📖 Opening Theory

♞ King's Gambit — as White

1.e4 e5 2.f4 (ECO C30–C39). White offers the f-pawn to pull Black's e-pawn away from the center, open the f-file for a rook once castled, and aim straight at f7 — the square only Black's king defends. It's the most aggressive way to meet 1...e5, at the cost of White's own king safety and a permanent pawn deficit if the attack doesn't land.

Move order & the major branches

After 2.f4, Black's two main replies are accepting the pawn (2...exf4, the King's Gambit Accepted) or declining it. Each branch below is collapsed by default — click to expand a line.

King's Gambit Accepted — 2...exf4

King's Gambit Declined

A note on the game citations above: a specific, verified historical game is linked for the Classical/Kieseritzky, Muzio, Bishop's Gambit, and Falkbeer lines. The Quaade, Fischer Defense, Schallopp, Becker/Modern/Cunningham, and Classical Declined branches don't have an equally famous named game attached — rather than link something unverified, those sections stick to the trap/theory content above.

Middlegame plans for White

Across most of the Accepted lines, White's compensation for the pawn rests on the same handful of ideas:

• The open f-file. Once castled, a rook on f1 stares straight down at f7 — the one square only Black's king defends in the opening. Much of King's Gambit theory is about forcing Black to either weaken that square further or fall behind in development trying to hold it.
• The central pawn duo. A follow-up d2-d4 (often paired with recapturing the gambit pawn via Bxf4) gives White a full e4/d4 center and lasting space, even in lines where the direct kingside attack doesn't land.
• Development speed over material. Nf3, Bc4, O-O, and sometimes d4 all come before White worries about the missing pawn — the whole opening is a bet that piece activity outweighs it.
• The tradeoff: king safety. f2 is permanently weaker after 2.f4, and ...Qh4+ (or ...Bh4+/...Be7-h4+ in some lines) is a recurring resource for Black that can cost White castling rights. Most named traps above exploit White's willingness to accept this risk — and Black's failure to.

Against the Declined lines (Falkbeer, Classical Declined), these gambit-compensation themes don't apply the same way since Black hasn't taken the pawn — plans there look more like ordinary open-game development, with White's main technical task being to neutralize Black's active bishop/central pawn rather than to attack f7.

Endgame themes

The King's Gambit is built around a middlegame attack, not a favorable endgame — there isn't a body of King's Gambit "endgame theory" the way there is for, say, a Ruy Lopez structure. If the direct attack is defused and material stays level or White remains a pawn down, the compensation that carries into an endgame is mostly the same positional edge as the middlegame: better development, a more active king (once safely tucked away), and better outposts for the knights, particularly one that's already reached e5 or g5 in the Kieseritzky/Muzio-family lines. In practice, a King's Gambit player who hasn't converted the attack by the middlegame is usually playing to hold a slightly worse but resourceful position, not to win a technical endgame.

♟ Ruy Lopez, Schliemann Defense — as Black

1.e4 e5 2.Nf3 Nc6 3.Bb5 f5 (ECO C63). Rather than meet the Ruy Lopez with a quiet main line, Black immediately counter-attacks the e4-pawn, accepting long-term risk to the kingside pawn structure and king safety for early piece activity and open lines. It's the most combative reply to 3.Bb5 — "positionally risky, especially against a strong and prepared opponent," in Wikipedia's phrasing, but a real practical weapon: Teimour Radjabov used it against the world's elite from 2008–11, scoring +1 −0 =10 against opponents including Carlsen, Anand, Topalov and Karjakin.

A naming footnote: English-language sources call this the Schliemann Defense after 19th-century player Adolf Schliemann, but the opening was actually first analyzed by the Finnish-Russian theorist Carl Jaenisch in 1847 — Schliemann's own published analysis was of a different line entirely (3...Bc5 4.c3 f5). Russian and German sources correctly call it the Jaenisch Gambit; the misattribution just stuck in English.

Move order & the major branches

After 3...f5, White's four main tries are 4.Nc3 (sharpest, requires the most preparation), 4.d3 (quieter, now White's most popular choice), 4.Bxc6 (the Exchange Variation), and 4.exf5 (grabbing the pawn, generally considered White's weakest try). Each branch below is collapsed by default — click to expand a line.

A note on the game citations above: verified games are linked for the 4.Nc3 lines (Pillsbury–Tarrasch 1903, Bagirov–Kholmov 1961) and 4.d3 (Adams–Radjabov 2011). The Exchange Variation (4.Bxc6) and Gambit Accepted (4.exf5) branches don't have an equally famous named game attached — rather than link something unverified, those sections stick to the theory and trap content above.

Middlegame plans for Black

Across the Accepted-pawn-back lines, Black's compensation rests on the same handful of ideas:

• The open f-file and e-file. Once the center opens up, Black's rooks get active quickly, often before White's own pieces are fully coordinated.
• The bishop pair. Most main lines see White trade a bishop for a knight (via Bxc6) — Black's two bishops, once developed, are a long-term asset even in structurally messy positions.
• Central and kingside activity over material. Black is frequently a pawn down for real, structural reasons — the whole opening is a bet that piece activity and White's slightly loosened kingside outweigh it, not an accident to be corrected.
• The tradeoff: king safety and structure. Black's own king often can't castle short comfortably and the kingside pawns get damaged in several lines (the g6/h-pawn structures above). Black is trading long-term structural soundness for short-term activity — a bet that has to pay off before the middlegame ends, or the weaknesses start to matter more than the activity.

