9...a5 Against the Bayonet: An Opening Answer With No Statistics Attached

Twenty-two moves. That was the entire baggage of a 16-year-old sitting across from me at an open tournament in Guangzhou last summer: a Bayonet line memorised to move 22, and the belief that Black would answer 9...Nh5 as every book he had read said. Black played 9...a5. The boy spent 41 minutes on move 23, then lost in a position the engine scored as equal.

I bring it up because this week the opening world received a product that lands exactly in that gap: a lecture library for Black, teaching how to meet the Bayonet with 9...a5. The promotional page calls it a “clear case” — a clean answer, few lines, heavy on strategy, light on memorisation. To a King’s Indian player tired of recalling eighteen moves of a Mar del Plata line, that greeting sounds very pleasant.

The first thing I did was open the data section. No win rate. No draw rate. No ACPL, no game sample, no engine match rate. Only words.

Let us set the board first. The King’s Indian is a hypermodern system: Black concedes the centre and the space, then strikes back with ...e5 and ...c5. Mar del Plata is the sharpest branch of that system, born at the tournament of the same name in 2026, defined by ...Nc6 after White builds the Classical setup. The Bayonet is White’s most ambitious reply: 9.b4. Instead of letting Black choose the plan freely, White pushes the queenside first, threatens b5-b6, prepares c4-c5 to clamp the centre, and strips Black of key squares. In ECO classification the whole complex sits between E97 and E99.

Black’s main move against the Bayonet is 9...Nh5, heading for ...f5, opening the kingside and accepting a race. That is the theoretical highway: thousands of games, hundreds of analyses, and enough forced variations to discourage anyone above 1800.

9...a5 is different. It strikes directly at the advanced b4-pawn, grabs queenside space, prepares ...Nc5 or ...a4, and turns the question of the game from “who calculates faster” into “who understands the structure better”. That trade is exactly what the product is selling.

I came to chess from the board, not the keyboard. In 2026 I was still sitting in the organiser’s and arbiter’s chair before writing my first analytical line. That job taught me something that still holds: people describe moves through feeling, but the move itself lives in the pawn structure.

Based on my experience following games at Asian open tournaments, the Bayonet is the most common trap club-level King’s Indian players run into — not because it is the strongest, but because it is the easiest to prepare. One evening with an engine is enough for a 2026 player to carry twenty moves into the game.

The structure of 9...a5 fits in one sentence: Black accepts a weakened b6 square in exchange for an open a-file and a frozen b4-pawn. After ...a5, White must decide immediately. Push b5 and c5 slips out of control. Stand still and Black has ...axb4 followed by ...Nc5 aimed at d5. It is pressure that does not kill at once but squeezes — the kind the engine scores 0.00 while humans score “uncomfortable”.

The product describes the move with five claims: the positions are sound, there are very few concrete lines, the focus is strategy rather than tactics, computers do not know how to handle these positions, and if Black knows what to do, White can get into danger quickly.

Four of those five can be tested. The fourth cannot, and it is the only purely promotional claim in the set.

My three-source discipline is explicit: before believing a claim about an opening, I need three independent sources — actual games, engine assessment, and expert opinion with no commercial stake. With 9...a5, all three roads lead to the same place: the seller. I counted three, but they are one. That is the trap I fell into once, and it has been on my checklist since the xG piece in 2026.

People call that a shock; I call it unread data. The move 9...a5 does not stun anyone because it is new — it stuns because it is absent from the preparation file. And a move absent from the preparation file is always stronger than its true value, for exactly one game.

The fourth claim deserves its own paragraph. “Computers don’t understand” is the standard line of every opening product sold to players who want to believe human intuition still beats silicon. Yet the same author, elsewhere, concedes that after 9...Nh5 “computer analysis becomes important again”. Both sentences sit in the same document. If the engine evaluates the main line well, it evaluates the sideline well too. The difference is not that the machine fails to understand; it is that the machine scores both the same. When both are equal, the only advantage left is an information advantage.

