The triangle is the submission you finish while you appear to be losing — three limbs and an opponent's own shoulder turned into a noose, thrown from flat on your back. Born in early-20th-century judo groundwork and rebuilt in Rio in the 1970s, it remains the position every guard passer is taught to fear.
On 7 August 2010, at the Oracle Arena in Oakland, Anderson Silva had lost twenty-three of the previous twenty-five minutes. Chael Sonnen, the challenger for Silva's UFC middleweight title, had taken him down in every round, held top position, and thrown enough punches to be comfortably ahead on all three judges' cards heading into the fifth. Silva was on his back, bleeding, doing the one thing a striker on the bottom is not supposed to be able to do to a wrestler: threatening.
With roughly two minutes left, Silva threw a leg over Sonnen's shoulder from the bottom, trapped the far arm across his own centerline, and locked his shin behind his own knee. Sonnen postured, drove, tried to pull the leg off. It was already too late. Silva finished with a triangle — the choke tightening as he pulled the trapped arm across and angled his hips — and reached up for the arm to add a joint lock as insurance. Sonnen tapped with 1:50 remaining. Silva had been losing the fight, on paper, until the instant he won it.
That is the argument the triangle choke makes, and has been making since it left Japan. It is the submission of the person underneath. It works from the position most people mistake for a losing one — flat, on your back, with a heavier opponent stacked on top and passing. And it uses that opponent's own posture against him: the harder he drives forward to pass the guard, the more he feeds his own neck and shoulder into the trap.
This is not a story about one comeback in Oakland. It is a story about a shape — three limbs and a shoulder — that has survived a century of people trying to pass around it.
The triangle choke is grappling's proof that the bottom is not the losing position. It is the only high-percentage submission that gets tighter the harder your opponent tries to escape by passing.
The triangle choke is not a Brazilian invention, and the sport's own folklore has spent decades quietly correcting the people who think it is.
The technique's home is early-twentieth-century Japanese judo, specifically the newaza (ground-fighting) laboratory that grew up alongside — and partly in rebellion against — the Kodokan's standing-throw orthodoxy. Jigoro Kano's judo emphasized tachi-waza, the standing game, at roughly an eighty-twenty split over groundwork. But from 1914, a parallel competition circuit known as kosen judo, contested among Japan's higher preparatory and technical schools, tilted the ratio toward the ground and gave specialists room to invent. It ran through 1943, and the grapplers who lived in that circuit produced much of the early leg-choke and leg-entanglement vocabulary that grappling still uses.
The triangle — sankaku-jime, "triangle strangle" — belongs to that lineage. Its development in the mid-1910s to early 1920s is most often credited to Kodokan groundwork specialists Tsunetane Oda and Kanemitsu Yaichibei, both direct students of Kano. The earliest commonly cited competitive appearance is a kosen judo tournament in Kobe, Hyogo, in November 1921, associated with Kanemitsu and his apprentice. Earlier Japanese names for the shape — matsuba-gatame, sankaku-garami — circulated before the vocabulary settled on sankaku-jime. The exact chain of invention is genuinely murky; what is not in doubt is that the triangle arrived in Brazil as an inherited judo technique, not a local discovery.
That inheritance is where a common BJJ myth needs correcting. The story often told is that the triangle "did not exist" in Brazilian Jiu-Jitsu until Rolls Gracie — the most technically influential Gracie of his generation — found it in an old judo book in the 1970s. Rolls' own black belts have pushed back on the tidy version of that tale: Mário Tallarico, one of Rolls' senior students, has said the triangle was an old judo technique present in Brazilian Jiu-Jitsu from early on. The more accurate account is not discovery but systematization. Through the 1970s, at the Rio de Janeiro academy of Osvaldo Alves — a judo-rooted teacher who cross-pollinated with the Gracie schools and coached a generation of innovators — grapplers began chaining the triangle to the armbar (juji-gatame) from the guard. Pairing the two attacks made each one dramatically harder to defend: posture up to escape the triangle and you feed the armbar; pull the arm out to escape the armbar and you feed the choke. That combination, refined in Rio in the late 1970s and early 1980s, is the moment the triangle became a system in BJJ rather than a technique someone occasionally hit.