Endgame themes

Like the King's Gambit, the Schliemann is fundamentally a middlegame weapon, not an endgame-favoring structure — Black's whole strategic bet is to generate enough activity that the game is decided well before an endgame is reached. When it does reach an endgame (as several of the games above do, including two long ones), the deciding factor is usually whichever structural damage has proven more costly: White's slightly exposed king and awkward piece placement from the early complications, versus Black's weakened kingside pawns and the pawn deficit in lines where material was never won back. Both games above that Black won were decided in long rook-and-minor-piece endgames, not quick kingside attacks — a reminder that even in a sharp opening built around activity, technique in the resulting endgame still matters.

♘ Caro-Kann, Two Knights Attack — as White

1.e4 c6 2.Nc3 d5 3.Nf3 (ECO B10–B11). Rather than the main-line 2.d4, White develops both knights before committing the d-pawn at all — a flexible, less-theoretical way to meet the Caro-Kann that keeps options open between d3 and d4 depending on how Black sets up. It was a favorite of a young Bobby Fischer before falling out of top-level fashion; a well-known trap buried in one of its main tries has caught two former World Champions.

Move order & the major branches

After 3.Nf3, Black's most common tries are 3...Bg4 (pinning the knight immediately), 3...dxe4 (resolving the center at once), 3...Nf6, and 3...d4 (chasing the knight away). Each branch below is collapsed by default — click to expand a line.

A note on the game citations above: verified games are linked for the Mindeno Variation's two lines, the classical-transposition trap (twice, since it caught two different World Champions the same way), and the 3...d4 line. The solid 3...dxe4 4.Nxe4 Nf6 continuation and the 3...Nf6 4.e5 sideline don't have an equally famous named game attached — rather than link something unverified, those sections stick to the theory and trap content above.

Middlegame plans for White

• Flexible d-pawn timing. The whole point of delaying d4 is choosing between d3 (solid, keeps options open, matches the modern main lines) and d4 (more space, more theory) based on what Black actually plays — there's no single "right" central structure the way there is in the main-line Caro-Kann.
• Punishing the bishop trade. When Black trades on f3 with ...Bxf3, White's doubled or shifted pawns are a long-term consideration, but the bishop pair and the extra tempo the recapturing queen gains are usually worth more in practice, especially with quick central and kingside expansion.
• The kingside pawn storm. h4-g4(-g5) ideas recur constantly — in the Retreat Line to trap the bishop on h5, and in the classical-transposition lines to hunt down the bishop on g6/h7 before Black can castle. Several of the sharpest lines above are built entirely around this plan.
• Development speed. Two knights out before the bishop or the d-pawn is the whole identity of the system — White is consistently willing to spend a move on a slightly unusual recapture (9.fxg3!, 6.Qxe4) to keep pieces active rather than worrying about structural purity.

Endgame themes

As with the King's Gambit and the Schliemann, there's no dedicated "Two Knights Attack endgame theory" — the system doesn't steer toward a particular favorable endgame structure the way an isolated-queen's-pawn or minority-attack opening might. What it does produce, repeatedly in the games above, is long technical fights: both Fischer–Petrosian (68 moves) and Keres–Tartakower (50 moves) were decided well into deep middlegame-to-endgame territory, not by an early tactical knockout. The trap line is the one clear exception — when it lands, it's over in 11-12 moves. Outside of that specific trap, White's practical winning chances come from the same long-game factors as any other Caro-Kann structure: whichever side's minor pieces end up better placed once the position simplifies.

♙ Caro-Kann, Accelerated Panov Attack — as White

1.e4 c6 2.c4 (ECO B10). White plays c4 on move two, before touching the d-pawn at all — a move-order trick aimed straight at the heart of the Caro-Kann. The threat is simple: if Black plays the natural 2...d5, White meets it with 3.exd5 cxd5 4.d4, transposing directly into the well-known Panov-Botvinnik Attack (normally reached via 1.e4 c6 2.d4 d5 3.exd5 cxd5 4.c4) — but a move faster, and with an extra independent try available along the way that the main-line move order doesn't offer.

Move order & the major branches

After 2.c4, Black's replies split into three very different categories: playing 2...d5 as normal (by far the most common, but branching further depending on how Black recaptures), the rare 2...e5 counter-strike, or simply transposing into an ordinary Caro-Kann with a quieter second move. Each branch below is collapsed by default — click to expand a line.

A note on the game citations above: verified games are linked for the Modern Variation (twice, showing both the 5.Nc3 and 5.Qa4+ move orders). The transposition line simply inherits the vast body of standard Panov-Botvinnik Attack games rather than needing its own separate citation, and the pseudo-Scandinavian trap and Open Variation don't have an equally famous named game attached — rather than link something unverified, those sections stick to the theory and trap content above.

