The Frontier
A recurring comment about our diagnostics tool: the pinch test doesn't seem to say much. Climbers who score well on it don't necessarily climb higher grades. Climbers who score poorly rarely fail because of it. The obvious conclusion follows: drop the test, it isn't measuring anything useful.
We didn't take that conclusion at face value. It skipped a question. If pinch strength doesn't matter, why is there such a clear, recognizable difference between climbers who pinch well and climbers who don't? That difference is real. The open question is why it fails to show up in a simple correlation with climbing grade. What follows is the working model that came out of chasing that question, built on literature, on our own data, and on two moments where we had to walk our own conclusion back.
What the literature says
Three recent, peer-reviewed studies specifically examined the relationship between grip-type-specific strength and climbing performance.
Söderqvist et al. (2024, International Journal of Sports Physiology and Performance) tested 32 male, advanced-to-elite boulderers across six grip techniques against bouldering performance. Half crimp was the strongest predictor (R² = 0.58). Pinch scored R² = 0.26–0.44, weaker than crimp but far from zero, with excellent test-retest reliability (ICC 0.79–0.96).
Van Bergen et al. (2022, Research Quarterly for Exercise and Sport) found, in 22 advanced-to-elite climbers, r = 0.58–0.67 for crimp and r = 0.66–0.72 for sloper with general ability. More importantly, strength on one hold type did not significantly predict performance on routes built around the other hold type. Grip-specific strength turned out to predict grip-specifically.
Buraas et al. (2025, European Journal of Applied Physiology) found, in 19 men, that climbing-specific finger strength correlated very strongly with bouldering performance (r = 0.89) and redpoint (r = 0.67), while general handgrip and pull-up strength scored noticeably weaker.
Three small, independent samples, one consistent pattern: grip-type-specific strength predicts climbing performance, to varying degrees per grip type, with pinch consistently on the weaker end.
What our own data says
We pulled 141 profiles from our diagnostics tool, filtered out four clearly unusable entries (a placeholder body weight of 200 kg paired with zero values on every test), and kept only measurements validated during a live climbing competition: not self-reported, not coach-estimated. Strength was normalized to body weight, and climbing level was coded on a 7-point scale spanning French/Font bouldering grades from 4 to 8b+ (roughly V0–V2 to V13–V14 on the V-scale).
| Test | n | r | R² | p-value |
|---|---|---|---|---|
| Pulling strength | 135 | 0.69 | 0.48 | 2.1 × 10⁻²⁰ |
| Crimp | 137 | 0.60 | 0.36 | 7.7 × 10⁻¹⁵ |
| Sloper | 136 | 0.59 | 0.35 | 4.5 × 10⁻¹⁴ |
| Pinch | 136 | 0.51 | 0.26 | 3.5 × 10⁻¹⁰ |
Our pinch R² of 0.26 lands almost exactly on the lower bound Söderqvist et al. report for the same grip, in a completely different population, measured with a different method. Two independent datasets converging on the same number isn't coincidence. It's about as strong a confirmation as you get outside a controlled experiment.
That could have been the end of the story: pinch matters, just less than the rest. Two things pushed us further.
Why pinch would lag behind even if it mattered just as much
Before going any deeper into the numbers, it's worth naming two explanations for pinch's last-place finish in that table that have nothing to do with the quality of the test itself. Neither is proven. Both should be stated plainly rather than left implicit.
The first is relevance. Crimps simply show up far more often than pinches in the typical diet of boulders and routes a climber works through. You can reach a high grade almost entirely on routes that never seriously demand pinch strength; pinch is rarely the hold that decides whether you send. A test that measures a skill relevant to most of what you climb will, by construction, explain more variance in general climbing level than a test measuring a skill relevant to only a fraction of it, regardless of how precisely either test captures the underlying ability. Weak correlation with grade doesn't mean weak measurement. It can just mean the skill is seldom load-bearing.
The second is training stimulus, and it follows from the same scarcity from a different angle. Because pinches come up so rarely, climbers rarely get the incidental exposure, and rarely make the deliberate choice, to train them the way hangboarding and crimp work are trained as a matter of course. Crimp and sloper strength improve almost as a byproduct of ordinary climbing. Pinch strength doesn't, because there isn't enough repeated demand to drive that adaptation. For most climbers, whatever pinch strength they have is closer to whatever they started with than to something anyone deliberately built.
Together, these two point at the same pattern in the table above. It isn't that pinch doesn't matter. It's that pinch rarely gets the chance to matter, and rarely gets the chance to develop. Two separate mechanisms, same direction, same rough size of effect. On their own, they'd be a plausible full explanation for why pinch sits at the bottom. What they don't tell us is whether the pinch ability people do have is still meaningful once you look past the population average, which is the question the rest of this piece tries to answer.
The turn: what happens when you look at everything together
Crimp, pinch, and sloper correlate strongly with each other (r = 0.68–0.83), which isn't surprising since they largely share the same finger-flexor chain. That raises an obvious question: does pinch explain anything crimp and sloper haven't already explained, or does it just ride along with them?
We entered all four tests into a single regression model together. The model as a whole explained more (R² = 0.53) than any single test alone. But the individual contributions were underwhelming: pulling stayed strong (β = 0.63, p < 0.001), crimp remained just significant (β = 0.29, p = 0.029), while both sloper (β = 0.14, p = 0.36) and pinch (β = −0.13, p = 0.26) lost significance. Pinch even flipped negative.