The no-gi and MMA eras did the rest. When a relatively unknown brown belt named Eddie Bravo submitted Royler Gracie with a triangle at the 2003 ADCC World Championship — more on that below — the technique acquired a second origin story, this one about the sport's future rather than its past.
Roughly a century after Kobe, the shape is unchanged. What changed is the map of positions it can be thrown from.
A triangle choke is a blood choke — it closes the carotid arteries on both sides of the neck, cutting blood flow to the brain, rather than crushing the windpipe. A competent triangle renders an opponent unconscious in seconds if he does not tap, and it does so with almost no strength from the arms. The compression comes from the legs and, critically, from the opponent's own body.
Here is the defining detail, the one that separates the triangle from every other choke in the sport: one side of the neck is closed by your leg; the other side is closed by the opponent's own shoulder. You trap one of his arms inside the triangle and leave the other outside. The trapped arm's shoulder is jammed against one carotid; your shin and the back of your knee close the other. His own limb is doing half the work. This is why the triangle is a low-effort, high-percentage finish when the geometry is right — and why it does nothing at all when the geometry is wrong.
The classic configuration is sankaku-jime from the guard:
Every triangle in the sport belongs to one of two families, defined by which arm is trapped and how the entry unfolds:
The triangle is also the connective tissue of an entire attack web. It shares its setup with the armbar (the Osvaldo Alves combination), flows to and from the omoplata when the opponent hides the near arm, and — in the no-gi and 10th Planet worlds — lives inside the high-guard systems where the legs are already climbing the torso. A player who has one of these usually has all of them, because they are the same isolation of the same arm from four inches of adjustment apart.
Understanding the mechanics is mostly understanding the failure modes, because a loose triangle is a gift to the opponent — it frees a hand, exposes your own back, and burns your legs. Three things break it: the wrong number of arms trapped, the missing angle, and a defended posture that lets the opponent stack you and stand. All three are addressed below in Common Mistakes.
In 2026, the triangle occupies a strange dual status: it is simultaneously the most taught submission from the bottom and one of the harder ones to land clean against elite guard passers who have spent their careers learning to keep both arms and their posture out of the trap.
At the highest levels of no-gi — ADCC, the Craig Jones Invitational, the WNO circuit — the pure closed-guard front triangle is less visible than it was in the 2000s, for the same reason closed guard itself is used sparingly there: elite passers rarely give you the clean arm-in/arm-out posture the classic entry needs, and the leg-lock era has pulled the bottom player's legs downward toward the hips rather than upward toward the shoulders. But the triangle did not decline; it migrated. Its center of gravity moved to the back (the rear triangle as a finishing structure inside back control), to mounted triangle setups off failed upa escapes, and to the transition — a triangle threatened not as a finish but as a lever to force an arm free for an armbar, or to spin to the back, exactly the way it was designed to chain in Osvaldo Alves' gym half a century ago.
In the gi, and across the vast middle of competition — blue, purple, and brown belt at IBJJF, where most of the sport's matches are actually contested — the front triangle remains one of the highest-percentage finishes from guard, precisely because most passers below the world elite still surrender an arm under pressure. And in MMA, the triangle has never been more valuable: it is the answer to the wrestler who takes you down and postures to ground-and-pound, and the highlight reel of the last two decades — Silva–Sonnen among the most famous — is full of fighters who were losing on the cards until a leg came over a shoulder.
The instructional market tracks this. Triangle-specific content sits inside nearly every major guard and back-attack instructional released in the last decade, and the technique is a fixture of the leg-entanglement-to-strangle systems that define modern no-gi. The triangle did not get replaced by the heel hook. It got networked into the same graph.
A short list of triangles that changed how the technique is understood — not a ranking, and deliberately limited to finishes that are cleanly documented.
Beyond these two, the triangle's fingerprints are everywhere the guard is played — from Ryan Hall building an entire competitive identity around the triangle-and-leg-attack web, to countless IBJJF finals decided by a leg coming over a shoulder in the last minute. The two matches above are the ones nearly everyone in a gym can name.
