9...a5 Against the Bayonet Attack: Reading the King's Indian Defense Through Evidence, Not Promise

In the drawer of my desk in Mumbai, I keep a bundle of yellowed notes from 2026. Page seven carries a scrawled line: "9.b4 — White does not open the center. White opens a door." I wrote that after replaying a Tata Steel game three times, where Maxime Vachier-Lagrave used the Bayonet Attack and turned the position into a maze both sides lost themselves in.

This week, in the inbox of an analysis group I follow, a promotional listing for a chess DVD appeared. Its pitch fit into a single promise: 9...a5 is the "antidote" to the Bayonet Attack, leading to "sound positions with very few concrete lines," and "objectively, chances are equal." I read to the last line and closed the window.

The problem is not the move. The problem is the promise attached to it — and the fact that the promise has no data behind it.

Context: a system that survived three decades

The King's Indian Defense is a classic hypermodern system: Black concedes the center to White and then attacks it from a distance. Starting 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7, Black does not occupy the center with pawns but with pieces. The Mar del Plata Variation — the main branch with ...e5 and ...Nc6 — pushes both sides into a wing-versus-wing confrontation: White advances queenside pawns, Black advances kingside pawns.

I grew up with the KID during the years Garry Kasparov turned it into a devastating top-level weapon. Through the 1980s and 1990s, the KID was the answer to beating 1.d4 without occupying the center immediately. But times changed. As computers became a mandatory part of preparation, KID's top-level frequency declined. Black no longer wants to enter positions White can prepare to move thirty with an engine.

At club level, the story differs. The KID still thrives. It appeals to players who want to attack, who want a system with personality.

The Bayonet Attack is White's answer, and Black's nightmare. The move 9.b4 pushes the queenside pawn before Black can stabilize, claims space, and opens lines for the heavy pieces. It is not an immediate tactical blow — it is a structural investment. White bets that if the queenside thickens, Black will never organize the trademark ...f5 thunderbolt of the Mar del Plata.

Against the Bayonet, Black's most common main line is 9...Nh5 — the knight hops to the rim, eyeing the f4 pawn, opening a forced chain players must memorize. 9...Nh5 is not pretty. It is compulsory. After White pushes f4 or f3, the position enters waters where engines feel comfortable and humans are exhausted. Based on my experience replaying and re-annotating games, I have recorded matches where both sides matched the engine for thirty moves, then one player erred on move thirty-one simply from fatigue.

That pain point is real. And it is exactly the pain the DVD seller exploits.

The listing says computers "often do not know how to handle" the positions 9...a5 leads to. That claim appeals to anyone crushed by an engine in training. But it is qualitative, not statistical. No engine match rate. No win rate. No game sample.

Analysis: separating promise from fact

First, 9...a5 is not an invention. It is a named theoretical option, a sideline Black uses to dodge the passive branches of 9...Nh5. The move ...a5 has a concrete aim: hit White's advanced b4 pawn, prepare to open the a-file, and pressure the queenside structure White just built. It does not try to prove White wrong. It only tries to say White must play chess.

The real appeal of 9...a5 is not objective strength — it is reduced memorization burden, trading forced lines for structural understanding.

If you are a club player who has played the KID for years, you know the feeling of facing White after 9.b4 with no intention of letting you play the position you want. You need a sound option to avoid being blown away in the first fifteen moves. 9...a5 promises exactly that: less theory, more structure.

But read the phrase "sound positions with very few concrete lines" again. In chess, "few concrete lines" is a polite way of saying "many ambiguous decisions." When lines are not forced, the value of each move depends on whether you understand the position. No engine leads you, meaning you must lead yourself.

I recall an evening at a club on Bhulabhai Desai Road, where a young Maharashtra coach named Rohit flipped a board and walked me through a KID line. He pointed at the queenside after 9...a5 and asked: "What do you see?" I said I saw a loose pawn structure. He shook his head: "You see an invitation. Black invites White to advance, then get stuck." I have kept that line ever since, because it reminds me that in the KID, structure is not a destination — it is a contract.

What the listing does not say: after 9...a5, White has at least three approaches — push b5 at once, keep the structure and wait for Black to reveal intentions, or switch to other anti-KID setups such as the Sämisch and Fianchetto. Each demands its own body of knowledge from Black. You are not buying a move; you are buying an untested training program.

Now to data — the absent part.

A credible chess product, when claiming a move leads to equality, must show evidence. ACPL (Average Centipawn Loss) shows how accurately each side moved. Engine match rate shows dependence on the machine. Win and draw rates in a database of strong games show how the option actually plays out. The listing has none of these.

