...g6, the winner when it comes to percentage scores, plummets when moves are ranked by rating performance.
Two other front-runners percentage-wise, 1...Nf6 and 1...Nc6, also do badly rating-wise.
But 1...c5, which came second in percentage score, also does reasonably well rating-wise, although by no means as well as I thought it would.
Some French Defence devotees may have expected to see the success of 1...e6, although I suspect even they will be surprised 1...e6 comes close, statistically, to giving equality.
(To be continued)

Friday 21 August 2020

Best By Test

ONE of world champion Bobby Fischer's most famous quotes is that 1.e4 is "best by test."
His view would likely have been supported by the great majority of 19th-century players.
But there have been times when 1.e4 has been out of favour, especially among the world's elite.
It was interesting to note in my Champion Repertoire series how the different world champions opened their games with the white pieces.
Fischer made his remark long before the advent of chess databases and engines.
It is pointless asking engines which is the best first move - their opening evaluations are very unsophisticated and there are just too many variables for them to come up with a useful verdict.
Databases, however, can give answers, at least from the viewpoints of popularity and statistical success.
So with the help of ChessBase's 2020 Mega database it is possible to discover what in modern chess is "best by test."
There are 238,275 games from 2019, which is the most-recent year available in Mega20.
In those games White scored 94,216 wins, 66,220 draws and 77,824 losses (+94,216=66,220-77,824). That is an overall score of 53%.
To make sure 2019 was not an exceptional year, I did the same search for the two years 2018-19 and the five years 2015-19. In both cases White scored 53%.
So it can be said with a reasonable degree of confidence that the 238,275 games from 2019 in Mega20 are a fair reflection of modern chess, at least in those events important enough to have their scores collected.
In those games the white players had an average rating of 2046 and scored a performance rating of 2057, ie +11.
Chess success can be measured statistically by percentage score and rating performance.
It is important to use both measures, even over a large sample such as 238,275, as percentage scores alone may be skewed by strong players being more likely on the white or black side of a particular opening variation.
There are 20 ways in which White can open a game (16 pawn moves and four knight moves), and all of them were played in 2019!
However some were played so infrequently (1.f3 occurs four times, 1.Na3 twice) that they can be excluded as statistically insignificant.
I decided that for a move to appear in table one below, it had to occur in at least 0.1% of games.

Table One: Openings By Popularity
Move.......Number of times played
1.e4......................114,450
1.d4........................79,279
1.Nf3.......................21,713
1.c4.........................16,456
1.b3..........................1,513
1.g3..........................1,323
1.f4...........................1,222
1.b4.............................425
1.Nc3..........................373

As can be seen, 1.e4 is by far the most-popular first move - it has many more appearances than the next two most-popular moves (1.d4 and 1.Nf3) combined.
Slightly surprising, at least to me, is how 1.Nf3 easily beats 1.c4 for third place. I suspect this is partly because 1.Nf3 is often used as a wait-and-see transpositional device that is considered more flexible than 1.c4.
The picture in table two, which shows White's opening moves by their percentage success, is very different.
Two fairly non-forcing kingside moves, 1.Nf3 and 1.g3, lead the way, with 1.e4 not doing all that much better, statistically speaking, than equality.
But at least 1.e4 does not give Black an edge, as 1.b4 appears to do with its score of 46%.

Table Two: Openings By Percentage Score (ties broken by frequency of appearance)
Move.........Score
1.Nf3..........57%
1.g3............57%
1.Nc3.........56%
1.c4............55%
1.d4............54%
1.e4............52%
1.b3............52%
1.f4.............50%
1.b4............46%

But, as I warned, it is dangerous to read percentage scores in isolation.
A very different picture emerges in table three, which lists White's opening moves according to how the players performed rating-wise.

Table Three: Openings By Rating Performance
Move..........Rating Performance
1.d4.....................+26
1.e4.....................+20
1.c4.....................+14
1.Nf3...................+11
1.b4.....................+6
1.f4......................+4
1.b3.....................+2
1.Nc3...................-17
1.g3.....................-23

White's most-popular moves, 1.e4 and 1.d4, lead the way, although "best by test," contra Fischer, is 1.d4.
Note that with 1.c4 claiming third spot, it could be said that it is the most-active moves, in the sense of moves that restrict Black's replies, which come out on top.
And look how badly 1.Nc3 and 1.g3 do - a complete reversal of looking at those moves by percentage score alone.
As an early conclusion we can say 1.e4 performs very well, but why does it lag six percentage points behind 1.d4?
(To be continued)

