King's Indian Mar del Plata: The 9...a5 Gamble Against the Bayonet Attack and the Price of the "Low-Theory" Promise

On the analysis screen, Black's ninth move appears: a5. I press pause, rewind, press play. No typo.

It is not Nh5.

For years, every time I sat in front of this position — the King's Indian Defense, the Mar del Plata system, the Bayonet Attack branch — my head held only one reflex for Black: the knight goes to h5, prepares f5, and tears open White's kingside. That move is the main line in almost every database I have ever searched. It is the move I once taught students with the certainty of someone who believed opening theory is a one-way road, with no turn-offs.

Then an opening product put another option on my desk: 9...a5. An old move, a secondary branch, repackaged as an "antidote" to the Bayonet. What made me stop was not the move itself. What made me stop was the way it was being sold.

I once loaded the wrong SHB Da Nang tape, and that taught me football does not forgive carelessness. Chess is the same, except a board has no stands to hide mistakes. Every error stays in the scoresheet, waiting for a more careful reader to find it.

Mar del Plata: the system of those willing to concede the centre

To understand why a move like 9...a5 matters, we have to go back to where the system begins.

The King's Indian Defense starts with 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7. Black deliberately does not occupy the centre. Black lets White build a large pawn centre, then attacks that very centre with ...e5 and pressure from the pieces. This is hypermodern thinking, a theory that in the 1920s was still treated as heresy.

The Classical variation adds 5.Nf3 0-0 6.Be2 e5 7.0-0 Nc6 8.d5 Ne7. Both sides have locked the centre. White owns space on the queenside; Black will attack on the kingside. This is a position every player recognises by name: Mar del Plata.

The name belongs to the Argentine city of Mar del Plata, where the system reached the highest level in 2026. Miguel Najdorf and Svetozar Gligoric laid the foundations. Later, Mikhail Tal and Bobby Fischer turned it into a familiar weapon in major events. The ECO range for this branch sits between E97 and E99.

The geometry is so simple it is easy to read too quickly. White will push queenside pawns, open the c-file, and bring rooks in. Black will push f5, then g5, g4, ripping apart the kingside. In most classical Mar del Plata games, both sides sprint on opposite wings and the first to arrive wins. That is why the system is famous as an all-in affair: White attacks the queenside without a safe king, Black attacks the kingside without a centre.

At a high level the price is an enormous amount of theory. Black must remember exact sequences in forcing lines, because a single tempo off and the kingside collapses before White's queenside pawns can decide anything. This is the territory of good memorisers.

Then the Bayonet Attack appeared.

The Bayonet Attack is the 9.b4 branch. White pushes the b-pawn to the fourth rank, grabbing space in front of Black's pieces, preparing b5 to drive the knight off c6 and control key queenside squares.

It sounds simple. The real change lies elsewhere: 9.b4 slows Black's kingside attack. A knight on c6 loses its natural retreat to d7. Black must rearrange pieces, and while doing so, White has time to open files and advance.

From the 1990s onward, the Bayonet became a serious elite choice. It suited an era in which young players were raised on databases: instead of entering a calculation race on the kingside, White drags the game onto different ground, where space and pawn structure speak first.

Black has two answers.

The first is 9...Nh5, the main move. The knight jumps to the rim, prepares f5, keeps the kingside-attack philosophy intact, and opens forcing lines everyone must memorise. The character after 9...Nh5 is open, concrete, calculable.

The second is 9...a5, the move the product wants to sell.

9...a5: the mechanism of a quiet move

Before plans, the technical nature of 9...a5 must be stated clearly.

White's b4-pawn is the anchor of the entire queenside. It holds space, it prepares b5, it restrains Black's knight. The move 9...a5 strikes at that anchor with a flank pawn. No move is more direct.

The subtlety is that 9...a5 does not change the structure. It creates new tension. The a5-pawn and the b4-pawn stare at each other, and White must decide first: trade, push, or defend. In chess, forcing the opponent to decide first on a wing where you have invested nothing is a form of time advantage.

After 9...a5, Black no longer commits to a kingside attack. Black keeps the formation, waits to see White's choice, and only then deploys. The centre stays closed. The kingside stays intact. The position drifts into a closed type, where both sides must improve piece placement in small increments.

That is why the product markets it with the phrase "very few concrete lines". If there are no forcing lines, there is nothing to memorise. If there is nothing to memorise, a club player has a chance against an opponent who knows the book.

Theoretically, one fact deserves mention: 9...a5 is not a new invention. It is a long-recognised branch, a secondary option to 9...Nh5. The product itself claims no novelty. It sells a practical solution, not a discovery.

White's three choices

After 9...a5 10.bxa5, White opens the b-file for the rook, but Black gains ideas against the centre. In such positions White usually cannot keep the extra pawn, and opening the b-file is not automatically an advantage if the minor pieces are not placed yet.