That is where I place my analytical bet: the real value of 9...a5 is not that it is stronger than 9...Nh5, but that it is less prepared against — and that advantage has an expiry date.

The second claim, “very few concrete lines”, is the most honest one and the most misread. Low forcing density means what, exactly? It means that after move 12 both sides have more than one reasonable move. For a player strong on structure, that is paradise. For a memoriser, it is a desert. A line with few branches is not easier; it is difficult in a different place.

I tested this on myself years ago, moving from learning lines to learning plans. The first sensation was loss: no correct move left to look up. The second, after about twenty games, was freedom. Those twenty games were the price, and no lecture library includes it in the purchase price.

The only quantitative word in the whole document is “equal”. The author concedes that objectively the chances are level. That is the kind of credible concession which shows the writer is not selling miracles. It also draws a very clear boundary: this product does not promise an advantage, it promises a playable position. Those two things are far apart, and buyers routinely read the first as the second.

The fifth claim is the most technically interesting. “If Black knows what to do, White can get into danger quickly” is true — but it is true of every opening line. What it actually says is that the product does not run itself. It needs a user who understands the plans, not one who remembers the moves.

On time control, this is the weakness least discussed. Strategic plans need a slow rhythm of thought. In rapid and blitz, where a move gets three seconds, a position that demands understanding is punished faster than one that demands calculation. A buyer should keep 9...Nh5 as the fast-format weapon and use 9...a5 for long games. That is the correct use, not the only use the marketing suggests.

On cyclicality: the practical value of a sideline depends on the popularity of the main line. When the Bayonet is in fashion at major events, 9...a5 has value. When the Bayonet cools, it becomes a line in a notebook. Opening theory is where the meta vanishes before the data can be printed in a book — and that holds for both sides of the board.

On the market, there is a line I keep using about sports economics: the transfer market is the only place where people pay for unverified numbers. The opening market is close, except here people pay for unverified lines, and the seller writes the statistics table.

Numbers are asceticism: you have to give up comfort before you can see the truth. With 9...a5 the truth is simpler than the advertising: this is a legitimate, playable sideline, and nobody has proved it equal with independent data.

The counterintuitive angle sits here: “equal” is not a gift to the weaker side.

An equal position is one where the better player wins. If 9...a5 really leads to equality with no structural edge for either side, then the biggest beneficiary is the player with better positional skill — which is to say, the player who does not need to buy a lecture library. The buyer, by definition, is someone trying to close a gap with packaged knowledge. The product sells them a position where packaged knowledge does not help much.

That is the paradox of the entire “understanding over memorisation” segment. It is real, it is useful, and it cancels out its own customers.

The second angle: the risk is not that 9...a5 is bad. It is that the “low theory” label conceals a “high understanding” requirement. Players buy it to save time, then discover that using it costs more time than they saved. Nothing is technically wrong here — only the expectation is wrong.

The third angle, and this is where I argue with myself: if 9...a5 is simply a legitimate sideline, having no independent statistics is not its fault. Every sideline has few games. The problem is not missing data; it is missing data alongside very confident claims. That deficit is not in the move. It is in the way the move is sold.

Three signals I will track over the next six months all sit outside the seller’s control. The frequency of the Bayonet at open events and national championships decides the practical value of 9...a5: if it rises, the product is worth more than its asking price. An engine project showing a clear White edge would collapse the whole argument in a single update. And independent win-draw statistics on an open database will tell us whether “equal” is a concession or just politeness.

Club-level King’s Indian players should buy this product, study the plans section, then check it against an open database before carrying it into a rated game. That is how you use a packaged tool without being used by it.

And the 16-year-old in Guangzhou — if he still plays the Bayonet, I hope he learns one more thing: his opponent is reading the advertising too. The opening race does not stop at move 22; it just moves somewhere else, where no book points the way, and there the player who understands structure always arrives first.