Thursday 20 August 2020

Non-Champion Repertoire

TO see ourselves as others see us is a skill not acquired by many, as Robert Burns observed.
But it is a skill available to many chess players, at least from a purely chess viewpoint.
All that is needed is a popular database - ChessBase's 2020 Mega database is the obvious choice - with a sufficient number of our own games.
I have 923 games in Mega20 which, considering it is easily twice more than Paul Morphy's total in the same database, is a good sample size of what I play.
Preparing to, as it were, play against oneself is a good way to see our chess selves as others see us.
Here is a repertoire based on my games in Mega20, using the same formula I employed for world champions in my Champion Repertoire series.
White
Open 1.e4, aiming to play the Spanish Exchange: 1...e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.Nc3!? Meet 5...f6 with 6.d4 exd4 7.Qxd4 Qxd4 8.Nxd4, continuing against 8...Bd7 and 8...Bd6 with 9.Be3 and continuing against 8...c5 with 9.Nb3. Meet 5...Bg4 with 6.h3 Bxf3 7.Qxf3 Qf6 (the only move faced by Spanton in Mega20) 8.Qg3. Meet 5...Bd6 with 6.d4 exd4 7.Qxd4 and meet 5...Qd6 with 6.d4 exd4 7.Qxd4.
Against the Schliemann: 3...f5 play 4.d4!? fxe4 5.Bxc6 dxc6 6.Nxe5.
Against the Smyslov: 3...g6 play 4.d4 exd4 5.Bg5, meeting 5...Be7 with 6.Bxe7 and meeting 5...f6 with 6.Bh4.
Against the Classical: 3...Bc5 play 4.c3 Nf6 5.d3, meeting 5...d6 and 5...Qe7 with 6.Be3.
Against the Berlin: 3...Nf6 transpose into the Spanish Four Knights with 4.Nc3, meeting 4...Bb4 with 5.0-0 0-0 6.d3 d6 7.Bg5, meeting 4...Bc5 with 5.Nxe5!? Nxe5 6.d4 and meeting 4...d6 with 5.d4.
Against the Bird: 3...Nd4 play 4.Nxd4 exd4 5.0-0 bc5 6.d3.
Against the Petrov: 2...Nf6 play 3.Nxe5 d6 4.Nf3 Nxe4 5.Qe2!? Qe7 6.d3.
Against the Latvian Gambit: 2...f5 play 3.Nxe5 Qf6 4.d4.
Against the Philidor: 2...d6 play 3.d4.
Against the Sicilian: 1...c5 play 2.Nf3 and aim for Bb5(+) lines.
After 2...d6 3.Bb5+ meet 3...Bd7 with 4.Bxd7+, continuing against 4...Qxd7 with 5.0-0 Nc6 6.Qe2!? and continuing against 4...Nxd7 with 5.c4. Meet 3...Nc6 with 4.d4 cxd4 5.Qxd4 Bd7 6.Bxc6, continuing against 6...Bxc6 with 7.Nc3 and continuing against 6...bxc6 with 7.c4. Meet 3...Nd7 with 4.d4 Ngf6 5.Nc3 cxd4 6.Qxd4.
After 2...Nc6 3.Bb5 meet 3...g6 with 4.Bxc6 dxc6 5.d3 Bg7 6.h3, continuing against 6...Nf6, 6...e5 and 6...Nh6 with 7.Nc3. Meet 3...e6 with 4.Bxc6 bxc6 5.d3, continuing against 5...d6 with 6.Nc3 and continuing against 5...Ne7 with 6.Qe2. Meet 3...e5 with 4.Bxc6 dxc6 5.0-0, while 3...d6 transposes to a line already covered.
After 2...e6 transpose to a King's Indian Attack with 3.d3, meeting 3...Nc6 with 4.c3!?, meeting 3...d5 with 4.Qe2!?, meeting 3...g6 with 4.c3 and meeting 3...b6 with 4.g3.
After 2...g6 play 3.d4 cxd4 4.Nxd4 Nc6 5.Nc3 Bg7 6.Nxc6!? bxc6 7.Bc4.
After 2...Nf6 play 3.e5 Nd5 4.d4!? cxd4 5.c4.
Against the French: 1...e6 play for a King's Indian Attack with 2.d3 d5 (2...c5 3.Nf3 transposes to Sicilian lines with 2...e6) 3.Nd2, meeting 3...Nf6 and 3...c5 with 4.Ngf3, usually followed by 5.g3.
Against the Caro-Kann: 1...c6 2.d4 d5 play the Panov-Botvinnik Attack: 3.exd5 cxd5 4.c4 Nf6 5.Nc3, meeting 5...e6 and 5...Nc6 with 6.Nc3 and meeting 5...g6 and 5...Bf5 with 6.Qb3.
Against the Pirc: 1...d6 2.d4 Nf6 3.Nc3 g6 and the Modern: 2...g6 3.Nc3 Bg7 play 4.Nf3. Meet the Pirc set-up (...d6, ...g6, ...Bg7, ...Nf6) with 5.Be2 0-0 6.0-0 Nc6 7.Bg5 and meet the Modern set-up (...d6, ...g6, ...Bg7, ...a6) with 5.a4, continuing against 5...b6 with 6.Bc4 and continuing against 5...Nc6 with 6.a5!? (apparently a Spanton novelty).
Against the Alekhine: 1...Nf6 play 2.Nc3!?, meeting 2...d5 with 3.exd5 Nxd5 4.Bc4 and meeting 2...e5 with 3.Nf3 and a likely transposition to the Spanish Four Knights or Philidor.
Against the Scandinavian: 1...d5 2.exd5 Nf6 play 3.Nc3!? when 3...Nxd5 transposes to the Alekhine line 2.Nc3!? d5. Against 2...Qxd5 3.Nc3 Qa5 (no games with 3...Qd6 or 3...Qd8) play 4.d4, meeting 4...Bf5 with 5.Nf3 and meeting 4...e5 (no other moves played against Spanton in Mega20) with 5.dxe5.
Against 1...Nc6 play 2.Nf3, meeting the independent 2...d6 with 3.d4.
Black v 1.e4
Play 1...e5, planning to meet the Spanish: 2.Nf3 Nc6 3.Bb5 with the Berlin: 3...Nf6.
In the mainline Berlin Wall: 4.0-0 Nxe4 5.d4 Nd6 6.Bxc6 dxc6 7.dxe5 Nf5 8.Qxd8+ Kxd8 meet 9.Nc3 with 9...Ke8, continuing against 10.b3 with 10...a5, continuing against 10.h3 with 10...Be6 and continuing against 10.a3 with 10...Be7. Meet 9.Bg5+ and 9.Bf4 with 9...Ke8.
Against 5.Re1 play 5...Nd6, meeting 6.Nxe5 with 6...Be7 and meeting 6.Bxc6 dxc6 7.Nxe5 with 7...Be7.
Against 5.Qe2 play 5...Nd6 6.Bxc6 dxc6 7.Qxe5+ Qe7.
Against 4.d3 play 4...Bc5, meeting 5.0-0 with 5...Nd4 and meeting 5.c3 with 5...0-0.
Against a transposition to the Spanish Four Knights with 4.Nc3 play 4...Bb4.
Against the Scotch: 3.d4 exd4 4.Nxd4 play 4...Bc5, meeting 5.Nxc6 and 5.Be3 with 5...Qf6 and meeting 5.Nb3 Bb6 6.Nc3 (only move faced by Spanton in Mega20) with 6...Nf6, meeting 7.Bg5 with 7...h6 and meeting 7.Bd3 with 7...0-0.
Against the Scotch Gambit: 4.Bc4 play 4...Bc5, meeting 5.c3 with a transposition to the Giuoco Piano with 5...Nf6.
Meet the Italian Game: 3.Bc4 with 3...Bc5, when 4.c3 Nf6 5.d4 exd4 is the old mainline Giuoco Piano. Continue against 6.cxd4 Bb4+ 7.Bd2 with 7...Bxd2+ 8.Nbxd2 d5. If 7.Kf1 (no games with 7.Nc3) play 7...d5. Continue against 6.e5 with 6...d5 7.Bb5 Ne4 8.cxd4 Bb4+ 9.Nbd2 (no  games with 9.Bd2) Bd7!?
In the modern Italian Game with 4.d3 Nf6 5.c3 (White's fourth and fifth moves are often transposed) play 5...a6, meeting 6.0-0 and 6.b4 with 6...Ba7 but meeting 6.Bb3 with 6...0-0.
Meet the Evans Gambit: 4.b4 with 4...d5!?
The Spanish Four Knights: 3.Nc3 Nf6 4.Bb5 has already been covered as a transposition from the Spanish Berlin. Meet the Scotch Four Knights: 4.d4 exd4 5.Nxd4 with 5...Bb4.
Meet the Ponziani: 3.c3 with 3...d5 4.Qa4 (only move faced by Spanton in Mega20) f6.
Meet the King's Gambit: 2.f4 with 2...Bc5 3.Nf3 d5!?, continuing against 4.exd5 with 4...e4 and against 4.Nxe5 with 4...dxe4!?
Meet the Vienna: 2.Nc3 with 2...Nf6, continuing against 3.Bc4 with 3...Nxe4, continuing against 3.g3 with 3...d5 and continuing against 3.f4 with 3...d5 4.fxe5 Nxe4, and if 5.Nf3 play 5...Be7, if 5.d3 play 5...Nxc3 6.bxc3 d4 and if 5.Qf3 play 5...Nxc3 6.bxc3 Be7.
Meet the Bishop's Opening: 2.Bc4 with 2...Bc5.
Against 1.d4 and Others
Against 1.d4 play 1...d5 with the aim of heading down the main line of the QGD Exchange: 2.c4 e6 3.Nc3 Nf6 4.cxd5 exd5 5.Bg5 Be7 6.e3 0-0 7.Bd3 Nbd7, meeting 8.Nge2, 8.Nf3 and 8.Qc2 with 8...Re8.
Against 4.Nf3 play 4...c5 5.cxd5 cxd4!?, meeting 6.Qxd4 exd5 7.Bg5 with 7...Be7. If 7.e4 play 7...dxe4 or 7...Nc6 and if 7.e3 or 7.g3 play 7...Nc6. Meet 6.Nxd4 with 6...Nxd5 and meet 6.Qa4+ with 6...Bd7 7.Qxd4 exd5.
Against 4.Bg5 play 4...c5!? 5.cxd5 cxd4, meeting 6.Qxd4 with 6...Be7 and meeting 6.Qa4+ with 6...Qd7 7.Qxd4 Nc6. If 5.e3 play 5...cxd4 6.exd4 (only move played against Spanton in Mega20) Bb4.
Against 3.Nf3 play 3...Nf6, meeting the Catalan: 4.g3 with 4...c5!?, continuing against 5.cxd5 with 5...cxd4!? 6.dxe6 Bxe6 and continuing against 5.Bg2 with 5...Nc6 6.0-0 cxd4!?
Against 2.Nf3 play 2...Nc6!?, meeting 3.Bf4, 3.e3, 3.c3, 3.g3, 3.b3, 3.c4 and 3.a3 with 3...Bg4!?
Against the New London: 2.Bf4 play 2...Nf6, meeting 3.e3 and 3.Nf3 with 3...e6 followed by ...Bd6.
Against the Trompowsky: 2.Bg5 play 2...Nf6, meeting 3.e3 (the only move against Spanton in  Mega20) with 3...Nc6!?
Against the Veresov: 2.Nc3 Nf6 3.Bg5 play 3...c6 or 3...c5.
Against the Blackmar-Diemer Gambit: 2.e4 dxe4 3.Nc3 play 3...e5!?
Against the English: 1.c4 play the Symmetrical: 1...c5, meeting 2.Nf3 Nc6 3.Nc3 (White's second and third moves are often transposed) with 3...g6. Against an early g3 imitate White's moves, eg 2.g3 g6 3.Bg2 Bg7 4.Nc3 Nc6 5.Nf3 Nf6 6.0-0 0-0. If 7.d4 play 7...cxd4 8.Nxd4 Nxd4 9.Qxd4 d6. If 6.d4 play 6...cxd4 7.Nxd4 0-0 with a likely transposition. If 5.e3 play 5...d6 6.Nge2 e5 7.0-0 Nge7. If 5.a3 play 5...d6 6.Rb1 a5.
Against 1.Nf3 play 1...Nc6!?, meeting 2.c4, 2.g3 and 2.e4 with 2...e5. If White stops ...e5 with 2.d4, play 2...d5, transposing into lines already covered.
Against the Bird: 1.f4 play the From Gambit: 1...e5.
Against 1.g3 play 1...c5 2.Bg2 Nc6 with a likely transposition into an English or a Closed Sicilian.
Against 1.b3 play 1...d5 2.Bb2 Nf6, meeting 3.e3 and 3.Nf3 with 3...Bf5.
Against 1.b4 play 1...e5.