The second option is 10.b5, pushing the pawn, maximising White's space, but turning b5 into a fixed target. Black can pin it with the a-pawn, bring a knight to central squares, and conduct the kingside plan in silence. The position shifts fully into blockade mode: White is strong on the queenside but must prove that space converts into points.

The third option is reinforcing b4 with 10.Ba3, keeping the structure and waiting. Black can then trade on a and exploit the empty squares, or delay and continue development. Both sides must play pure positional chess: improving each piece, fighting for each square, waiting for a structural error.

Three scenarios, three different games. That is the strength of 9...a5 for a buyer: it does not force Black to memorise a single line, but opens a family of positions to handle.

That is also its weakness. A family of positions does not mean easy. It means difficult in a different way.

9...Nh5 and 9...a5: two philosophies, one board

Placed side by side, the difference is clear.

9...Nh5 takes the game into the concrete zone. After the knight reaches h5, Black prepares f5, and White must react precisely. Lines here can be calculated to the end. The upside: Black controls the tempo. The downside: if White is better prepared, Black loses quietly because the book has been read.

9...a5 takes the game into the strategic zone. Black gives up forcing the tempo in exchange for flexibility. The upside: White cannot beat Black with a single line. The downside: White keeps space, and in chess the side with space usually has more ways to win.

This product chooses the second path and markets it with an argument that sounds persuasive: "computers often do not know how to handle complex strategic positions".

Based on my experience following and analysing games, I have to be blunt: that argument is dated.

It was true before neural-network engines. Back then, machines evaluated positions through search depth, so closed Mar del Plata positions with slow manoeuvring and few forcing moves were genuinely a machine blind spot. But current engine generations handle long-term strategic positions far better than that sentence implies. The boundary the product builds between "calculation" and "strategy" was dissolved long ago.

This is marketing language that has outlived its own technical reality.

What is proven and what is not

A serious opening product should present evidence: win-draw-loss rates for 9...a5 in databases, engine evaluations at key moves, and at least a few sample games at a high level.

This product has none of that.

No engine evaluation. No database statistics. No move-quality metric. No cited games. No named author with results in the Mar del Plata itself.

The claim that "chances are equal" after 9...a5 is a positional judgement, not a proof. It is plausible, but it stands alone, without support.

A tape misloaded early in life is the most expensive lesson: the eye always needs verification. In analysis work I learned that the most dangerous thing is not false information, but true information placed in a false frame. 9...a5 is a real move. The "antidote" is a frame.

For a club player the consequence is concrete. You buy a product believing you will no longer have to calculate much. You get a closed Mar del Plata position where every move still must be calculated, only differently. Expectation and reality diverge on page one.

The blind spot lies elsewhere

Stopping at whether 9...a5 is strong misses the larger question.

The larger question is: why must the opening market keep producing "antidotes"?

The transfer market resembles an endgame: whoever reads the role first breathes easily. The opening-theory market works the same way, except what must be read first is not a player's role but a learner's fear.

The club player's fear is specific: fear of being out-prepared. Fear that the opponent checked a database while you did not. Fear that you are playing chess from memory, and someone always has a longer memory.

A product selling "few concrete lines" is not selling theory. It is selling relief. That is why it works commercially, whatever the real technical value of its lines.

Here is the counter-intuitive point: precisely because 9...a5 has no forcing lines, it is harder to train. You cannot memorise your way through. You can only understand your way through. For most buyers of opening products — people trying to reduce memorisation load — that is a higher demand, not a lower one.

The biggest risk of this branch is not being refuted. It is the player becoming complacent.

In opening chess, every secondary branch has a lifespan. Once enough people play it and enough games reach databases, White will respond. That is the law of a theory cycle. Black is one tempo ahead in the silence, until the silence ends.

What to watch

For a serious player, verifying 9...a5 does not mean rereading the product. It means watching three observable indicators.

First, the frequency of 9...a5 in top-level classical games. If it appears in major events at 2700+, that is a real signal. If it lives only in club games and commercial products, its reference value is limited.

Second, the frequency of the Bayonet Attack 9.b4. A defensive branch only matters as long as White keeps playing 9.b4. If the trend shifts elsewhere, the whole discussion closes itself.

Third, the White response. Once concrete counter-lines are published, pressure returns to Black. In opening chess, every small advantage has an expiry date.

An empty stadium turned out to be the most honest mirror of modern football. A chessboard has the same kind of silence: no commentary, no noise of opinion, only the bare structure. After everything, what remains is the position, not the advertisement.

Conclusion

9...a5 is a sound move, a real branch, a reasonable choice for anyone wanting to leave forcing lines behind. But "less theory" does not mean "less thinking".

The question left for the next game is not whether to choose 9...Nh5 or 9...a5. The question is: sitting in front of a position with no line to memorise, what do you have left?