The Principle
The dog fight is the canonical no-gi scramble — the knee-to-knee, underhook-to-underhook exchange that emerges when a half guard player comes up to defend the pass and the top player fights to come down. The name captures the position: both players on one knee, chest-to-chest, fighting for the underhook and the angle. The scramble resolves in seconds; whoever wins the angle lands on top.
The dog fight is the single most common scramble in modern no-gi grappling. Every underhook-heavy half guard exchange passes through it; every failed pass against a committed underhook ends in it; every deep half back take attempt that the top player survives produces one. Understanding the dog fight is not optional for a no-gi game — it is the central scramble from which many positional exchanges resolve.
Invariables Expressed
Scramble positions resolve in favour of the player with the prepared next position.
The dog fight is the clearest case of INV-SC01. Both players have the same raw materials — underhook, knee up, hip connection. The player who has drilled the next position (back take from the dog fight, or sprawl-and-pass against the dog fight) wins; the player who is improvising loses.
The dominate-neutralise-capitalise hierarchy resolves scrambles in that order.
The scramble’s correct strategy follows Jones’ hierarchy: first, dominate the angle by getting to the back side; second, neutralise the opponent’s re-entry attempts; third, capitalise by converting the angle into a pin, back take, or submission. Skipping to capitalise before dominating produces positional loss.
Underhooks elevate the opposite shoulder and break the line of base.
The dog fight is an underhook battle. The player who locks the deeper underhook elevates the opponent’s shoulder on that side and wins the angle. The underhook depth decides the scramble — shallow underhooks are whizzered out; deep underhooks rotate the opponent.
A defender cannot defend two threats simultaneously when each defence creates the other’s opening.
From the dog fight, the attacker threatens both the back take (rotation behind) and the sweep (drive through to top half). The opponent’s defence to one opens the other — the scramble contains the half guard back-take vs sweep dilemma embedded within the scramble exchange.
Dominate — Winning the Angle
The dominate phase is the knee-up-and-underhook race. The first player to post up on one knee, hip in, and lock the deep underhook controls the scramble. The underhook depth is decisive — forearm past the opponent’s back, not just under the armpit. Deep underhook = structural elevation of the opponent’s shoulder, which breaks their base and commits their weight forward.
The dominate move is the pummel: underhook-in, whizzer-out, step the knee up, hip in, chest onto the opponent’s chest. The player who completes the pummel first is ahead; the player who completes it second is already losing.
Neutralise — Stopping the Opponent
Once the angle is won, the neutralise phase denies the opponent’s re-entry. The most common re-entry is a whizzer-and-sit-back — the opponent tries to re-establish a seated guard by whizzering the underhook and sitting through. The neutralise response is to drop the hips and deny the sit-through, keeping the opponent’s far hip pinned while the back-take setup develops.
A secondary neutralise is against the opponent’s stand-up attempt. The dog fight’s losing-side player will often try to stand up to reset; the winning side denies the stand-up by hanging on the underhook and keeping their weight on the front-side knee. The opponent cannot stand with the weight still committed.
Capitalise — Converting to Back
The capitalise phase is the back take or sweep conversion. Back take fires when the opponent’s back is exposed during their whizzer defence — the winner spins behind and locks the seatbelt. Sweep-to-top-half fires when the opponent stays square but their base is broken — the winner drives through to top half or mount.
The choice between back take and sweep is not pre-selected. It follows from the opponent’s defence: if they expose the back (whizzering out), take the back; if they stay square (denying the spin), sweep forward. Both finishes are prepared during the dominate and neutralise phases; the capitalise phase simply commits to whichever the opponent has exposed.
Deploying the System
When to enter
The dog fight is entered — willingly or unwillingly — whenever a half-guard bottom player commits to the underhook and the top player commits to the whizzer. Three entry paths. First — bottom-player-initiated: half-guard bottom times the underhook, hips in, and posts up onto the back-side knee; the top player is dragged into the dog fight by the rising underhook. Second — top-player-initiated: top half-guard with a collapsed whizzer, where the top elects to come up into the dog fight rather than accept a flat-back defensive posture. Third — transition-produced: a failed sweep (old-school, knee-lever, John Wayne) whose defence left both players knee-up with matched underhook chases.
The dog fight is the wrong place to be when your underhook is already losing the depth battle — a shallow underhook against a deep opponent’s underhook is a pre-lost position; accept the whizzer hop-back to knee-shield or abandon the dog fight to guard-recovery rather than let the loss deepen. It is also wrong when your base leg is vulnerable to a leg entry: knee-up dog fight exposes the posting leg’s foot, and a leg-entanglement opponent will convert the dog fight into an ashi entry on your base leg.
Live reads inside the system
Four reads. First — whose underhook is deeper? Deeper underhook = structural advantage = about to win the scramble; a forearm past the opponent’s back outranks an armpit-only underhook. Second — whose head is higher? Higher head = longer spine = winning the posture war. Lower head = forehead pressure into chest = committing to a squeeze-style conversion. Track both and adjust. Third — is the opponent cheek-to-cheek or cheek-to-chest? Cheek-to-cheek is the dog fight proper; cheek-to-chest means they have begun the back-take rotation on you — stuff the head immediately or race your own rotation. Fourth — where is the outside (whizzered) foot? If your whizzered foot is free, the back-step exit to north-south is live; if it’s trapped under the opponent’s knee, you are committed to resolving via underhook-fight only.
When the system stalls
The canonical stall is the whizzer-deadlock: both players with matching underhooks and whizzers, neither gaining an angle. The tactical response is the head-position break — fighting for head inside position, since the deadlocked grip battle is won by whoever gets the head under the opponent’s chin. A second stall is the sit-back-to-guard: losing-side player sits their hips back out of the dog fight and returns to seated guard. Do not pursue — recover to pass-initiation posture (headquarters) rather than dive into the seated guard’s gripping game. A third stall is the third-party scramble entry: a dog fight can sometimes produce a leg-entanglement for either player during the knee-fight. The leg entanglement continue vs reset dilemma applies — choose between continuing the dog fight or converting to a leg-entanglement finish based on which is farther along.