Here Spanton uses his unusual line against the King's Gambit to score a quick win.
Richard S Jones (2075) - Spanton (2065)
Newport (Wales) 1997
King's Gambit Declined
1.e4 e5 2.f4 Bc5 3.Nf3 d5!?
This counter-gambit initiates sharp play and can be thought of as a delayed Falkbeer.
4.exd5
The most-popular choice in Mega20 but 5.Nxe5 is a major alternative.
4...e4 5.Ne5 Nf6 6.Nc3!?
Again the most-popular choice in Mega20, but my main analysis engines Stockfish11 and Komodo11.01 prefer 6.d3 or 6.d4.
6...0-0 7.d4
7.Bc4!? to hold on to the d5 pawn does not appear in Mega20. The engines reckon Black has an edge after 7...Re8 8.Na4 Bd6.
7...exd4!?
Komodo11.01 likes this but Stockfish11 prefers 7...Bb4, which was played by a 2300 in the only previous game to reach this position in Mega20.
8.Bxd3
Subsequent games mostly saw 8.Qxd3!?, when Black also seems to have good compensation for the pawn-minus.
8...Re8 9.Ne4?
White indirectly protects d5 by attacking the black dark-square bishop, one point being that 9...Nxe4 10.Bxe4 gives d5 two defenders. There is a fatal flaw, but even so the best the engines come up with is 9.Kf1, when Black's compensation is obvious.
Black to play and win
*****
*****
*****
*****
*****
9...Nxe4 10.Bxe4 Rxe5! 11.fxe5 Qh4+ 12.Kd2 Qxe4 13.Re1?!
This accelerates the end but White is busted anyway.
13...Qd4+ 14.Ke2 Bg4+ 0-1

Wednesday 19 August 2020

Champion Repertoire (part 16)

