Out-Preparing the Candidates with Fat Fritz Part 1
by Tanmay Srinath - 21/03/2020
As the Candidates headed into the rest day, one thing that is obvious is how each player is fighting unto their very last reserves for that extra half point! We have seen decisive games in all the 3 rounds till now, and with such zest it is impossible to pick a clear favourite. One thing that stands out however, is the high level of opening preparation. Out of the 12 games played, 10 games have witnessed the absolute peak as far as homework is concerned. However, Tanmay Srinath set out to beat the preparation of the Candidates with the help of Fat Fritz and come up with some really deep ideas that can be used by any player in their future games.
After the advent of the Corona Virus, we have been stuck in our homes, waiting for a respite. In such a situation, the FIDE Candidates 2020 is a godsend! In the 1st 3 rounds, there have been 5 decisive games, and in the remaining grandiose battles have left me mesmerised at the level of the players playing in the event. There has been extensive coverage on ChessBase India with round reports and videos, and here in these series of articles appearing on every rest day and after the closing ceremony I will attempt to improve on the opening preparation of the players using Fat Fritz, the strongest engine in the world currently.
All the analysis in this article has been done by using Fat Fritz engine that comes along with Fritz 17
Now let me be honest with the readers before proceeding. Compared to the seconds of these players who had months to prepare their lines, I have had only three solid days to prepare my notes, and unlike Deep Mind I don't have 4 super powerful TPUs. What I do have however, is a lot of experience as an analyst for ChessBase India, and working with engines. The ideas I have managed to find have been in part due to me actively participating with the engine and understanding each and every position, i.e. working with the engine and not for it. This is important, as it will not only result in analysis that will promise the best practical results.
It's always a combination of the human and machine that will bring you the best results! | Photo: Vectra AI
With that in mind, let me list out the games where I found concrete improvements - MVL - Caruana and Grischuk - Alekseenko. The latter is more of an alternative way to play, so we will look at it second, and concentrate on how MVL could have posed more problems for his opponent to solve!
MVL - Caruana
Maxime Vachier Lagrave had only two weeks to prepare for the Candidates, and while his openings still need a bit of work, it is clear that the break from chess has been beneficial for him! He has come fresh and invigorated after a 2019 where he conquered titles but missed out on direct qualification by the smallest of margins. However, fortune favours the brave, and MVL has taken the chance that Radjabov gave up, and by the quality of his play so far he is definitely is going to fight for the title. His opponent, Fabiano Caruana, on the other hand, has come with a killer instinct turned on - I have never seen Fabiano Caruana play chess this aggressively, looking for possibilities to pose maximum problems for his opponent, and not caring for the objectivity as much as he used to! A sad loss notwithstanding, I still see Fabi as the favourite to win this thing - he is the rating favourite and has played the best chess so far.
The man on a mission! | Photo: Lennart Ootes
Things however, could have turned out differently had MVL taken a deeper look at the Neo-Archangel, Fabi's weapon of yore, and an uncompromising system where Black gives White tempi to build up c3 and d4 in the center but keeps a set of active bishops, and chances to play for a win. Time to dive deep!
This is the modern day tabiya of the Neo- Archangel : Black keeps his light squared bishop flexible at the cost of giving up a pawn, and modern theory suggests that after the critical 10.axb5 axb5 11.Na3 he seems to be holding his own. Thus, MVL's choice 10.a5!? seems more pragmatic. In the press conference both sides called this a very sharp line. It seemed to me (the viewer) that MVL was still under-prepared in these lines, and it showed just 4 moves later.
This is a critical position, and perhaps the most instructive moment of the game. Maxime went for 15.Ng5 here, protecting the e4 pawn and threatening Nxf7, which Caruana allowed in the game, but the former went wrong almost immediately and White didn't get much from there on. Can you do better than Maxime? White to play.
The right move is 15.Qd2! This move has been seen previously in a correspondence game, so it is an over the board novelty. What is White's idea? Well for starters the e4 pawn is indirectly protected, as the b4 knight will hang at the end of the variation 15...Nxe4 16.Nxe4 Bxe4 17.Qxb4. Secondly, White simply completes his development by connecting the rooks, and the onus is on Black to find the best possible moves. He has to play 15...c5! (otherwise White just continues Rad1) 16.dxc5 dxc5 17.Rad1! Qxa5!? (critical, probably safer is 17...Qxd2, but we shall see this later).
If there is one position that Maxime could have been afraid of over the board, it has to be this - Black has simply won a pawn and White has seemingly no immediate compensation. Add this to the fact that Fabi would have known all the good moves here, and it makes for a scary proposition for MVL! However after 18.e5! Rbd8! (only move, can you see what happens after 18...Bxf3? 19.gxf3 Rbd8 20.Qe2! Nh5 21.Bg5! Rxd1?)
