Squat bar path is one of the few technical cues that has genuinely made it into mainstream gym knowledge. Most people who lift seriously have heard that the bar should travel in a straight vertical line over the middle of the foot, and a fair few have filmed themselves from the side to check.
That’s good. It’s a useful cue and it’s basically correct.
But there’s a problem with checking your squat from a camera at the side of the rack, and it’s the same problem we ran into when we wrote about bench press bar path a couple of blogs back. A side-on view shows you one plane of a movement that happens in three. Everything travelling forwards and backwards, you can see. Everything happening left to right, you can’t.
And it turns out quite a lot is happening left to right. More than most lifters would guess, including us.
What a good squat bar path looks like
Start with the part everyone agrees on.
In a back squat, the bar should stay stacked more or less vertically over your midfoot for the whole rep. That’s not an aesthetic preference. It’s a balance requirement. Your midfoot is roughly where your base of support sits, and a loaded barbell that drifts away from that line has to be dragged back over it before you can stand up.
The reasoning is the same as it is for the bench: horizontal drift is wasted work. If the goal is to move a bar from the bottom of a squat to the top, any distance the bar travels forwards or backwards is effort spent going nowhere. Worse, the further the bar sits from your midfoot, the longer the lever your hips and back have to fight. A bar a few centimetres forward of balance at the bottom of a heavy squat is a meaningfully harder squat than the same weight sitting where it should.
Two things that are commonly misunderstood are:
A) “Vertical” doesn’t mean your torso is upright. Bar path and torso angle are separate things. A high bar squat will usually look more upright, and a low bar squat will involve more forward lean, because the bar sits further back on your shoulders and your hips have to travel further behind you to keep it balanced. Both can have a perfectly vertical bar path. Your build matters too: long femurs generally mean more forward lean, and that’s fine. Chasing an upright torso your proportions don’t allow is a good way to end up with a worse squat.
B) The way up should mirror the way down. This is the part people skip. Consistency between the descent and the ascent tells you more than either phase alone. If your bar tracks nicely on the way down and then wanders on the way up, that’s not a bar path problem, it’s a strength or coordination problem revealing itself under load.
The three ways it usually goes wrong
Squat bar path faults are almost always a symptom of something else. The bar is just the thing that makes the underlying problem visible.
Hips shooting up first. You come out of the bottom, your hips rise faster than your shoulders, and the squat turns into a good morning. The bar drifts forward because your torso has folded over to meet it. This is the most common fault at heavy loads and it usually means your quads are giving up before your hips and back do.
Drifting forward onto the toes. The bar tracks toward the front of your foot on the descent, your heels get light, and you finish the rep balanced on your toes. Ankle mobility is often blamed, and sometimes fairly, but it’s just as often a bracing or setup issue.
Bar path that doesn’t match on the way up. You descend along one line and ascend along a different one. This is the sneaky one, because each individual phase can look acceptable and the mismatch only shows up when you overlay them.
If you want a thorough treatment of the front-to-back stuff, PowerliftingTechnique’s guide covers it well and we’d rather point you at good existing work than pad this out here.
Because the more interesting question is what’s happening in the plane you haven’t been looking at.
The dimension the camera misses
Here’s the thing that surprised us when we started reading the research properly.
Bilateral asymmetry in the squat, meaning a meaningful difference between what your left and right sides are doing, is not a niche problem affecting people coming back from injury. It appears to be close to universal.
A study of long jumpers published in the Journal of Strength and Conditioning Research found bilateral asymmetry in hip joint angle and torque under every load tested, from moderate right up to near-maximal, with ankle torque differences emerging as the weight got heavy. These were trained athletes. The asymmetry wasn’t a beginner problem they’d grown out of.
Asymmetry shows up even without a barbell. Work published in the International Journal of Sports Physical Therapy found peak force asymmetries during plain bodyweight squats in a healthy, active population.
And the finding that genuinely made us sit up: one investigation comparing trained lifters against novices found that the regular weight trainers showed greater bilateral asymmetry than the beginners, not less. If you read that the same way we did, then you might also feel slightly uneasy. Thousands of reps don’t necessarily grind an asymmetry away. They can just as easily grind it in.
What ties these studies together is the equipment they needed. Motion capture rigs, laboratory setups, multiple cameras and a researcher to interpret the output. The asymmetries were real and they were measurable, but measuring them took a university lab.
That’s the gap. Not that lifters are ignoring their asymmetries. That most of us have no practical way to see them.
Why this is hard to spot without instrumentation
Think about what’s actually available to you at the rack.
Feel is unreliable, and gets worse the longer the asymmetry has been there. A pattern you’ve repeated for five years doesn’t feel like a lean. It feels like standing up. Your nervous system has long since normalised it. Plenty of lifters only discover a side preference when someone films them from behind, or when they try a split squat and find one leg is comfortably stronger.