MAGNUS Carlsen has been world champion since defeating Vishy Anand in 2013.
Carlsen is well-known for varying his openings, but his games in ChessBase's 2020 Mega database reveal distinct patterns.
White
Open 1.e4, intending to play a mainline Spanish: 1...e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Be7 6.Re1 Be7 7.Bb3, and if 7...0-0 avoid the Marshall Attack by playing 8.a4, meeting 8...b4 with 9.d3 d6 10.Nbd2 and meeting 8...Bb7 with 9.d3 d6 followed by either 10.Nbd2 or 10.Bd2.
Against 7...d6 play 8.c3 0-0 9.h3, meeting 9...Na5 with 10.Bc2 c5 11.d4 Qc7 12.Nbd2, meeting 9...Nb8 with 10.d4 Nbd7 11.Nbd2 bb7 12.Bc2 Re8 13.Nf1 Bf8 14.Ng3 g6 followed by either 15.a4 or 15.b3, meeting 9...Bb7 with 10.d4 Re8 11.Nbd2 Bf8 12.d5, and meeting 9...Re8 with 10.d4 Bb7 11.Nbd2, which is a transposition to the line with 9...Bb7.
Against the Open Variation: 5...Nxe4 play 6.d4 b5 7.Bb3 d5 8.dxe5 Be6 9.Nbd2, meeting 9...Nc5 with 10.c3, continuing against 10...d4 with 11.Bxe6 Nxe6 12.cxd4 Ncxd4 13.a4 and continuing against 10...Be7 with 11.Bc2, meeting 9...Be7 with 10.c3 0-0 followed by either 11.Bc2 or 11.Qe2, and meeting 9...Bc5 with the somewhat unusual 10.c3!? 0-0 11.Bc2.
Against 5...b5 6.Bb3 Bc5 play 7.c3 d6 8.d3!? If 6...Bb7 play 7.d3.
Against 5...Bc5 play 6.c3, meeting 6...0-0 with 7.d3!? and meeting 6...b5 with 7.Bb3 Nxe4 (7...d6 transposes to 5...Bc5 6.Bb3 Bc5) 8.Qe2!? d5 9.d3.
Against the Steinitz Deferred: 4...d6 play 5.c3 Bd7 (no games with 5...f5) 6.d4.
Against the Berlin: 3...Nf6 play 4.d3, meeting 4...Bc5 with 5.Bxc6 dxc6 6.h3!? and meeting 4...d6 with 5.0-0, continuing against 5...Be7 with 6.c3 0-0 7.Re1 and continuing against 5...g6 with 6.d4 exd4 (no games with 6...Bd7) 7.e5 dxe5 8.Nxe5.
Against the Schliemann: 3...f5 play 4.d3 fxe4 (no games with 4...Nf6) 5.dxe4 Nf6 6.0-0, meeting 6...Bc5 with 7.Bxc6, 7.Qe2!? or 7.Qd3 and meeting 6...d6 with 7.Nc3!?
Against the Petrov: 2...Nf6 play 3.Nxe5 d6 4.Nf3 Nxe4 5.d4 d5 6.Bd3, meeting 6....Bd6 with 7.0-0 0-0 8.c4, meeting 6...Nc6 with 7.0-0 Be7 8.c4 Nb4 9.Be2 0-0 10.Nc3 and meeting 6...Be7 with 7.0-0 0-0 (7...Nc6 is a transposition to 6...Nc6) 8.c4 Nf6 (no games with 8...c6) 9.Nc3.
Against the Philidor: 2...d6 play 3.d4 exd4 4.Nxd4 Nf6 5.Nc3 Be7 6.g3!?
Against the Sicilian: 1...c5 play 2.Nf3, 3.d4 and (after 3...cxd4) 4.Nxd4.
Against the Najdorf: 2...d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 play 6.Be3 e5 7.Nf3!?, continuing against 7...Be7 with 8.Bc4 0-0 9.0-0 Be6 10.Bb3 and continuing against 7...Qc7 with 8.a4. If 6...Ng4 play 7.Bg5 h6 8.Bh4 g5 9.Bg3 Bg7 10.Qd2!? If 6...e6 play 7.f3 b5 8.Qd2, meeting 8...Nbd7 with 9.g4 and meeting 8...b4 with 9.Na4.
Against the Dragon: 5...d6 play 6.Be3 Bg7 7.f3, meeting 7....0-0 with 8.Qd2 Nc6 9.0-0-0, continuing against 9...d5 with 10.exd5 Nxd5 11.Nxc6 bxc6 12.Bd4, continuing against 9...Bd7 with 10.g4 Rc8 (or 10...Qa5 11.Kb1 Rfc8 12.h4) 11.h4 and continuing against 9...Nxd4 with 10.Bxd4 Be6 11.g4!? Qa5 12.Kb1 Rfc8 13.a3. Meet 7...Nc6 with 8.Qd2 Bd7 (8...0-0 transposes to 7...0-0) 9.0-0-0, when 9...0-0 (the only move Carlsen has faced in Mega20) again transposes to 7...0-0. Meet 7...Nbd7 with 8.Qd2 h5 (the only move against Carlsen in Mega20) 9.Bc4.
Against 5...Nc6 play 6.Bg5, meeting 6...e6 with 7.Qd2, continuing against 7...a6 with 8.0-0-0 Bd7 f4 or 8...h6 9.Nxc6!? bxc6 10.Bf4 d5 11.Qe3 and continuing against 7...Be7 with 8.0-0-0 0-0 9.f4 Nxd4 10.Qxd4 Qa5 11.Bc4 Bd7 12.Kb1!? or 8...Nxd4 9.Qxd4 a6 10.f4 b5 11.Bxf6 gxf6 12.Be2!? Meet 6...Bd7 with 7.Qd2 or 7.Be2 (just one game of Carlsen's with each move in Mega20). Meet 6...g6 with 7.Bxf6 exf6 8.Bc4 Bg7 9.Ndb5 0-0 10.Qxd6 f5 11.0-0-0. Meet 6...e5 with 7.Bxf6 gxf6 8.Nf5.
Against the Scheveningen: 5...e6 play 6.g4!? h6 (only move played against Carlsen in Mega20) 7.Bg2!? Nc6 8.h3.
Against 2...e6 3.d4 cxd4 4.Nxd4 Nc6 play 5.Nc3, meeting 5...Qc7 with 6.Be3, continuing against 6...a6 with 7.Bd3 Nf6 8.0-0 and continuing against 6...Nf6 with 7.Bd3, when Carlsen has only faced the transpositional 7...a6. Meet 5...a6 with 6.Be3, continuing against 6...Nf6 with 7.Bd3, when Carlsen has only faced the transpositional 7...Qc7, and continuing against 6...Nge7 with 7.Nb3 Ng6 8.Be2!? Meet 5...d6 with 6.Be3 Nf6 7.f4!? Be7 8.Qf3. Meet 5...Nf6 with 6.Ndb5, continuing against 6...Bb4 with 7.a3 Bxc3+ 8.Nxc3 d5 9.Bd3!? and continuing against 6...Bc5 (no games with 6...d6) with 7.Bf4!?
Against 2...e6 3.d4 cxd4 4.Nxd4 a6 play 5.Nc3 with likely transpositions. Meet the independent 5...b5 with 6.Bd3 Qb6 (no games with 6...Bb7) 7.Nf3!?
Against 2...e6 3.d4 cxd4 4.Nxd4 Nf6 play 5.Nc3, again with likely transpositions. Meet the independent 5...Qb6 with 6.Nb3 Bb4 7.Bd3.
Against 2...Nc6 3.d4 cxd4 4.Nxd4 Nf6 play 5.Nc3, meeting 5...d6 with 6.Bg5, which is a transposition to lines already covered. Meet 5...e5 with 6.Ndb5 d6 7.Bg5 a6 8.Na3 b5 9.Bxf6  gxf6 10.Nd5, continuing against 10...f5 with 11.Bd3 Be6 12.0-0 and continuing against 10...Bg7 with 11.Bd3 Ne7 12.Nxe7 Qxe7 11.0-0.
Against 2...Nc6 3.d4 cxd4 4.Nxd4 e5 play 5.Nb5, meeting 5...d6 with 6.N1c3 a6 7.Na3 b5 8.Nd5, continuing against 8...Nge7 with 9.c4 and continuing against 8...Nf6 with 9.Nxf6+!? Qxf6 10.c4.
Against 2...Nc6 3.d4 cxd4 4.Nxd4 g6 play 5.c4, meeting 5...Bg7 with 6.Be3 Nf6 7.Nc3 0-0 8.Be2 d6 (Black's last two moves are often transposed) 9.0-0 Bd7 10.Qd2 Nxd4 11.Bxd4 (Carlsen also reaches the same position via the move-order 9...Nxd4 10.Bxd4 Bd7 11.Qd2) Bc6 12.f3, continuing against 12...a5 with 13.b3 Nd7 14.Be3 (also reached vis 12...Nd7 13.Be3 a5 14.b3) Nc5 15.Rab1 Qb6 16.Rfc1. Meet 5...Nf6 with 6.Nc3 Nxd4 7.Qxd4 d6 8.f3!? Bg7 9.Be3 0-0 10.Qd2.
Against 2...g6 3.d4 cxd4 4.Nxd4 Bg7 (4...Nc6 transposes to lines already covered) Carlsen has almost equally played 5.Nc3 and the repertoire-friendly 5.c4. He has also reached the position after 5.Nc3 via 2...g6 3.d4 Bg7 4.Nc3 cxd4 5.Nxd4, the games continuing 5...Nc6 6.Be3 Nf6 7.Bc4, after which meet 7...0-0 with 8.Bb3 a5 9.0-0 a4 10.Nxa4 Nxe4 11.Nb5. If 8...Qa5 play 9.0-0 d6 10.h3 Bd7 11.Re1 and if 8...e6 (the only other move Carlsen has faced in Mega20) play 9.f4!? d6 10.Qf3. Meet 7...Qa5 with 8.0-0 0-0 9.Bb3 (a transposition to the line with 7...0-0 and 8...Qa5).
Against 2...Nf6 play 3.e5 Nd5 4.Nc3, meeting 4...Nxc3 (the only move Carlsen's faced) with 5.dxc3 Nc6 6.Bf4.
Against 2...a6 play 3.c3, meeting 3...d5 with 4.exd5 Qxd5 5.d4 and meeting 3...e6 with 4.d4 d5 5.e5.
Against the French: 1...e6 2.d4 d5 play 3.Nc3, meeting the Classical: 3...Nf6 with 4.e5 Nfd7 5.f4 c5 6.Nf3 Nc6 7.Be3, continuing against 7...a6 with 8.Qd2 b5 9.a3!?, continuing against 7...Be7 with 8.Qd2 0-0 9.g3!?, continuing against 7...cxd4 with 8.Nxd4 Bc5 9.Qd2 and continuing against 7...Rb8 (the only other move played against Carlsen in Mega20) with 8.a3!? and if 8...c4 (the only move faced by Carlsen) then 9.g4.
Meet the Winawer: 3...Bb4 with 4.e5 c5 5.a3 Bxc3+ (no Carlsen games with 5...Ba4) 6.bxc3, meeting 6...Ne7 with 7.Qg4, continuing against 7...Nf5 with 8.Bd3 h5 9.Qf4, continuing against 7...Qc7 with 8.Bd3!? cxd4 9.Ne2 dxc3 10.Qxg7 Rg8 11.Qxh7 and continuing against 7...0-0 with 8.Bd3 Nbc6 (no other moves played against Carlsen in Mega20) 9.Qh5 Ng6 10.Nf3 Qc7 11.Be3 c4 12.Bxg6 fxg6 13.Qg4. Meet 6...Qa5 (the only move apart from 6...Ne7 faced by Carlsen in Mega20) with 7.Bd2 Qa4 8.Qg4. Against 4...b6 play 5.a3 Bf8 6.Nf3.
Meet the Rubinstein: 3...dxe4 with 4.Nxe4 Nd7 (only move faced by Carlsen in Mega20) 5.Nf3, continuing against 5...Ngf6 with 6.Bg5 Be7 7.Nxf6+ Nxf6 (no games with 7...Bxf6) 8.Bd3 and continuing against 5...Be7 with 6.Bc4!? Ngf6 7.Nxf6+ Nxf6 (no games with 7...Bxf6) 8.0-0 0-0 9.Ne5!?
Against the Caro-Kann: 1...c6 2.d4 d5 play 3.e5, meeting 3...Bf5 with 4.Be3!? e6 5.Nd2 and meeting 3...c5 with 4.Nf3!? Nc6 5.c4.
Against the Pirc: 1...d6 2.d4 Nf6 3.Nc3 g6 Carlsen has varied, playing 4.f4, 4.Nf3 and 4.Be3 twice each and playing 4.Bg5 once. Perhaps the easiest repertoire fit is 4.Nf3 as Carlsen has also reached this position via various Modern move-orders, eg 2...g6 3.Nf3 Nf6 4.Nc3. After 4...Bg7 play 5.Be2 0-0 6.0-0, meeting 6...Bg4 with 7.Be3 Nc6 8.Qc1!?, meeting 6...c6 with 7.h3 Nbd7 8.e5, meeting 6...Nc6 with 7.Be3!? and meeting 6...a6 with 7.h3!?
Against the Alekhine: 1...Nf6 play 2.e5 Nd5 3.d4 d6 4.Nf3, meeting 4...Bg4 with 5.Be2, meeting 4...dxe5 with 5.Nxe5 c6 6.Be2 and meeting 4...g6 with 5.c4!? Nb6 6.exd6 cxd6 7.Be2 Bg7 8.0-0.
Against the Scandinavian: 1...d5 2.exd5 Qxd5 play 3.Nc3, meeting 3...Qd6 with 4.d4 Nf6 5.Nf3 g6 (the only 3...Qd6 line faced by Carlsen) 6.Nb5, meeting 3...Qa5 with 4.d4 Nf6 5.Bd2 and meeting 3...Qd8 with 4.d4 Nf6 5.Nf3 Bg4 6.h3 Bxf3 7.Qxf3 c6 8.Qd3!? Against 2...Nf6 play 3.d4 or 3.Nf3.
Black v 1.e4
Play 1...e5 aiming to meet the mainline Spanish: 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Be7 6.Re1 b5 7.Bb3 0-0 8.c3 d6 9.h3 with 9...Nb8 10.d4 (the only move played against Carlsen in Mega20) Nbd7, continuing against 11.Nbd2 with 11...Bb7 12.Bc2 Re8, meeting 13.a4 with 13...Bf8 14.Bd3 c6 and meeting 13.Nf1 with 13...Bf8 14.Ng3 g6. If 11.c4 play 11...c6 12.Nc3 b4 13.Na4 c5 14.d5 Re8.
Against 9.d4 play 9...Bg4, meeting 10.Be3 with 10...exd4 11.cxd4 Na5 12.Bc2 c5 and meeting 10.d5 with 10...Na5 11.Bc2 c6 12.h3 Bc8.
Against the Anti-Marshall: 8.a4 play 8...b4!?, against 8.h3 play 8...Bb7 9.d3 d6 and against 8.d4 play 8...d6 9.c3 Bg4.
Against 5.Qe2 play 5...Be7 6.c3 d6!? 7.0-0 0-0.
Against the Exchange Variation: 4.Bxc6 dxc6 5.0-0 (the only move faced by Carlsen in Mega20) play 5...Bd6 6.d4 exd4 7.Qxd4 f6 and against 6.d3 play 6...Ne7!?
Against the Italian Game: 3.Bc4 play 3...Bc5 4.c3 Nf6 5.d3 a6, meeting 6.0-0 with 6...d6, continuing against 7.a4, 7.Re1, 7.Bb3, 7.h3 and 7.Nbd2 with 7...Ba7, meeting 6.Bb3 with 6...0-0!? 7.0-0 Ba7 and meeting 6.Nbd2 with 6...Ba7. Against 5.d4 play 5...exd4 6.e5 (the only move played against Carlsen in Mega20) d5, meeting 7.Bb5 with 7...Ne4 8.cxd4 Bb6 and meeting 7.Be2 with 7....Ne4 8.cxd4 Bb6.
If 4.0-0 or 4.d3 play 4...Nf6 with likely transpositions, and if 4.Nc3 play 4...Nf6 5.d3 d6.
Against the Evans Gambit: 4.b4 play 4...Bb6 5.a4 a6.
Against the Scotch: 3.d4 exd4 4.Nxd4 play 4...Bc5, meeting 5.Nxc6 and 5.Be3 with 5...Qf6 and meeting 5.Nb3 with 5...Bb6, and against the Scotch Gambit: 4.Bc4 play 4...Nf6.
Against the Scotch Four Knights: 3.Nc3 Nc6 4.d4 exd4 5.Nxd4 play 5...Bb4 and against the Spanish Four Knights: 4.Bb5 play 4...Bb4.
Against the Ponziani: 3.c3 play 3...Nf6, against the Vienna: 2.Nc3 play 2...Nc6, against the Bishop's Opening: 2.Bc4 play 2...Nf6 and against the King's Gambit: 2.f4 play 2...exf4 3.Nf3 (only move faced by Carlsen in Mega20) g5.
Black against 1.d4 and Others
Against 1.d4 aim for a Nimzo-Indian: 1...Nf6 2.c4 e6 3.Nc3 Bb4, meeting 4.Qc2 with 4...d5, meeting 4.e3 with 4...0-0, meeting 4.Nf3 with 4...d5, meeting 4.f3 with 4...d5, meeting 4.Bg5 with 4...b6 and meeting 4.a3 Bxc3+ 5.bxc3 with 5...Nc6.
Against 3.Nf3 transpose to a Queen's Gambit Declined with 3...d5, meeting 4.Nc3 with 4...Bb4 (Carlsen has played 4...Be7 with equal frequency but scored much better with 4...Bb4, which fits in with the Nimzo-Indian repertoire), meeting the Catalan: 4.g3 with 4...Be7 5.Bg2 0-0 6.0-0 dxc4 and meeting 4.Bg5 with 4...Be7 5.Nc3 (the only move faced by Carlsen in Mega20) h6, continuing against 6.Bh4 with 6...0-0 7.e3 b6 and continuing against 6.Bxf6 with 6...Bxf6. If then 7.e3 play 7...0-0 8.Rc1 c6, if 7.Qc2 play 7...dxc4 8.e3 c5 and if 7.cxd5 play 7....exd5 8.b4 (the only move Carlsen has faced) 0-0 8.e3 c6.
Against 3.g3 play 3...d5 with likely transpositions.
Against the London: 2.Nf3 d5 3.Bf4 play 3...c5, against the New London: 2.Bf4 play 2...e6 3.e3 b6 and against the Colle: 2.Nf3 d5 3.e3 play 3...e6, meeting 4.c4 with 4...a6!?, meeting 4.b3 with 4...b6 and meeting 4.Bd3 with 4...Bd6.
Against the Trompwosky: 2.Bg5 play2...d5, meeting 3.Bxf6 with 3...exf6 and meeting 3.e3 with 3...c5.
Against the Veresov: 2.Nc3 d5 3.Bg5 and the Jobava-Prié: 3.Bf4 play 3...g6.
Against the English: 1.c4 play 1...Nf6, meeting 2.Nc3 with 2...e5 3.Nf3 Nc6, continuing against 4.g3 with 4...d5 5.cxd5 Nxd5 6.Bg2 Nb6, continuing against 4.e3 and 4.e4 with 4...Bb4 and continuing against 4.d3 with 4...d5. Meet 3.g3 (no other moves have been faced by Carlsen in Mega20) with 3...Bb4 4.Bg2 0-0. Against 2.g3 (no games with 2.Nf3) play 2...e6 3.Bg2 d5 or 2...c6 3.Bg2 d5.
Against 1.Nf3 play 1....Nf6, looking for transpositions. Meet the independent 2.g3 with 2...g6 3.Bg2 Bg7, continuing against 4.0-0 0-0 5.d4 and 4.c4 0-0 5.0-0 with 5...d6 but continuing against 4.c4 0-0 5.Nc3 with 5...c5. Meet 2.b3 with 2...c5 3.Bb2 Nc6.
Against 1.b3 Nf6 2.Bb2 play 2...g6.