At first glance, pinch adds nothing unique. But we checked the variance inflation factor, the standard multicollinearity diagnostic, and it was extreme: 30 to 55, where 10 is already considered a red flag. Crimp, pinch, and sloper are so tightly entangled that a regression model can no longer reliably tease them apart. The model essentially distributes the shared explanatory power arbitrarily across the three, and pinch happened to draw the short straw. That's a statement about the statistical instability of this particular model, not about the value of pinch.
This is exactly the kind of moment that separates a working model from a finished conclusion. We could have stopped here with "pinch adds nothing," and we would have been wrong.
Asking the question differently
If three tests are largely measuring the same underlying finger strength, the meaningful question isn't "what does pinch explain across the whole population that the others don't." It's "how much of pinch is even left once you subtract the shared finger strength."
We modeled pinch as a function of crimp and sloper combined. Those two explained 59% of the variation in pinch score (R² = 0.59); sloper carried most of that weight (β = 0.45, p < 0.001). That leaves 41% unexplained. That residual is, per individual, how far someone's pinch score deviates from what their own crimp and sloper level would predict, and that deviation varies substantially, from −2.1 to +3.9 standard deviations.
One climber in our data has solid crimp (0.72) and sloper (0.65), but a pinch score of only 0.27 against a predicted 0.40: a real, individual pinch weakness despite generally strong fingers. Another climber has modest crimp and sloper (0.55 and 0.38), but a pinch score of 0.52 against a predicted 0.27: a natural pincher who doesn't stand out on any other test.
That residual is the real answer to the original observation. Pinch correlates weakly with overall climbing level not because pinch ability doesn't exist, but because most of it is already explained by something we already measure (general finger strength). What's left is a small but real, individual signal in pure pinch ability, separate from general finger strength, the kind a population-wide correlation with climbing grade can't capture. Whether that signal actually shows up on the wall, whether it predicts anything about how someone climbs, is a separate question this data doesn't answer. We haven't tested it, and nothing in the three studies above tests it either.
Why the thumb is probably the bottleneck
Here we leave confirmed data and enter reasoning, flagged explicitly as such. It also gives the training-stimulus argument above a sharper, more specific shape: not just "pinch as a whole is under-trained," but a guess at exactly which part of it is.
Pinching is mechanically a closed system: fingers and thumb press against each other, and total force is capped by whichever side is weaker, not the stronger one. The fingers are driven by the flexor digitorum superficialis and profundus, large forearm muscles with substantial cross-sectional area. The thumb is driven by a much smaller muscle group: adductor pollicis, flexor pollicis, opponens pollicis. That asymmetry in muscle mass already exists in non-climbers.
In climbers, it's probably larger still. Nearly all climbing-specific training (hangboarding, crimping, slopers) loads the finger-flexor chain hard and barely touches the thumb, because in a crimp or sloper the thumb contributes almost nothing to the force. Years of training build extremely strong fingers while the thumb stays relatively untouched. In our own data, pinch max strength is lower than crimp max strength in 93% of climbers, with a median ratio of 0.46: pinch typically delivers about half of what the same person produces on a crimp.
If the thumb really is the bottleneck, then the 41% residual from the previous section is largely a proxy for underdeveloped thumb-opposition strength. That's testable, and untested: targeted thumb training should, if the hypothesis holds, raise pinch score disproportionately more than what general finger training alone would predict.
Technique as a hidden variable
One layer further into speculation, and still without data underneath it.
On a route with a vertical line of pinches, most climbers lean hard to one side or the other. Doing so rotates the pulling vector on the hold so that part of it lands against an edge or asymmetry, turning what would otherwise be a pure friction grip into something closer to a side-pull and trading raw squeeze force for mechanical advantage. A climber with very strong pinch strength may not need that and can simply climb straight through.
That means technique can partially substitute for raw pinch strength, but not everywhere. On a symmetric, rounded hold with no edge anywhere (the clearest example: an overhand or underhand sloper where the thumb presses from the opposite side) there's nowhere to lean into. There, raw squeeze strength becomes effectively unavoidable.
If that's right, it partly explains why pinch strength correlates so weakly with performance on an average "pinch route." Part of what happens on those routes isn't a pure strength test. It's a technical puzzle that can substitute for strength. It would also mean an isolated, symmetric pinch-block test (like the one in our diagnostics tool) sits closer to the true, inescapable strength demand than "performance on an arbitrary pinch route" ever could. The test isolates exactly the scenario where technique offers no way out.
The model, open for refinement
Where this leaves us: pinch strength correlates significantly with climbing level, in two independent datasets, with a strikingly consistent effect size. A large share of that strength is shared with crimp and sloper through the finger-flexor chain. What's left is a smaller, individual signal that we suspect, but haven't proven, is mostly thumb-opposition strength, underdeveloped because climbing-specific training largely skips it. On the wall, that signal is further muddied by technique, which, where hold geometry allows it, can substitute for strength.
What's still open: whether targeted thumb training actually raises pinch score disproportionately. Whether performance specifically on pinch-dominant routes correlates more strongly with our test than general climbing level does, which would settle the relevance question. And whether the distinction between "leanable" and "non-leanable" pinches holds up against video or performance data rather than reasoning alone.
None of these questions undermine what we found so far. They're the next step in the same model.
A few more worth sitting with, that this piece doesn't try to answer:
What if climbers trained, on purpose, the one thing the wall itself never trains?
What does a route's grade actually say about pinch difficulty, if the person who set it never trained pinch either?
What are we even measuring pinch strength against, if the entire population sharing our baseline is undertrained on it?
If you want to see where you personally land on any of this, your relative pinch score, how your pinch compares to other climbers, find it in the diagnostics tool.