Mat Strength

What people mean by it

The phrase describes a mismatch. An athlete whose gym numbers predict one thing and whose grappling predicts another — someone you out-lift comfortably and cannot pass. It is used as praise, and occasionally as an excuse, and almost never as a claim anyone expects to be checked.

It is worth checking, because if the mismatch is real it means the gym is measuring the wrong thing, and that has consequences for how you train. The honest answer from the research is that the mismatch is real, it is narrower than the folklore suggests, and the quality underneath it has a name that is less romantic than mat strength: it is mostly strength endurance, expressed through the arms and hands, held for a long time.

The squat does not separate them

Start with the claim that sounds most like an excuse, because it is the one with the cleanest evidence behind it.

Marinho and colleagues compared eighteen BJJ brown and black belts — eight classified elite, ten non-elite — across body composition and a battery of physical tests [1]. Absolute bench press separated the groups clearly (111 ± 6 kg vs 98 ± 6 kg, p < 0.001). Then the pattern breaks:

• Squat 1RM: 91 ± 8 kg vs 88 ± 7 kg — p = 0.410, not significant
• Squat relative to body weight: 1.20 vs 1.23 × BW — p = 0.700, and numerically lower in the elite group
• Bench relative to body weight: 1.46 vs 1.36 × BW — p = 0.096, not significant

What did separate them: one-minute sit-ups (46 ± 4 vs 40 ± 3, p = 0.007, d = 1.70), one-minute push-ups (41 ± 3 vs 36 ± 3, p = 0.018, d = 1.67), body fat (11.2% vs 14.8%, p = 0.049) — and, with the largest effect in the entire study, flexibility (sit-and-reach 40 ± 3 vs 32 ± 3 cm, p < 0.001, d = 2.67).

Read that list again. In a comparison of elite and non-elite grapplers, the biggest measured difference was how far they could reach past their toes, and the squat did not register at all.

Now the caveat, because it matters. Eight athletes versus ten is a small study, and a small study is bad at proving absences. A p-value of 0.096 with an effect size of 0.40 means "we did not detect this," not "this does not exist." A meta-analysis pooling the available comparisons did report an overall strength advantage for elite athletes — but it pooled only three studies, rated just 5% of its own evidence as high certainty on GRADE, and its funnel plot was asymmetric. Its back squat row was not significant either. The correct summary is not "squatting is useless." It is: once everyone in the comparison is already a trained grappler, maximal squat strength stops being what tells them apart.

Grip: the most misread quality

Ask any grappler what mat strength is made of and grip comes up first. They are right about the location and wrong about the quality, and there is a measurement that shows exactly where the mistake is.

Andreato and colleagues had ten advanced athletes fight matches of 2, 5, 8 and 10 minutes, testing them before and after each [2]. After the ten-minute match:

• Maximal grip force was essentially intact. Handgrip dynamometry went from 49 ± 6 kgf to 47 ± 10 kgf — not significant.
• Grip endurance had collapsed. Time hanging from a gi fell from 30 ± 13 s to 21 ± 8 s (p = 0.002, η² = 0.38) — roughly a 30% loss.
• Countermovement jump did not decline at all. Athletes localised their fatigue to the fingers and forearm.

A squeeze test at the end of a hard round tells you almost nothing, because the capacity to produce one maximal contraction is the last thing to go. What leaves is the ability to produce a submaximal one repeatedly.

The discrimination data points the same way. Maximal handgrip did not differ between the winners and losers of actual BJJ matches [3], and in elite judoka it was unrelated to Special Judo Fitness Test classification (r = −0.18, p = 0.47) [9]. Meanwhile the gi grip endurance tests separate levels sharply: 56 ± 10 s versus 37 ± 11 s for the static hang, and 15 ± 4 versus 8 ± 3 gi pull-ups, between elite and non-elite [4]. Across four groups spanning beginner to advanced, those tests produced effect sizes of η² = 0.747 and η² = 0.627 [5].

One detail in that second study deserves more attention than it gets. The tests separated competitors from non-competitors cleanly — but they did not separate advanced athletes from non-advanced ones (18 ± 3 vs 17 ± 3 reps). The gradient is competitive exposure, not eliteness. Grip endurance appears to be a threshold you cross by competing, not a ladder you keep climbing.

What "grip" actually is

Here the research says something genuinely unexpected. Follmer and colleagues tested fifteen BJJ athletes on the gi grip tests and on isokinetic dynamometry, to find out what those tests are actually loading [6]. The strongest correlations were not with anything in the hand:

• Gi pull-ups vs eccentric elbow flexor peak torque: r = 0.82 (p < 0.001)
• Static gi hang vs eccentric elbow flexor peak torque: r = 0.81 (p < 0.001)
• Gi pull-ups vs elbow extensor torque: r = 0.82 (p < 0.001)

What the gi tests measure is substantially the capacity to hold a bent arm under load and resist it being straightened. That is not a hand quality. It is the same action as keeping a sleeve grip while someone strips it, or holding an underhook while they drive away — and it is eccentric, which is to say it is the strength of losing slowly.

This reframes the fatigue athletes report. The forearm is where it is felt; the elbow flexors may be where it is happening. Nobody has separated the two during live grappling, so treat that as the most plausible reading rather than a settled one.

The isometric question

Grappling is full of positions held rather than moved through, which makes isometric strength an obvious candidate for the missing quality. The evidence is suggestive and genuinely contradictory, and you should know both halves.