A side-on camera is blind to it by definition. Filming from 90 degrees is the standard advice for bar path, and it’s fine for the forward-and-back plane. But a camera at the side cannot see one hip rising before the other, one knee travelling further inward, or the bar tilting a couple of degrees in the frontal plane. You’d need a second camera behind you, level, with a clear view, on every set.
Bar-mounted trackers measure the bar, not you. A device clipped to the barbell tells you where the bar went. It has no idea which leg did the work of getting it there. You can produce a beautifully vertical bar path while your right side quietly handles more of the load than your left, because the bar is a rigid object connecting two sides of a body that are perfectly capable of compensating for each other.
That last point is the one that shaped Korvi’s approach more than anything. Track the bar and you learn about the bar. Track the limbs and you learn about the lifter.
The bit where this got personal
We should be upfront that this isn’t a topic we picked off a keyword list.
Vish had a suspicion about his squat, the sort of thing you half-notice for years without ever confirming, because confirming it requires equipment you don’t have.
Then we put a v3 prototype sensor on him.
The asymmetry was there in the data, clearly enough that nobody had to squint at a chart to be convinced. Something he’d vaguely suspected for years turned into a number on a screen in the space of a few sets.
We’re being deliberately careful about how much we make of this. It’s early prototype hardware, it’s a sample size of one founder in a testing session, and it is a very long way from a validated clinical measurement. We’re not claiming otherwise. But it was the moment the feature stopped being a line on a roadmap for us. There’s a real difference between believing asymmetry detection would be useful and watching it tell one of your own founders something true about his squat that he couldn’t otherwise have proved.
That’s roughly the standard we’re building against: does this tell you something you actually couldn’t have found out on your own?
If you’d rather see this than read about it, we’ve put some of it up on our socials this week. There’s a squat session with the sensors on over on Instagram and TikTok, and we’ve posted the longer version with a bit more of the data on Reddit if you want to poke holes in it. Genuinely, do. The comments on that thread will be more useful to us than another month of testing between ourselves.
What you can do about it right now
You don’t need our sensors to make progress here, and we’d rather give you something useful today than a reason to wait.
Film from behind as well as the side. Set the phone on the floor a couple of metres back, dead centre, and squat. Watch whether the bar stays level, whether one hip rises first out of the bottom, and whether one knee travels inward more than the other. It takes one extra set and it’s the single highest-yield thing on this list.
Check where your foot pressure sits. Not just heel versus toe, but left versus right. Most people find something once they actually pay attention.
Use unilateral work as a diagnostic. Split squats, Bulgarians, single leg RDLs. If one side gives out several reps earlier at the same load, you’ve found something a bilateral squat was happily hiding by letting the strong side pick up the slack.
Watch what happens when you fatigue. Asymmetries tend to be cleanest on your first set and messiest on your last. If your form is symmetrical at rep one and drifts by rep eight, that drift is the useful information, not the rep one that looked fine.
Don’t panic about small differences. The research is fairly clear that some degree of asymmetry is normal and present in essentially everyone. The goal isn’t perfect balance, which doesn’t exist. It’s knowing roughly where you sit and whether it’s getting worse.
Where Korvi fits
Our sensors are worn on the body rather than clipped to the bar, and that’s the whole point of the design.
Six-axis motion data from each limb means the system sees each side of your body as a separate story rather than inferring one story from the barbell. Bar path falls out of that, in three dimensions rather than the one a side-on camera gives you. So does the left-to-right comparison: how each side moves through the rep, how that changes as you fatigue through a set, and whether the gap between your sides is widening or closing across weeks of training.
The same logic runs through everything we’ve built. The training engine currently sets your weights from what you log, because logged sets are the best input available today. Richer movement data makes every layer above it sharper, and asymmetry is a good example of a signal that simply cannot be derived from a training log no matter how diligently you fill it in.
We’re not there yet. The sensor system is still in development and there’s real engineering left between a prototype that catches an asymmetry in a testing session and a product that does it reliably for thousands of people in noisy commercial gyms. That’s the work.
But squat bar path is a good illustration of why we think the wearable approach earns its place. The vertical line over your midfoot is worth getting right, and you can check it this week with a phone and a bit of patience. It just isn’t the whole picture. The other half of your squat has been going unmeasured, not because it doesn’t matter, but because nobody handed you a practical way to look at it.
Korvi is building wearable sensors for strength training, with full 3D motion tracking, the potential for bar path analysis, and AI-powered coaching. The app is free and available now on the App Store and Google Play. The sensor system is in development, with a Kickstarter launch planned for 2027.