Here Carlsen destroys a strong grandmaster's 'drawish' Berlin.
Carlsen (2837) - Sergey Karjakin (2760)
World Blitz Championship (Riyadh, Saudi Arabia) 2017
Spanish Berlin
1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.d3 Bc5 5.Bxc6
Carlsen has played six other moves here, including the somewhat bizarre 5.Ba4, but the text, which was played as long ago as 1870, is fashionable among the elite.
5...dxc6 6.Nc3
Carlsen's only outing with this move in Mega20, where he has scored a win and a loss with 6.Qe2, a win and two draws with 6.Nbd2 and four wins with 6.h3.
6...0-0 7.Be3 Bd6
My main analysis engines Stockfish11 and Komodo11.01 prefer 7...Bxe3!? but strong players have shied away from half-opening the f file for White.
8.Bg5 Re8 9.h3 c5!?
Weakening d5 will cost Black the bishop-pair.
10.Nd5 Be7 11.Nxe7+ Qxe7 12.0-0 h6 13.Be3 Nd7
The engines' 13...c4!? may be an improvement.
14.Nd2 Nb8?!
Carlsen is clearly aiming for a kingside attack, and it may be Karjakan thought his knight had time to redevelop via c6 to d4 or e5, but 14....Nf8 was more prudent.
15.f4 exf4 16.Rxf4 Nc6 17.Qh5
White develops with tempo thanks to threatening the c5 pawn.
17...b6 18.Raf1 Rf8 19.Nf3
Even stronger is 19.R4f3, threatening Bxh6 and Rg3+.
19...Be6 20.Rh4 f6 21.Qg6 Qf7?!
Defence is especially hard at blitz. The engines suggest 21...Bf7 22.Qg3 Kh7 with a small edge for White, according to Stockfish11, although Komodo11.01 rates the position as level.
22.Qg3
Position after 22.Qg3
Karjakan may have underestimated this retreat. Despite having all his kingside pawns on dark squares, Black is weak on the dark squares thanks to White having the only dark-square bishop.
22...Nb4?
The engines give 22...h5 but much prefer White.
23.Bxh6 Nxc2
There is no defence.
24.Ne5 fxe5
Tantamount to resigning, which Karjakin might well have done at a slower time control.
25.Rxf7 Rxf7 26.Qg6 Bxa2 27.Bg5 Rff8 28.Rh7 Rf7 29.Bf6 1-0

Tuesday 18 August 2020

Champion Repertoire (part 15)