White belt learns the shape and, mostly, learns to lose it. The lesson at this rank is the arm count: one arm in, one arm out, every time. A white belt who can reliably identify which arm to trap has learned the position's first truth.
Blue belt learns the angle. The realization that a triangle thrown straight up the back does not choke — that you must pivot your hips off the centerline toward the trapped arm — is the single upgrade that turns a blue belt's triangle from a stalling leg-lock into a finish.
Purple belt learns the chains. The triangle stops being a standalone attack and becomes one door in a room: triangle to armbar, triangle to omoplata, triangle to a sweep when the opponent stacks. The purple belt threatens the triangle to open the other attacks, exactly as the technique was systematized to do.
Brown belt learns the entries no one hands you. Mounted triangle off a failed escape, rear triangle from the back, reverse triangle from the top, the triangle finished when the opponent thinks he has already passed. The brown belt hunts the arm-in/arm-out split from positions where the opponent does not know he has offered it.
Black belt finishes tight triangles on people who know exactly what is coming and have world-class posture, by controlling the head and the far arm before the legs ever lock — and, just as often, uses the threat of the triangle to steer a scramble somewhere else entirely.
1. Trapping the wrong number of arms. Both arms inside your legs is not a choke — it is a body triangle the opponent will eventually posture out of. Both arms outside leaves no shoulder to compress the far carotid. Exactly one arm in, one arm out, or you have a leg entanglement, not a strangle.
2. Forgetting the angle. The most common blue-belt error and the reason most "failed" triangles fail. A triangle locked square, with the opponent's spine aimed at your tailbone, leaves a gap on the trapped-arm side and lets him breathe and stack. You must rotate your hips off the centerline toward the trapped arm so your leg cuts across the neck. If your opponent can talk while you have "locked" the triangle, you have not cut the angle.
3. Pulling the head instead of managing posture. Yanking on the head with two hands while the opponent's posture is strong is a losing tug-of-war that opens your elbows and invites a stack pass. Break the posture first — with grips, with the legs, by disrupting his base — then the head control becomes a finishing detail rather than a survival grab.
4. Leaving the free hand a home. If the opponent can plant his free hand on the mat or on your hip and drive, he can stack your shoulders over your own head and either pass or relieve the choke. Kill the post — trap it, off-balance him, or angle away — before you commit to squeezing.
5. Squeezing with the knees apart. The finish is knees pinched, locked leg curling down, trapped arm pulled across. Wide knees and a lazy figure-four turn a fight-ending choke into a leg workout while the opponent works free. If it is not finishing, check the angle and the knees before you assume the opponent is simply tough.
At open mat this week, isolate the entry, not the finish. Start in closed guard against a same-rank partner. One rule: your only goal is to establish the arm-in/arm-out split and get one leg over the shoulder — no locking the figure-four, no squeezing. Your partner's only goal is to keep both arms in or both arms out and to posture. Fifteen minutes. You are drilling the fight for the arm, which is where every triangle is actually won or lost.
Then add the angle. From a pre-locked triangle your partner gives you, do nothing but pivot your hips off the centerline and re-lock across the neck, twenty times per side, feeling the difference between a square lock (loose) and an angled one (tight). Your partner taps or says "tight" when it changes. That contrast — the same lock, four inches of rotation apart, going from nothing to a finish — is the whole technique in one repetition.
Repeat weekly. The triangle rewards the angle more than the squeeze.
The triangle is the flagship attack of the guard and the hub of an entire submission web.
Every one of these is four inches of adjustment from the others. Learn the arm-in/arm-out split and the angle, and you have not learned one submission — you have learned the grammar of an entire attacking system.
Publication date will be set at Sanity ingest. This log will be appended to over the life of the article. Any factual corrections will be dated and specific.
Publication draft prepared 2026-08-04. Awaiting hero image (needs permissioned photo — placeholder pre-launch). Two-pass edit not yet complete. Source-check pass required before status: scheduled in Sanity.
Known open items: