Data analysis

Every expensive movement in the library returns less than it costs

Fifty-nine of the 272 movements in the Forta database give back less stimulus than they charge in neural cost. Fifty-eight of those are, simply, every movement priced at 8 or above. There is not one exception — and four of the 59 still belong in almost every programme.

Alfredo Silva · Sports Scientist MSc Physical Education · MSc High Performance Training 59 of 272 movements

Every movement in the database carries two numbers: neural cost, from 1 to 10, and stimulus, also from 1 to 10. Divide the second by the first and you get the return — what the exercise pays you for every unit of systemic fatigue it charges. Below 1.0 means it takes more than it gives.

Fifty-nine movements sit below 1.0. That is 21.7 per cent of the library. Their mean return is 0.58.

I expected that list to be interesting. What I did not expect was the shape of it.

There is no exception

The library holds 58 movements at a neural cost of 8 or higher. All 58 of them return less than 1.0. Every single one. Not a single expensive movement anywhere in 272 entries pays for itself.

The 59th is the Kipping Pull-Up — cost 6, stimulus 5.5, return 0.92. It is the only movement below 1.0 that is not in the top price bracket, and it misses by a hair.

So the sub-1.0 list is not really a list of bad exercises. It is the price bracket. The two things are the same set.

59movements return less than 1.0

58of those are every movement at cost 8+

0.58mean return across the 59

0.94the best of them — Romanian Deadlift

The full list

All 59, grouped by the return they produce. Within a band the numbers are identical, so the order inside each row means nothing.

Debit · return below 1.0 return = stimulus ÷ neural cost

Every movement in the database that gives back less than it charges.
Return	Cost	Stim	n	Movements
0.44	8–9	3.5–4.0	2	Handstand Walk · Ring Muscle-Up
0.45	10	4.5	27	Barbell Deadlift · Power Clean · Hang Power Clean · Squat Clean · Hang Clean · Power Snatch · Snatch · Hang Snatch · Muscle Snatch · Dumbbell Snatch · Snatch Balance · Clean & Jerk · Push Jerk · Split Jerk · Power Jerk · Clean High Pull · Snatch High Pull · Thruster · Turkish Get-Up · Box Jump · Box Jump-Over · Broad Jump · Tuck Jump · Depth Jump · Pogo Hops · Skater Bound · Medicine Ball Slam
0.50	10	5.0	4	Front Squat · Zercher Squat · Sumo Deadlift · Trap-Bar Deadlift
0.55	10	5.5	2	Deficit Deadlift · Stiff-Leg Deadlift
0.56	8	4.5	1	Sumo Deadlift High Pull
0.60	10	6.0	1	Barbell Back Squat
0.63	8	5.0	1	Dumbbell Thruster
0.67	9	6.0	1	Overhead Press
0.72	9	6.5	1	Barbell Bench Press
0.75	8	6.0	15	Overhead Squat · Box Squat · Hip Thrust · Barbell Hip Thrust · Barbell Bent-Over Row · T-Bar Row · Pendlay Row · Meadows Row · Upright Row · Push Press · Z-Press · Landmine Press · Decline Bench Press · Close-Grip Bench Press · Skullcrusher
0.81	8	6.5	2	Good Morning · Incline Barbell Bench Press
0.92	6	5.5	1	Kipping Pull-Up
0.94	8	7.5	1	Romanian Deadlift

Two rows carry most of the list. Twenty-seven movements sit at exactly 0.45 — cost 10, stimulus 4.5 — and fifteen at exactly 0.75. That is the Olympic lifts, the jumps and the barbell presses and rows, priced as one block.

One entry needs flagging. The database holds the Dumbbell Thruster twice: once at cost 8 and stimulus 5.0, which is the 0.63 above, and once at cost 4 and stimulus 5.5, which returns 1.375 and is not in this group at all. That is a duplicate in my data, not a finding about the exercise, and I would rather point at it than quietly pick the convenient one.

Why the top of the scale behaves this way

Part of this is honest physiology and part of it is the model, and it matters which is which.

The physiology: systemic fatigue and local stimulus come apart as load becomes axial and coordinated. A cable fly asks one joint to move under tension with the machine handling balance. A Snatch asks for a near-maximal, whole-body, technically fragile effort in under a second. The second one leaves a bill the muscle you were training never sees.

The model: neural cost saturates at 10 while stimulus in this group tops out at 7.5, so once a movement is priced at the ceiling there is no stimulus value that would lift it above 1.0. The finding is therefore partly a statement about my scale, not only about the body. I would rather say that plainly than dress it up.

The useful reading is not "these 59 are bad". It is: there is no such thing as an expensive movement that is also efficient. If you want efficiency you buy it cheaply, and if you want what only the expensive lifts give, you pay full price.

The four that still belong

Four of the 59 go into most programmes I write, at their full price, and I would defend each one against anything in the cheap tier.

Romanian Deadlift — 0.94. The best return in the group, and the highest stimulus of any movement priced at 8 or above: 7.5. A loaded hip hinge under a long hamstring, which nothing on a machine reproduces at the same lengths.

Barbell Bench Press — 0.72. Cost 9, stimulus 6.5. The cable fly at 4.50 builds more chest per unit of fatigue and builds no horizontal pressing strength at all. Those are different purchases.

Barbell Back Squat — 0.60. Cost 10, stimulus 6.0. Deep knee flexion under axial load, which is the one quality the entire cheap tier cannot sell you.

Barbell Deadlift — 0.45. The worst return of the four, and still the best maximal posterior-chain pull there is. If a programme needs that, there is no substitute at a discount.

What those four have in common is not their return. It is that the thing they produce is not on sale anywhere else in the library.

The accounting error underneath

Impellizzeri, Marcora and Coutts argued in 2019 that training load has an external component — what you did, the kilos, the metres, the reps — and an internal one, what it cost the person who did it, and that treating the first as a proxy for the second is a measurement problem, not a detail.

That is the same error, in miniature, in every programme that counts sets. Four sets of leg curls and four sets of Snatches are four sets. They are not the same week.

What to do with this on Monday

Keep the expensive four, or your own version of them. Then ask of every other expensive movement in the plan a blunt question: what is this buying that something at 3.0 or better cannot?

Often the answer is real — rate of force development, technical competence, maximal strength — and you keep it. Often there is no answer, and the movement is there because it looks like training. Drop that one, and the week has room again.

Methodology note

The 272 movements are characterised by a documented, auditable model: 20 exemplars hand-authored from the peer-reviewed literature, and a rules engine that applies the same logic — isolation vs. compound, axial load, stretch bias — to the remaining 252. Every figure in this article was computed directly from that dataset.

A counting caveat: the dataset holds 272 scored entries under 267 distinct names, five of them duplicated. Every count in this article is a count of entries. Four of those duplicates carry the same scores twice and change nothing here; the fifth is the Dumbbell Thruster noted above.

These are working estimates for deciding order and day, not physiological constants. As the section above says, the behaviour at the top of the cost scale is in part an artefact of a scale that stops at 10. Contest it line by line if you like — that is what publishing the numbers is for.

Forta does this arithmetic every morning

A ten-second check-in — sleep, mental energy, soreness — and a score from 0 to 100. The expensive movements appear on the days you can afford them, and not on the days you cannot.

Download Forta