By 2015 I was tired of endurance sports and shifted my athletic focus to weight lifting. I targeted the deadlift, for a single goal to optimize. It works most of your body and is the strongest lift, so you get big, satisfying numbers.
My first training program was just: go to the gym and do deadlifts. Eventually progress stalled. I read about form and experimented with sumo stance (legs spread wide), as opposed to feet roughly shoulder-width. This got me through my first plateau, and hinted that form matters, a lot.
I read Starting Strength by Mark Rippetoe. He harped on optimal bar path—keeping the bar’s as close to straight up-and-down as possible. Otherwise, you’re wasting effort to move the bar horizontally, when the point of the deadlift is to get weight from floor to the position that locks out your knees and lower back.
Doing overhead press after reading Starting Strength gave me my first brush with injury. I thought I’d hurt my rotator cuff. Luckily, it recovered fine—no surgery required. Rippetoe himself has had rotator cuff injuries. This got me wondering if there was more to optimal lifting than just optimizing bar path. Something that incorporates the limitations of human anatomy.
Keeping the bar path vertical reduces leverage, which radically increases the amount of work you must do without changing the weight. But leverage doesn’t just apply to the bar’s position relative to your center of mass—it applies at each joint. Joints have different angular ranges determined by anatomy and flexibility.

Model of joint angle ranges in deadlift.
To perform the deadlift, your mind orchestrates the contraction of muscles to torque each joint: ankles (A), knees (B), hips (C), shoulders (D). With a good set of weights and bar, the weights will rotate smoothly around the bar (W) as you lift, so you don’t have to deal with torque there.
The farther the load is from the joint, the more torque you have to apply at the joint to lift the weight. So optimal position should minimize torque.
Combining joint angle constraints with the equation for total torque (which includes body segment lengths) should give a system of equations that you can optimize with the Simplex algorithm.
Combine that with athletic gear that integrates joint angle measurement, or camera-based systems that measure joint angles, and it should be possible to develop automated coaching systems for weight lifters.
Is anyone doing this?
2 Responses to Mathematics of the Deadlift
Interesting idea! As one who injured myself from improper deadlift form this would be really helpful.
There’s a bunch of golf apps out there that can look at your body position and determine if you’re in proper alignment or not. I don’t see why applying something like opencv and some basic measurements couldn’t get you far enough to create something of value.
Sorry to hear about your injury. I hope you’re on your way to a full recovery.
Yeah, I think the computer vision aspect of it is very doable. The actual model for optimal form has to be a bit more complex than the one I sketched out, to account for the strength of each joint, but I’m hoping an exercise science student sees this one day and takes it to home base.