Well, White has the superb tactical shot 22.Bxf7+!! and he is winning in all the variations succeeding this sacrifice - check the notes provided at the end of the diagrams.
Now we turn our attention to Black's only move that keeps him in the game - 18...Rbd8! Here White has two moves - the untested 19.Qe1!?N, which is Fat Fritz's choice, and involves accepting an exchange sacrifice, and my move 19.Qe2, which is the main line, as it keeps more sanity in the position.
After a sequence of forcing moves from 19.Qe2 we reach this position, which is critical for the evaluation of the line. White wins a piece, but Black has no obvious weaknesses, and a dangerous 3 vs 1 majority on the queenside. However, with this position still having a lot of pieces on the board, I prefer White, due to the fact that a piece is better than two pawns in general, and especially in the middlegame. The computer agrees, giving White around 0.50 here, and that shows while Black has some compensation, it is insufficient for equality.
Returning to White's novelty 19.Qe1!?, after 19...Bxf3 20.gxf3 Rxd1 21.Rxd1 Rd8 22.Qb1!? (22.Qe2 should simply transpose - check the notes) we reach this position. White is a pawn down and has a fractured structure, but he is the one creating threats, and soon Black is forced to give up an exchange. We reach the following diagram if both sides play their best moves:
White does have the exchange for a pawn, but this position didn't appeal to me personally because of the following reasons - Black doesn't have a clear weakness in his position, his pawns restrict our bishop and the knight can find a good spot on f4 eventually, and that with his queenside majority will make this tough to win. However, if the reader is fine with grabbing material, this line should suit her/him.
I have attached my analysis here. I won't be stupid enough to call it perfect, but I think this should be a reasonable file to use and improve on:
We will be keeping an eye on MVL's preparation in the rounds to come | Photo: Lennart Ootes
Does this much of chess scare you? I hope not! The real innovations are yet to come!
This was the most intriguing clash of Round 1. Would experienced Grischuk prevail over his young compatriot, or would Kirill show that his performance at the Grand Swiss was not a fluke? Turned out it was the latter, but only just, as in Sasha's time trouble he missed a golden opportunity to press for eternity. However, we are not concerned with the entire game. The opening, however, does merit our attention. Let us quickly enter the critical position:
I find Danill Dubov to be a really good commentator - he is objective and keeps an open mind. Here he mentions that it was him that first vouched for Black in these structures back in 2014/15. Well, it seems to me that he is right in saying that black has a playable position, but I don't think he equalises, and while 8.Qc2 by Grischuk in the game is a perfectly good move, I think my novelty here leads to positions that should be slightly more pleasant for White, and while it is not a strict improvement, it seems to leads to better chances practically, as black has numerous ways to go wrong:
8.a4!? is my proposition for White here. It has never been tried before, but it leads to positions that are promising for the White player in my opinion. What is the idea you might ask? Well, White wants to play Nf3 and castle, but doing so immediately runs into e4!. So he has to wait. How does he do it? Well 8.Qc2!? is one way, but as I said previously it might be better just to keep the queen flexible. Thus what to do? When in doubt push the rook pawn! (Alpha Zero's teachings!) Also, the deeper meaning of this advance is explained in notes after 8.a4!? a5.
I will proceed now to show some sample positions that can be reached. Please check the analysis if you want to know how exactly this position is reached!
You will find this move h6 played by the machine numerous times in these structures. Basically Black wants to play a useful move while waiting for White to define his piece configuration. Here White should take on b6, and play Nd2, to improve the placement of the knight and the g2 bishop.
More tension = less advantage? Not quite! Here White is benefited by the fact that he is yet to choose his pawn configuration in the center, so he can switch plans and aim for e4 and f4, rather than d4
This is perhaps the critical test of the line - what if Black responds to 8.a4 with 8...a5? Well, the inclusion of these moves seem to favour White for a very specific reason - Black loses the flexibility of his queenside majority, while White's move is a useful one to make - it restricts the b5 break and allows for Ba3 in some lines. The notes provide more detailed explanations, but this is the gist.
This is an extreme example of when both Black and White try to get what they want, but White should be the happier side here - good central control and chances for a quick and devastating kingside attack.
With all that covered let us look at the critical move according to Fat Fritz - 8...Re8. Here I recommend the surprising 9.Qc2! - again White wants to avoid being hit with e4, and while this is not the first choice of the engine it seems to lead to positions that are easier to play for White due to the better pawn structure he will possess.