For it: Coswig and colleagues tested both an isometric and a dynamic gi chin-up against what athletes actually did in simulated matches [7]. In the BJJ group, the isometric test was the stronger predictor of effective actions (β = 0.62, p = 0.002) and of effectiveness as a percentage (β = 0.67, p = 0.02). The dynamic test added little beyond it.

Against it: in judo, the isometric gi hang failed to separate elite from non-elite athletes (35 ± 18 s vs 39 ± 14 s, not significant) while the dynamic test succeeded (12 ± 5 vs 9 ± 4 reps, p < 0.05) [8]. That is the opposite result in a closely related sport.

Both studies are small, they use different populations and different outcomes, and neither measures competition results. The defensible position is that isometric capacity looks important in BJJ specifically and the evidence is one study deep. Nobody has ever directly pitted isometric strength against 1RM as competing predictors of who wins.

What wrestling data adds

The largest study of its kind in any grappling sport looked at 168 elite wrestlers, 85 of them medallists and 83 finishing fifth to eighth, and ranked 39 variables by how well they classified the two groups [10]. It reached 78.6% accuracy. The top two physical qualities:

• Upper-limb peak power (arm-crank Wingate)
• Strength endurance (pull-ups), ω² = 0.142

Maximal strength was not measured at all. The two qualities that rose to the top are upper-body repeated-effort power and upper-body strength endurance — the same family the BJJ grip findings point to, arrived at independently in a different sport with a sample sixteen times larger.

One honest complication from that paper, which its authors report and most summaries drop: the two technical variables in the model explained more variance than any physical one (ω² = 0.183 and 0.161, versus 0.142 for pull-ups). The physical qualities ranked highest on one importance metric; skill still explained more. That is worth holding onto before anyone decides mat strength is the whole story.

Your weight class changes the answer

Any statement about grappling strength has to specify whether it means absolute or relative, because across weight classes those two rank athletes in opposite directions.

In 406 judo athletes, summed handgrip strength rose with weight category — 85 kgf at 60 kg up to 121 kgf above 100 kg. Divided by body weight it fell, from 1.43 kgf/kg down to 0.98 [11]. Both trends were significant with nearly identical effect sizes (ηp² ≈ 0.31).

The heaviest athletes are the strongest and the weakest, depending on which number you report. In a sport with weight classes, relative is almost always the meaningful one — and it is the reason our own benchmark standards scale targets by body weight rather than holding every athlete to one number.

How to train it

Read this section knowing what it is. Everything above is descriptive research — it compares people who already differ. No study has trained any of these qualities in grapplers and measured what happened to their grappling. What follows is reasoning from the descriptive evidence, not a demonstrated protocol, and we would rather say that than dress inference up as a finding.

With that stated, the evidence points somewhere reasonably specific:

• Keep building maximal strength, and stop expecting it to be the difference. It did not separate elite from non-elite, but the studies finding that were small, and strength underwrites everything else you do. Treat it as the foundation, not the edge.
• Train the arms to hold, not just to pull. Given that the gi tests load eccentric elbow flexion harder than anything else, long holds and slow lowering under load resemble the measured demand more closely than fast reps do.
• Train grip for duration, not for a single maximal squeeze. Maximal grip force survived a ten-minute match nearly intact; endurance lost a third. Whatever you do, the variable to progress is time under tension.
• Use the gi tests as your tracking measure. Static hang for time and gi pull-ups to failure both have excellent test–retest reliability (ICC 0.99 and 0.97). No percentile table has ever been published, so they tell you your direction, not your rank.
• Do not ignore the flexibility result. The largest single effect in the elite-versus-non-elite comparison was sit-and-reach, d = 2.67. Nobody knows whether that is cause, consequence or confound — but it is the most conspicuous number in the study and it costs very little to address.

What nobody has measured

This is the part most articles on this subject leave out, and it is the part that should change how much weight you put on the rest.

• No intervention study exists. Not one trial has trained grip endurance, isometric strength or anything else in grapplers and shown improved match outcomes — or even improved sport-specific test scores. Every link above is correlational.
• Almost nothing is validated against winning. Predictors are checked against belt rank, simulated matches or technical-action counts. The one BJJ study that used real win–loss as the outcome found no physical differences at all [3].
• A validated grappling test correlated with nothing in the lab. A sport-specific BJJ work-capacity test shared essentially no variance with any neuromuscular or anaerobic measure applied to it — r ranged from −0.06 to 0.24 [12]. Either the lab tests miss what grappling asks, or the sport test does. Nobody knows which.
• Core endurance is unmeasured. Trunk endurance is near-universally named as a mat-strength component. No study has tested it against grappling performance or competitive level in BJJ, judo or wrestling.
• Gi versus no-gi is unmeasured. Every grip finding above is gi-only. No study compares the grip demands of the two rulesets.
• The samples are tiny. Ten, fourteen, fifteen, eighteen, twenty-four. The convergence across them is the reason to take the pattern seriously; no individual result deserves much weight on its own.
• This is male data. The systematic review covering 1,496 athletes located only two studies involving women, both body-composition only [13]. The strength meta-analysis excluded women entirely. The handgrip norms are male. Any number on this page is a male number, and we are not going to pretend otherwise.

What survives all of that is modest and still useful. Mat strength is not a myth and it is not magic. It is upper-body strength endurance, expressed through a bent arm held for a long time, and it is not the thing a one-rep max measures. That is why the athlete you out-lift can still hold you down — and it is why the number you should be watching is how long you can hold, not how much you can move.