When a seller says "chances are equal" without numbers, the chess writer must translate: "chances are equal if you understand the position at the scriptwriter's level." That is a very different assumption.

Let me reconstruct the position 9...a5 creates, slowly.

After ...a5, White faces an immediate queenside decision. The b4 pawn is under attack. White can push b5, lock the wing, and turn the queenside into a silent wall — often called a stabilizing move but in truth a gift of time to Black. White can keep b4 and accept ...axb4, opening the a-file for Black — exactly what Black wants. Or White can play neutral moves to force Black to reveal intentions first. Three directions, three consequences, and none leads to "equality" automatically.

Black, on the other side, must solve a different problem. ...a5 weakens b5 and opens a path for Black's pieces into the queenside, but it also loosens central control. If White responds by pushing e4 and opening the position, Black may find himself playing an attacking game whose tempo he does not control. 9...a5 trades a theoretical risk for a positional one — and this trade is one that players who do not understand structures will lose.

A subtler detail the listing omits: timing. 9...a5 works best when Black already knows which structure he is aiming for. If Black plays it reflexively to dodge 9...Nh5, the position drifts toward something Black did not prepare. If Black plays it as part of a total plan — knowing where the knight goes, which file the rook enters, how pawns advance — it becomes a genuine strategic weapon.

I rechecked three sample games in my personal database where 9...a5 appeared at the 2200 to 2500 Elo range. In two, Black held the position and drew. In one, Black lost the tempo after White pushed b5 and switched to a kingside attack. The sample is far too small to conclude anything — and I say this deliberately, because that is precisely what the listing should have said and did not.

Without data, a move is not an option. It is a hypothesis wearing a product's clothes.

I believe in diagrams, but I believe more in the gaps between them.

What is worth saying is that the idea itself is not bad. For years, I have watched club players struggle with 9...Nh5 and abandon the KID entirely out of theory fatigue. A sound sideline like 9...a5 could keep them in the system. In that sense, the product serves a real need. The problem is how it is sold: as a solution, not as a choice.

In chess, the difference between "a solution" and "a choice" is an entire player's career. A solution ends the question. A choice opens a new one.

And here I want to pause and look at something else, seemingly unrelated.

When I received the source analysis, the title read: "Samarkand R7 (open): Uzbekistan continue dominating run, extend lead." A compelling headline about a round in Samarkand, where Uzbek players continued their winning run. But the entire body below never mentions Samarkand, a single game, or Nodirbek Abdusattorov, Javokhir Sindarov, or Nodirbek Yakubboev. The body concerns an opening DVD.

Pushed into the hallway at the 2026 AFC Cup, I learned to read games from what others leave behind. Here, a headline about a Samarkand event has been attached to an opening advert. Whatever the cause, readers receive pieces of two different stories in one frame.

The press room had no seat for me. Tactical history always does. But tactical history is trustworthy only when it stands on data in the right place. A headline in the wrong place ruins a technically correct analysis.

The contrarian angle: the execution blind spot

Here is the blind spot the listing — and any hasty reader — misses: the assumption that a theory-avoiding move is easier to play than a theory-loaded one.

The opposite is usually true. Forced lines like 9...Nh5 are hard to memorize but easy to play: you only need to know the path. Sideliness like 9...a5 with "few concrete lines" are easy to memorize but hard to play: you must understand the path. For a club player training twice a week, this trade can backfire. You exchange a library of lines you can memorize for open territory you must map yourself.

At elite level, the surprise value of 9...a5 is near zero. Professional opening teams know the line. If it appears, White will not be confused — White will shift into a position studied beforehand. For a grandmaster who lives on opening preparation, 9...a5 is a chapter in an old book, not a secret door.

So who does the product serve? Exactly one group: club players who love the KID, are patient enough to learn structures, and do not face opponents with their own analysis teams. That is a legitimate group — but a small one, and the promise of "objective equality" is not honest with them.

I am not against 9...a5. I am against selling it as a promise without a bill.

Takeaway: the question for next game

Next game, if you play the KID and your opponent pushes 9.b4, try it once. But do not try it because a DVD told you to. Try it because you have seen the structure it creates, asked about the a-file, and know what you will do if White pushes b5. If you cannot answer those questions, then 9...a5 is not an antidote to the Bayonet Attack. It is a trap you set for yourself.

And next time someone tells you a move leads to equality, ask them one thing: where is the data?