VISHY Anand was world champion from 2007 until he lost to Magus Carlsen in 2013.
A repertoire based on his games in ChessBase's 2020 Mega database emphasises deep theoretical lines.
White
Open 1.e4, aiming to play a mainline Spanish: 1...e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Be7 6.Re1 b5 7.Bb3 0-0/d6 8.c3 d6/0-0 9.h3.
Meet 9...Nb8 with the mainline 10.d4 Nbd7 11.Nbd2 Bb7 12.Bc2 Re8 13.Nf1 Bf8 14.Ng3 g6 15.a4 c5 16.d5 c4 17.Bg5.
Meet 9...Na5 with the mainline 10.Bc2 c5 11.d4 Qc7 12.Nbd2 cxd4 13.cxd4 Nc6, at which point Anand has twice played the main move 13.Nb3 and twice played 13.a3!?, doing better with the latter.
Meet 9...Bb7 with the mainline 10.d4 Re8 11.Nbd2 Bf8, at which point he has most often played 12.a4 but scored much better with the main alternative 12.d5. 
Meet 9...Re8 with the mainline 10.d4 Bb7 11.Nbd2, which is a transposition to Anand's play against 9...Bb7.
Meet 9...Nd7 with the mainline 10.d4 Bf6, at which point Anand has twice played 11.a4 and twice played 11.Be3, doing slightly better with the former.
Meet 9...Be6 with the mainline 10.d4 Bxb3 11.axb3 exd4 12.cxd4 d5 13.e5 Ne4 14.Nc3 f5 15.exf6 Bxf6 16.Nxe4 dxe4 17.Rxe4 Qd5 18.Rg4.
Against the Marshall Attack: 8.c3 d5 play the mainline 9.exd5 Nxd5 10.Nxe5 Nxe5 11.Rxe5 c6 12.d4 Bd6 13.Re1 (Anand has sometimes transposed White's 12th and 13th moves) Qh4 14.g3 Qh3 15.Be3 Bg4 16.Qd3 Rae8 17.Nd2.
Against the Open Variation: 5...Nxe4 play the mainline 6.d4 b5 7.Bb3 d5 8.dxe5 Be6, continuing 9.Nbd2 Nc5 10.c3 Be7 11.Bc2. If 9...Be7 play 10.c3 0-0, at which point Anand has done best with 11.Qe2!? Nc5 12.Nd4 Nxb3 13.Nxb3 Qd7 14.Nxc6 Qxc6 15.Be3.
Against 5...b5 6.Bb3 Bc5 play 7.a4!? If 6...Be7 play 7.d4!? d6 8.c3,  and if 6...Bb7 play 7.d3!? with a quick a4.
Against 5...Bc5 play 6.c3 b5 7.Bb3 d6, at which point Anand has done much better with the slightly unusual 8.a4 than with the mainline 8.d4.
Against the Steinitz Deferred: 4...d6 play 5.c3, meeting 5...Bd7 and 5...g6 with 6.d4 and meeting 5...f5 with 6.exf5 Bxf5 7.0-0 Bd3 8.Bc2 Bxc2 9.Qxc2 Nf6 10.d4.
Against 4...b5 5.Bb3 Na5 play 6.0-0 d6 7.d4.
Against the Berlin: 3...Nf6 head for the Berlin Wall: 4.0-0 Nxe4 5.d4 Nd6 6.Bxc6 dxc6 7.dxe5 Nxf5 8.Qxd8+ Kxd8 9.Nc3, meeting 9...Ke8 10.h3 h5 with 11.Bg5 or 11.Bf4, meeting 9...Ne7 10.h3 Ng6 with 11.Bg5+, meeting 9...h6 10.h3 Bd7 with 11.b3 Kc8 12.Bb2 b6 13.Rad1, meeting 9...Be7 with 10.Rd1+ Ke8 11.Bg5 and meeting 9...Bd7 10.h3 h6 with 11.b3, which is a transposition to Anand's line against 9...h6.
Against the Schliemann: 3...f5 play 4.d3 fxe4 5.dxe4 Nf6 6.0-0 Bc5 7.Bxc6!? bxc6 8.Nxe5.
Against the Classical: 3...Bc5 Anand won all three games in which he played 4.c3 and all three in which he played 4.0-0.
Against 3...g6 Anand has done best with 4.c3, and there is no other black third move he has faced more than twice in Mega20.
Against the Petrov: 2...Nf6 play 3.Nxe5 d6 4.Nf3 Nxe4 5.d4 d5 6.Bd3, meeting 6...Nc6 7.0-0 Be7 (Black can easily reverse moves six and seven) with 8.c4 Nb4 9.Be2 0-0 10.Nc3 and meeting 6...Bd6 with 7.0-0 0-0 8.c4 c6, at which point Anand played 9.Re1!? five times and 9.Qc2 four times, but scored 28 percentage points higher with the latter.
Against the Philidor: 3...d6 play 3.d4 exd4 4.Nxd4 Nf6 5.Nc3 Be7 6.g3!?
Against the Sicilian: 1...c5 play Open lines, ie 2.Nf3, 3.d4 and (after 3...cxd4) 4.Nxd4.
Against 2...d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 play 6.Be3, meeting 6...e6 with 7.f3 b5 8.g4!?, meeting 6....e5 7.Nb3 Be6 with 8.f3 and continuing 9.Qd2 against both 8...Be7 and 8...Nbd7, and meeting 6...Ng4 with 7.Bg5 h6 8.Bg4 g5 9.Bg3 Bg7, at which point Anand varied his responses, doing well with 10.Be2, 10.h3, 10.Qd2 and 10.Bc4.
After 5...Nc6 Anand usually played 6.Bg5 e6 7.Qd2, meeting 7...a6 and 7...Be7 with 8.0-0-0, but the one time he faced 7...Qb6 he beat it with 8.Nb3.
After 5...g6 play 6.Be3 Bg7 7.f3, meeting 7...0-0 with 8.Qd2 Nc6 9.Bc4 (Black often transposes moves seven and eight) Bd7, at which point Anand played equally 10.h4 and 10.Bb3 but did much better with the latter.
After 5...e6 play 6.g4 h6 (the only move Anand faced in Mega20) 7.h4 Nc6 8.Rg1.
After 2...Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 the move 5...d6 transposes to the 2..d6 line with 5...Nc6. If 5...e5 play 6.Ndb5 d6 7.Bg5 a6 8.Na3 b5 9.Bxf6 gxf6 10.Nd5, meeting 10...f5 with 11.c3 Bg7 12.exf5 Bxf5 13.Nc2 and meeting 10...Bg7 with 11.Bd3 Ne7 12.Nxe7 Qxe7 13.0-0 0-0 14.c4. If 5...e6 play 6.Ndb5, meeting 6...d6 with 7.Bf4 e5 8.Bg5, which transposes to the line with 5...e5, and meeting 6...Bb4 with 7.a3 Bxc3+ 8.Nxc3 d5 9.exd5 exd5 10.Bd3 0-0 11.0-0 d4 12.Ne2. If 4...g6 play 5.Nc3 Bg7 6.Be3 Nf6 7.Bc4, meeting 7...0-0 with 8.Bb3 and meeting 7...Qa5 with 8.0-0 0-0 9.Nb3. If 4...Qb6 play 5.Nb3 Nf6 6.Nc3 e6 7.Bd3!?
After 2...e6 3.d4 cxd4 4.Nxd4 a6 play 5.Bd3, meeting 5...Bc5 with 6.Nb3, continuing against 6...Be7 with the unusual 7.Be3!? and continuing against 6...Ba7 with 7.Qe2 Nc6 8.Be3. Meet 5...Nf6 with 6.0-0, continuing against 6...Qc7 with 7.Qe2 d6 8.c4 and continuing against 6...d6 with 7.c4. Meet 5...g6!? with 6.0-0 Bg7, which Anand has reached three times, winning with 7.c3, 7.Be3 and 7.Nb3. Meet 5...Qb6 with 6.c3!? and meet 5...Qc7 with 6.0-0 Nf6 7.Qe2, which is a transposition into the line 5...Nf6 6.0-0 Qc7.
After 2...g6 3.d4 cxd4 4.Nxd4 Bg7 play 5.Nc3 Nc6 6.Be3, transposing into the line with 2...Nc6 and 4...g6.
After 2...Nf6 play 3.Nc3, after 2...a6 play 3.c4 and after 2...b6 play 3.Nc3 Bb7 4.d4 cxd4 5.Nxd4.
Against the French: 1...e6 2.d4 d5 play 3.Nc3, meeting the Classical: 3...Nf6 with 4.e5 Nfd7 5.f4 c5 6.Nf3 Nc6 7.Be3, continuing against 7...a6 with 8.Qd2 b5, a position Anand reached six times with the white pieces, playing four different moves, including 9.h4 twice 9.a3 twice, doing better with the latter. If 7...cxd4 play 8.Nxd4 Bc5 9.Qd2 0-0 10.0-0-0. If 7...Qb6 play 8.Na4 Qa5+ 9.c3, meeting 9...cxd4 with 10.b4 and meeting 9...c4 with 10.b4 Qc7 11.Be2 Be7 12.0-0.
Against the Winawer: 3...Bb4 play 4.e5 c5 5.a3 Bxc3+ 6.bxc3, meeting 6...Ne7, 6...Qc7 and 6...Nc6 with 7.Qg4 (Anand played Nf3 slightly more often than Qg4 against 6...Ne7 but scored 21 percentage points and 399 elo better with the latter). If 6...Qa5 play 7.Bd2 Qa4 8.Qg4. If 5...Ba5 play 6.b4 cxd4 and now Anand played equally often and roughly equally successfully 7.Nb5 Bc7 8.f4 Bd7 and 7.Qg4!? Ne7 8.bxa5 dxc3 9.Qxg7 Rg8 10.Qxh7 Nbc6 11.f4.
Against the Rubinstein: 3...dxe4 play 4.Nxe4 meeting 4...Nd7 with 5.Nf3 Ngf6 6.Nxf3+ Nxf6 7.Bd3 c5 8.0-0 cxd4 9.Nxd4 and meeting 4...Bd7 with 5.Nf3 Bc6 6.Bd3 Nd7 7.0-0 Ngf6 8.Ng3.
Against the Caro-Kann: 1...c6 2.d4 d5 play 3.Nc3 dxe4 4.Nxe4, meeting 4...Bf5 with 5.Ng3 Bg6 6.h4 h6 7.Nf3 Nd7 8.h5 Bh7 9.Bd3 Bxd3 10.Qxd3 e6 11.Bf4, meeting 4...Nd7 with 5.Bc4 Ngf6 6.Ng5 e6 7.Qe2 Nb6 8.Bb3 h6 9.N5f3, continuing against 9...c5 with 10.Bf4 Nbd5 11.Be5 Qa5+ 12.Nd2 and continuing agasint 9...a5 with 10.a4 c5 11.Bf4 Bd6 12.Ne5, and meeting 4...Nf6 with 5.Nxf6+ exf6 (no games with 5...gxf6 in Mega20) 6.Nf3 Bd6 7.Bd3 0-0 8.0-0.
Against the Pirc: 1...d6 2.d4 Nf6 3.Nc3 g6 play 4.f4 Bg7 5.Nf3, meeting 5...0-0 with 6.Bd3 (Anand has played 6.a3 equally often but not so successfully) and meeting 5...c5 with 6.Bb5+ Bd7 7.e5 Ng4 8.e6, continuing against 8...Bxb5 with 9.exf7+ Kd7 10.Nxb5 Qa5+ 11.Nc3 cxd4 12.Nxd4 Bxd4 13.Qxd4 Nc6 14.Qc4 and continuing against 8...fxe6 with 9.Ng5 Bxb5 10.Nxb5!? Qa5+ 11.c3 Qxb5 12.Nxe6.
Against the Modern: 1...g6 2.d4 Bg7 3.Nc3 d6 play 4.Be3, meeting 4...a6 with 5.Qd2 b5 6.a4!? b4 7.Nd1 a5 8.c3, meeting 4...c6 with 5.Qd2 b5 6.f4 and meeting 4...Nf6 with the repertoire-consistent 5.Qd2 or 5.h3 (Anand has only faced 4...Nf6 twice in Mega20, drawing with 5.Qd2 and winning with 5.h3).
Against Alekhine's Defence: 1...Nf6 play 2.e5 Nd5 3.d4 d6 3.Nf3, meeting 4...Bg4 with 5.Be2, continuing against 5...e6 with 6.0-0 Be7 7.c4 Nb6 8.Nc3 0-0 9.Be3, and continuing against 5...c6 with 6.0-0 Bxf3 7.Bxf3 dxe5 8.dxe5 9.Nd2!?
Against the Scandinavian: 1...d5 with 2.exd5 Qxd5 3.Nc3 meet both 3...Qa5 and 3...Qd6 with 4.d4 Nf6 5.Nf3. Anand has only faced 2...Nf6 four times in Mega20, scoring wins with 3.Bb5+ and 3.c4 but a draw and a loss with 3.d4.
Black v 1.e4
Play 1.e5 aiming to meet the Spanish with the Marshall Attack: 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Be7 6.Re1 b5 7.Bb3 0-0 8.c3 d5. After 9.exd5 Nxd5 10.Nxe5 Nxe5 11.Rxe5 c6 meet 12.d4 with 12...Bd6 13.Re1 Qh4 14.g3 Qh3, meet 12.d3 with 12...Bd6 13.Re1 Bf5 14.Qf3 Qh4 15.g3 Qh3 and the one time Anand faced 12.Re1 he played 12...Bd6. If 9.d4 play 9...exd4 10.e5 Ne4 11.cxd4 Bf5.