Middlegame plans for White

• Choosing your structure, not just your opening. The whole point of 2.c4 is that White decides, move by move, whether to steer toward the well-trodden IQP middlegames of the standard Panov (rich in theory, well understood by both sides) or the more independent, less-charted Modern Variation structures — a real practical weapon precisely because Black has to be ready for both.
• The IQP compensation package, when it arises. In the transposition line, the same ideas as any Panov-Botvinnik or Isolated-Queen's-Pawn middlegame apply: piece activity and outposts (especially e5) outweigh the long-term structural weakness, provided White keeps pieces on and doesn't drift into a pure endgame where the isolated pawn becomes a simple target.
• Punishing move-order carelessness. The pseudo-Scandinavian trap is the clearest example, but the whole opening is built around Black needing to recognize that this isn't quite the position their memorized Caro-Kann or Scandinavian theory assumes.
• Development speed over structural symmetry. In the Modern Variation especially, material and structure end up level fast — White's edge, when there is one, comes from being a touch faster to develop and castle, not from a long-term pawn-structure advantage the way the IQP lines provide.

Endgame themes

The two faces of this opening lead to genuinely different endgame outlooks. In the transposition line, the classic Panov-Botvinnik IQP question applies directly: White's isolated d-pawn is a long-term target once queens and enough minor pieces come off, so White generally wants to avoid over-simplifying and keep enough pieces on the board for the pawn's dynamic benefits (space, outposts, open lines) to matter. In the Modern Variation, both games above ended in fairly level, simplified endgames (a drawn rook ending in Vaganian–Dreev, a balanced middlegame-to-endgame draw in Radjabov–Azmaiparashvili) — a fair reflection of a structure that's closer to equal than the IQP lines, where technique and small edges matter more than any built-in structural advantage.

♗ Evans Gambit — as White

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.b4 (ECO C51–C52). White offers a second pawn — on top of the Italian Game's usual central play — to deflect Black's bishop off the c5–f2 diagonal, then races to build a full pawn center with c3 and d4 while the bishop sorts itself out. Named for William Davies Evans, a Welsh sea captain who worked the idea out and introduced it in the game below; it became the single most fashionable weapon of the Romantic 19th century, played constantly by Anderssen, Morphy, and later revived by Mikhail Chigorin against the era's top defenders. Considered "refuted" by some late-19th-century analysis, it has never actually gone away — Garry Kasparov pulled it out to beat Viswanathan Anand in their 1995 World Championship match, over a century after its supposed refutation.

Move order & the major branches

After 4.b4, Black's main choice is accepting the pawn (4...Bxb4), which itself branches heavily depending on where the bishop retreats to on move 5. Declining with 4...Bb6 is also playable, and 4...d5 is a rare, direct counter-punch. Each branch below is collapsed by default — click to expand a line.

Evans Gambit Accepted — 4...Bxb4

Evans Gambit Declined & sidelines

A note on the game citations above: verified games are linked for the Anderssen Defence and Pierce Defense lines (via the Morphy–Anderssen and Evergreen games), the Compromised Defence, and the Anderssen Variation (5...Be7). The Mieses Defence, Lasker Defence/Tartakower Attack, the 5...Bc5 retreat, the Stone-Ware Variation, and both Declined sidelines don't have an equally famous named game attached — rather than link something unverified, those sections stick to the theory and trap/avoid content above.

Middlegame plans for White

• The battery on the long diagonal. Qb3 and Ba3 together are the recurring engine of White's attack across almost every accepted line — Qb3 hits f7 and b7 at once, while Ba3 both prevents Black from ever castling short safely and eyes the a3–f8 diagonal straight through wherever Black's king ends up.
• Development speed over pawn count. White is routinely two, and in the Compromised Defence three, pawns down at some point in these lines — the whole opening is a bet that the resulting lead in development and open lines is worth more, not a miscalculation to be corrected by winning the pawns back quickly.
• The full pawn center. Once c3 and d4 both go in, White often gets e4 and d4 (or e5 and d4) working together, cramping Black's pieces and supporting further advances like e5 that gain even more time.
• Forcing the king to stay in the center. Nearly every named trap and brilliancy above shares the same shape: keep Black from ever castling safely, then convert that alone into a mating attack, even if it costs real material along the way (as in the Evergreen Game, where White's own queen and a rook never leave their starting squares).

Endgame themes

Like the King's Gambit, the Evans Gambit is built to win before an endgame is reached, not to produce a favorable one — White's whole compensation (development, open lines, the long-diagonal battery) tends to matter less the more pieces come off the board. The Lasker Defence is the clearest illustration of this from Black's side: trading queens immediately, as early as move 9, is specifically designed to reach an endgame where White's structural investment (the missing pawn) simply outweighs whatever lingering development edge remains — Fine's "psychologically depressing" comment was about exactly this dynamic. When White's attack does land, it's almost always well before move 25 or so, as in every game cited above; an Evans Gambit that reaches a level middlegame with no attack in sight is usually just a pawn down for White with little to show for it.