It's in one of the sidelines of the 9.Qc2, but it shows what exactly White wants to do - he wants to trade his isolated a-pawn and get one pawn island compared to Black's two. This rather simple idea sometimes leads to large advantages - long term ones that is!
What about 9...a5? Well, as we can see from this position, the inclusion of a4 and a5 has favoured White, and he has a tangible advantage due to simply being more flexible.
In the main line Black is forced to capture away from the center to retain the best chances of equality - something which most people won't do willingly. Even if black finds all this, with best play we reach the following diagram:
White holds a small advantage here due to the following two aspects - better pawn structure and a clear weakness to attack. Black is struggling for a plan, and while some tests are required, it is clear that only White can be pressing here.
Here are the extensive notes on the novelty 8.a4!?
I know, I know - Fat Fritz's suggestions always seem to work! | Photo: Maria Emelianova
Here are two more interesting positions for the readers to finish this article -
Ding Liren - Wang Hao
Dubov praised Ding's next move 11.e4!?, calling it a supercharged Chigorin Ruy Lopez Reversed! Well, I tend to agree,but here are better options - 11.Bb2!? is one, trying to keep the game in Reti/ English Territory.
Caruana - Alekseenko
If there is one guy's opening play I can't improve on, it's Fabiano's! Here his move 13.d6! is the best practical chance, forcing Black to find only moves to survive. Well, what about 13.h3!?, which Fat Fritz likes initially? Well, this is one case where I have to bow my head to Caruana's team - they noticed what was wrong with the line almost immediately - 13.h3!? Bxf3! 14.Rxf3 Bd6 15.Bg5 Qe7 16.Nb5 and now the only move, which is easier (than the game) to find for a player of Kirill's ability:
16...Bb8! and Black is completely fine!
The 3rd Round did have a fascinating clash - Ding Liren managed to play like an engine and beat off Caruana's preparation almost effortlessly! However, there are a few points I would like to make here - Ding's move 10.Nxe5!? is probably not the best as far as Fat Fritz is concerned - it evaluates it at 0.64 after some thought, which suggests that the pawn sacrifice needs to be taken seriously. However, there are two other options that don't win material, but give White a much better position as far as static considerations are taken into account. Let us look at them here.
The idea that Fabi played in the game is really interesting and gutsy - Black simply gives up a pawn for long term compensation. However, there is a reason I consider this move to be a one game idea, and in the subsequent lines I offer White two simple ways to gain a long term positional advantage. When you look at this move, it becomes clear that somehow the concept is positionally not a 100 percent correct - Black is not that much better developed, and the White king is reasonably safe. Thus, when I checked it deeply with Fat Fritz, I found two interesting continuations.
I can confidently state that this move, along with the idea succeeding it, effectively refutes Black's idea. Why is it so? Well, the move e5 wants White to be greedy and take material, but why should I do it, when I have continuations that promise me a long term static advantage? Logically, this move makes perfect sense - counter-attacking in the center is the best defense against the provocative thrust 9...e5!. The point of this move is revealed in the following line - 10...exd4 11.Na2! b5(seems to be the only move) 12.Nxb4 bxc4 13.exf5 d3 14.Be3 Re8 and now the powerful defensive sacrifice 15.Bxd3!
White returns the extra material, but develops all his pieces and secures his king. After a few more logical moves we reach the following position
White is materially not better, but look at his pieces! All of them are centralised and active, and the c-file is ridden with Black weaknesses to attack. It is not a winning advantage, but definitely a tangible one, and Black should look forward to a torturous defense. Fat Fritz gives White around 0.65 here, which is almost a vertical plus minus, and shows how bad Black's position is here.
There is an alternative to this on the 10th move - 10.dxe5!?. The move is not as clear cut as 10.e4!, but Fat Fritz still gives White an evaluation of 0.62, and it should be an option to consider.
This position is reached after the critical line in 10.dxe5 Nfd7. It is easy to see why White is tangibly better here - he has a monster bishop on c4, better pieces in general and a dangerous central majority that can decide the battle in the middlegame. This was the safer option to consider.
You too can now use Fat Fritz for your analysis. It comes along with the Fritz 17 software:
About the author:
Tanmay Srinath has been writing for ChessBase India since quite some time now. His tournament reports and depth of analysis have been widely appreciated. Pursuing a full-fledged career in engineering Tanmay doesn't get enough time to pursue chess, but he loves to follow top level encounters and analyzes those games with his Fat Fritz engine. With so many complex games being played at the Candidates, we thought it would be a perfect time for Tanmay to use his expertise of working with Fat Fritz to check the analysis of these world class players. We hope you find his analysis useful in your games.