If White avoids the Marshall with 8.h3 play 8...Bb7 9.d3 d6 10.a3 Na5 11.Ba2 c5. Against 8.a4 play 8...b4 9.d3 d6, against 8.d4 play 8...d6 9.c3 Bg4 and against 8.d3 play 8...d6 9.c3 Na5 10.Bc2 c5 11.Nbd2 Re8 12.Nf1 Bf8.
Against the Exchange Variation: 4.Bxc6 play 4...dxc6, meeting 5.0-0 with 5...f6 6.d4 Bg4 7.dxe5 Qxd1 8.Rxd1 fxe5 9.Rd3 Bd6 and meeting 5.Nc3 (no other fifth moves in Anand's games) with 5...f6 6.d4 exd4 7.Qxd4 Qxd4 8.Nxd4 Bd6, after which he quickly castled kingside.
Against the Italian Game: 3.Bc4 play 3...Bc5, meeting 4.c3 Nf6 5.d3 (White's fifth and sixth moves are often reversed) with 5...d6, continuing against 6.0-0 with 6...a6 7.Bb3 Ba7. If 7.a4 or 7.Nbd2 play 7...0-0. Continue against 6.Bb3 with 6...0-0, continue against 6.Nbd2 with 6...a6 7.Bb3 0-0, continue against 6.b4 Bb6 7.a4 with 7...a5 8.b5 Ne7 and continue against 6.Bg5 with 6...a6!?
If 5.d4 play 5...exd4 meeting 6.e5 d5 7.Bb5 Ne4 8.cxd4 with 8...Bb6 and meeting 6.cxd4 Bb4+ 7.Nc3 (no games with 7.Bd2) Nxe4 8.0-0 Bxc3 9.d5 with 9...Ne5!?
Against the Evans Gambit: 4.b4 play 4...Bxb4 5.c3 Ba5.
After 4.0-0 Nf6 Anand has only faced in Mega20 5.d3 when 5...d6 will almost certainly transpose to lines already covered.
Against the Scotch: 3.d4 exd4 4.Nxd4 Anand has equally played 4...Nf6 and 4...Bc5 but done much better percentage-wise and rating-wise with the latter, and against the Scotch Gambit: 4.Bc4 he plays 4...Bc5, meeting 5.c3 with 5...Nf6.
Against the Scotch Four Knights: 3.Nc3 Nf6 4.d4 exd4 5.Nxd4 play 5...Bb4 and against the Spanish Four Knights: 4.Bb5 play 4...Bb4.
Against the Ponziani: 3.c3 play 3...Nf6 4.d4 Nxe4 5.d5 Ne7 6.Nxe5 Ng6.
Against the Bishop's Opening: 2.Bc4 play 2...Nf6, meeting both 3.d3 and 3.Nc3 with 3...Bc5.
Against the King's Gambit: 2.f4 play 2...exf4, meeting 3.Bc4 with 3...Nf6 4.Nc3 c6 and meeting 3.Nf3 with either 3...d5 or 3...g5 (Anand has played each move once, drawing both times).
Against the Centre Game: 2.d4 exd4 3.Qxd4 Nc6 4.Qe3 play 4...Nf6 and meet both 5.Bd2 and 5.Nc3 with 5...Bb4.
Black v 1.d4 and Others
Against 1.d4 aim for a Nimzo-Indian: 1...Nf6 2.c4 e6 3.Nc3 Bb4, meeting 4.Qc2 with 4...d5 5.cxd5 Qxd5, continuing against 6.Nf3 with 6...Qf5 7.Qxf5 exf5 8.a3 Be7 or 7.Qb3 Nc6 and continuing against 6.e3 with 6...c5 7.Bd2 Bxc3 8.Bxc3 cxd4 9.Bxd4 Nc6 10.Bc3 0-0 11.Nf3 Rd8 12.Be2 Qe4. If 5.a3 play 5...Bxc3+ 6.Qxc3 Ne4 7.Qc2 c5 8.dxc5 Nc6, at which point Anand has only faced 9.cxd5, meeting it with 9...exd5 10.Nf3 Bf5 11.b4 d4!? 12.g4 Bg6.
Against 4.e3 play 4...0-0, meeting 5.Bd3 with 5...d5, continuing against 6.Nf3 with 6...c5 7.0-0 dxc4 8.Bxc4 Nbd7, continuing against 6.cxd5 with 6...exd5 7.Nge2 Re8 and continuing against 6.a3 (no games with 6.Ne2) with 6...Bxc3+ 7.bxc3 dxc4 8.Bxc4 c5, meeting 5.Nge2 with 5...d5 6.a3 Bd6 (Anand has scored +2=0-0 with this but just one draw in three tries with the more popular 6...Be7) and meeting 5.Nf3 with 5...d5, after which the vast majority of games see 6.Bd3, which transposes to 5.Bd3 d5 6.Nf3.
Against 4.Nf3 Anand's most common choice has been 4...d5 but he has scored 25 percentage points and 115 elo better with 4...b6. Meet 5.Bg5 Bb7 6.e3 with 6...h6 7.Bh4 g5 8.Bg3 Ne4, meet 5.e3 with 5...Ne4!? 6.Qc2 Bb7 7.Bd3 Bxc3+!? 8.bxc3 f5 9.0-0 0-0 and meet 5.Qc2 with 5...Bb7 6.a3 Bxc3+ 7.Qxc3 0-0.
Against 4.f3 play 4...d5 5.a3, at which point Anand has scored a win and a draw with the main move 5...Bxc3+ and with the slightly unusual 5...Be7!?
Against 4.a3 play 4...Bxc3+ 5.bxc3 c5, and against 4.g3 play 4...0-0 5.Bg2 d5 6.Nf3 dxc4 7.0-0 Nc6.
If White avoids the Nimzo-Indian with 3.Nf3 Anand has played the Queen's Gambit Declined: 3...d5 and the Queen's Indian: 3...b6 almost equally often but has scored 12 percentage points more with the latter, which is a better repertoire fit bearing in mind his playing the Nimzo-Indian line 4.Nf3 b6.
Against 4.Nc3 the move 4...Bb4 transposes to the aforementioned Nimzo-Indian line.
Against 4.g3 play 4...Ba6, meeting 5.b3 with 5...Bb4+ 6.Bd2 Be7 7.Bg2 c6, continuing against 8.Bc3 with 8...d5 9.Ne5 Nfd7 10.Nxd7 Nxd7 11.Nd2 0-0 12.0-0 Nf6 or 9.Nbd2 Nbd7 10.0-0 0-0 11.Re1 c5 12.e4 dxe4 13.Nxe4 Nxe4 14.Rxe4 Bb7 and continuing against 8.0-0 with 8...d5. Meet 5.Nbd2 with 5...Bb4, continuing against 6.Qc2 with 6...Bb7 7.Bg2 Be4 and continuing against 6.a3 (the only other sixth move Anand has faced) with 6...Bxd2+ 7.Nxd2 Bb7 8.Nf3 d5. Meet 5.Qb3 with 5...Nc6 6.Nbd2 d5!? Meet 5.Qa4 with 5...Bb7 6.Bg2 c5 7.dxc5 Bxc5 8.0-0 0-0 9.Nc3 Be7. Meet 5.Qc2 with 5...Bb7 6.Bg2 c5 7.d5 exd5 8.cxd5 Nxd5 9.0-0 Be7, continuing against 10.Rd1 with 10...Qc8 and continuing against 10.Qe4 with 10...Bc6 (an Anand novelty that became the main line) 11.Ne5 Nf6 12.Nxc6 Nxc6 13.Qa4 0-0.
Against 4.a3 play 4...Bb7 5.Nc3 d5, meeting 6.cxd5 with 6...Nxd5, continuing against 7.Qc2 with 7...Nxc3 and answering both 8.Qxc3 and 8.bxc3 with 8...Be7, continuing against 7.e3 with 7...g6 8.Bb5+ c6, continuing against 7.Bd2 with 7...Nd7 and continuing against 7.Qa4+ with 7...Qd7!? 8.Qc2 Nxc3 9.bxc3 (Anand has not faced the rarer 9.Qxc3) Bd6.
Against 4.e3 play 4...Bb7, meeting 5.Bd3 with 5...d5 6.0-0 dxc4!? 7.Bxc4 a6 and meeting 5.Nc3 with 5...d5 6.cxd5 exd5 7.Bb5+ c6 8.Bd3 Bd6 9.0-0 0-0.
Against 4.Bf4 play 4...Bb7 5.e3 Nh5!? 6.Bg5 Be7 7.Bxe7 Qxe7 8.Nc3 f5.
If White avoids the Nimzo-Indian with 3.g3, transpose to a Catalan with 3...d5. After 4.Nf3 Be7 5.Bg2 (White's last two moves are often played in reverse order) 0-0 6.0-0 play 6...dxc4, meeting 7.Qc2 with 7...a6, continuing against 8.Qxc4 with 8...b5 9.Qc2 Bb7 10.Bd2 Ra7!? 11.Rc1 Be4 12.Qb3 Nc6 13.e3 Qa8 14.Qd1 b4, and continuing against 8.a4 with 8...Bd7 9.Qxc4 Bc6.
If 7.Ne5 play 7...Nc6, meeting 8.Bxc6 bxc6 9.Nxc6 with 9...Qe8 10.Nxe7+ Qxe7 and meeting 8.Nxc6 bxc6 9.Na3 (the only move Anand has faced) with 9...Bxa3 10.bxa3 Nd5 11.Qa4 Nb6 12.Qxc6 Rb8 13.Qc5 Bb7 14.e4 Qd6.
If 7.Qa4 play 7...a6 when 8.Qxc4 (practically the only move played) is a transposition to lines with 7.Qc2.
If 7.Na3 play 7...Bxa3 8.bxc3 b5!? 9.a4 a6.
Against 6.Qc2 play 6...c5, meeting 7.0-0 with 7...cxd4 8.Nxd4 Qb6!?, and meeting 7.dxc5 with 7...Qa5+ or 7...Bxc5 (Anand has played each move once, drawing both times), and against 6.Nbd2 (the only other sixth move Anand has faced) play 6...c5.
Against 2.Nf3 Anand's most-common response has been 2...d5 but a much better repertoire fit is 2...e6, with which he has scored 14 percentage points and 103 elo better. Nearly all his opponents responded to 2...e6 with 3.c4, which is a transposition to lines already covered. Similarly 3.g3 transposes to the Catalan after 3...d5. Against 3.e3 play 3...b6 with more transpositional possibilities. The only truly independent third move Anand has faced is  3.Bg5, which he once beat with 3...h6 and the other time with 3...c5.
Against the New London: 2.Bf4 play 2...d5 3.e3 e6, meeting 4.Nf3 and 4.Nd2 with 4...Bd6 and meeting 4.c4 with 4...Be7.
Against the Trompowsky: 2.Bg5 play 2...e6, meeting 3.e4 with 3...h6 and meeting 3.Nd2 with 3...Be7.
Against 2.Nc3 play 2...d5, meeting the Veresov: 3.Bg5 with 3...Nbd7 and meeting the Jobava-Prié: 3.Bf4 with 3...e6 4.Nb5 Na6 or 3...c6 4.e3 Bf5 (Anand has played each line once, winning both games).
Against 1.Nf3 Anand has played 1...d5 and 1...Nf6 almost equally often, but the latter is a better repertoire fit, eg 2.c4 and 2.d4 can be met by 2...e6. Against the King's Indian Attack continuation 2.g3 play 2...d5 3.Bg2 c6 3.0-0 Bf5. Against 2.b3 play 2...c5 3.Bb2 Nc6.
Against the English: 1.c4 play 1...e5, meeting 2.Nc3 with 2...Nf6, continuing against 3.Nf3 with 3...Nc6 4.g3 Bb4 5.Bg2 0-0 or 5.Nd5 Bc5 6.Bg2 0-0. If 4.e3 play 4...Be7 5.d4 exd4 6.Nxd4 0-0. If 4.a3 (the only other fourth move Anand has faced) play 4...g6 5.g3 Bg7 6.Bg2 0-0 7.0-0 Re8!?
Against 3.g3 play 3...Bb4 4.Bg2 0-0, meeting 5.Nf3 with 5...Re8 6.0-0 e4 7.Nd4 Nc6, meeting 5.e4 with 5...Bxc3 6.bxc3 (Anand has not faced 6.dxc3) Re8!? and meeting 5.Qc2 with 5...Re8.
Against 2.g3 (no Anand games in Mega20 with 2.Nf3) play 2...Nf6 3.Bg2 (3.Nc3 is a transposition to 2.Nc3 Nf6 3.g3), at which point Anand has played almost equally often 3...Nc6 and 3...c6 but has scored 12 percentage points and 192 elo better with the latter. After 3...c6 meet 4.d4 with 4...e4!?, continuing against 5.Nc3 and 5.Bg5 (the only fifth moves Anand has faced) with 5...d5. Meet 4.Nf3 (the only other fourth move Anand has faced) with 4...e4 5.Nd4 (Anand has not faced the rarer 5.Ng5) d5.
Against 1.b3 play 1...e5 2.Bb2 Nc6 3.e3 Nf6.

Here Anand faces the Spanish of French grandmaster Maxime Vachier-Lagrave.
Vachier-Lagrave (2760) - Anand (2789)
Norway Chess (Stavanger) 2018
Spanish Open
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Ba4 Nf6 5.0-0 Nxe4
No Marshall Attack today. The move 5...Be7 is more common in Anand's praxis, and he has scored much better with it percentage-wise and rating-wise, but like all of today's top players, Anand sometimes varies so as not to present a stationary target.
6.d4 b5 7.Bb3 d5 8.dxe5 Be6 9.c3 Be7
Anand has also played 9...Bc5, but has performed much better with the text.
10.Nbd2 0-0 11.Re1 Nc5 12.Nd4 Nxd4 13.cxd4 Nd3
Winning the bishop-pair with 13...Nxb3 is a serious alternative. The text first appears in ChessBase's 2020 Mega database in a game of Alekhine's from 1914.
14.Re3 Nf4
Carlsen preferred another way of winning the bishop-pair, namely by 14...Nxc1, in a loss to Shirov in 2007.
15.Nf3 Bg4 16.h3 Bh5 17.Rc3
White highlights the main strategic problem with the black position - a backward c pawn on a half-open file.
17...Ne6 18.g4!?
Breaking the pin but inevitably weakening the kingside.
18...Bg6 19.Be3 a5 20.Bc2 Bb4 21.Rb3 f5 22.exf6 Bxc2 23.Qxc2 Qxf6 24.Ne5
How can Black solve the problem of his backward c pawn?
*****
*****
*****
*****
*****
24...c5!?
Anand is prepared to give up the exchange in order to get rid of his c-file weakness.
25.Nd7 Qf7 26.Nxf8
Not 26.Nxc5? Bxc5 27.dxc5 d4 etc.
26...Rxf8
Black is down the exchange, but he has the better bishop and his pieces generally are more active.
27.Qf5
Later the same year the game Mihail Kobalia (2620) - Babu Lalith (2529), Rilton Cup (Stockholm) 2018, saw White give back the exchange by 27.a3!? c4 28.axb4 cxb3 29.Qc6 (½–½, 34 moves). Grandmaster Krisztián Szabó in Mega20 gives the likely improvement 29.Qxb3!?, continuing 29...Qf3 30.Qd1 Qxh3 31.bxa5, analysing the position out to equality.
27...cxd4
Szabó's suggestion of 27...c4!? seems stronger.
28.Qxf7+ Rxf7 29.Rxb4?
It seems Black had to play 29.a3 Nc5 (forced) 30.Rxb4 axb4 31.Bxd4 Nb3 32.Rd1 "and the position is drawish" - Szabó.
29...axb4 30.Bd2 b3 31.axb3
Material is temporarily equal, and White has the advantage - other things being equal - of having the better minor piece to work with a rook, but other things are not equal, as Black quickly demonstrates.
31...Rf3 32.b4 Rd3 33.Re1 Kf7 34.Bc1 Rxh3 35.Re5?!
Stockfish11 and Komodo11.01 prefer 35.Kg2 Rb3 36.Re5 Rxb4 37.Rxd5, but Black is a solid passed-pawn up.
35...Rd3 36.Kf1 Rd1+ 37.Re1 Rxe1 38.Kxe1 g6 39.f4 Nd8 40.g5 Ke6 0-1
White is a pawn down and helpless to a knight and/or king invasion on the light squares.