Do Heavy Negatives Actually Help? What a New Study Found

Quick answer: A meta-analysis by Yang and colleagues, published 18 June 2026 in Sports Medicine Open, pooled eight studies on accentuated eccentric loading, where extra weight is added to the lowering phase and stripped off before you press. The rep straight after was slower, not faster. By rep two the effect had vanished.

Somebody in your gym swears by heavy negatives. Load the bar past your max, lower it under control, have a partner strip the extra plates at the bottom, then drive it up. The bar is supposed to fly.

The technique has a proper name, accentuated eccentric loading, and coaches have used it for years on the promise that a heavy lowering phase primes the muscle for a faster lift. Eight studies have now measured whether the bar actually moves quicker.

What the researchers actually did

The team searched four databases for studies where lifters used free weights, loaded the lowering phase heavier than the lifting phase, and had their barbell speed measured against a normal set with matched lifting weight.

Eight studies survived, covering 106 strength-trained people. Seven of them measured bar speed with a linear position transducer, a sensor on the bar that records how fast it travels.

The extra load came off mid-rep. Most studies used weight releasers: hooks holding additional plates that drop away when the bar reaches the bottom. Eccentric loads ran from 100% to 120% of the lifter's normal one-rep max, the heaviest weight they can lift once.

The outcome was mean concentric velocity, meaning the average speed of the bar on the way up, averaged across the whole range of motion rather than at its fastest point. That distinction turns out to matter.

Only the back squat and bench press were tested.

What they found

The headline number is the effect size, a way of expressing a difference in units of variability so results from different studies can be pooled. Around 0.2 counts as small, 0.5 as moderate. A negative number here means the bar was slower.

The bar was slower on the rep after the extra load came off, then normal again

-0.3-0.2-0.100.1

-0.25

0.01

0.08

Rep 1

Rep 2

Rep 3

View the data

Group	Rep 1	Rep 2	Rep 3
Effect on bar speed vs normal loading	-0.25	0.01	0.08

So the promised speed boost did not appear. What appeared was a small slowdown, and it was gone one rep later.

The pooled estimate for the first rep is also unusually tidy. The studies agreed with each other almost perfectly, with none of the variation between them attributable to genuine differences (I² of 0%). That is rare in a meta-analysis and it makes the finding harder to dismiss.

Then the researchers went hunting for conditions where the technique might still work.

Eccentric load did not matter. Going to 110% or 120% of the one-rep max produced the same result as staying at 100% (effect sizes -0.25 and -0.24, p = 0.908). Neither did the size of the gap between the lowering and lifting weights (p = 0.353).

Exercise choice did matter.

The squat lost three times as much speed as the bench press

-0.4-0.3-0.2-0.10

-0.33

-0.11

Back squat

Bench press

View the data

Group	Back squat	Bench press
Effect on bar speed, first rep	-0.33	-0.11

Neither exercise sped up. The squat just paid more for it.

Why did the bar slow down?

Your muscles do store something during a hard stretch. Lower a weight and the tissue behaves a bit like a stretched elastic band, holding energy that can feed the lift back up. That is the stretch-shortening cycle, and it is the whole theory behind heavy negatives.

The problem is what a supramaximal load does to the way you lower the bar. Nobody drops 120% of their max quickly. You slow the descent down to stay in control, and holding the bottom position while the extra plates release stretches the pause out further.

Elastic energy leaks away with time. Pause at the bottom of a squat and the stored energy dissipates before you use it. The researchers point to exactly this: weight releasers lengthen the lowering phase, which blunts the stretch-shortening cycle.

There is a second, more mundane explanation. Losing 40 kg off the bar in an instant, at the bottom of a squat, changes what you are balancing. Re-stabilising costs you a fraction of a second before you can drive.

Both are the authors' proposed explanations rather than something this analysis measured. The pooled data show what happened to bar speed, not why.

That also explains why the squat suffered more. A bar on your back, halfway down, with plates falling off it, is a harder object to control than one on your chest.

What this study doesn't prove

• Bar speed is not muscle, and it is not strength. This analysis measured how fast the bar moved on the first three reps of a set. It says nothing about whether training this way for eight weeks builds more muscle or a bigger max. Those are different questions with different answers.
• Six of the eight studies had bias concerns. Only two were rated low risk. The most common problem was that lifters did the conditions in random order but the studies did not confirm the order was counterbalanced, so people may simply have got better at the exercise as the sessions went on.
• Only two exercises were tested. Back squat and bench press, free weights only. Flywheel devices, pneumatic machines and computer-driven eccentric rigs were all excluded, and they behave differently.
• The evidence gets thin fast. The rating for the first-rep result was moderate certainty. For the second rep it drops to low, and for the third rep to very low, because fewer studies reported those reps and they disagreed more.
• Publication bias was never checked. With fewer than ten studies the usual test is unreliable, so the authors skipped it. Nobody knows how many null results never got published.
• Strong lifters may respond differently. The authors note that people who squat at least twice their bodyweight seem more likely to get a speed benefit on the second rep, but there were too few of them to test properly.

On the reassuring side, the study received no funding and all five authors declare no competing interests.

What changed, and what didn't

Note the first row carefully. Earlier studies that found a benefit measured peak velocity, the single fastest instant of the lift. This analysis measured the average across the whole way up. A brief spike can exist inside a slower rep.

How to test this in your own training

1. Decide what you actually want from it. If the goal is a faster bar on the next rep, this analysis says the technique does not deliver. If the goal is exposure to heavier eccentric loads over months, that is untested here and worth its own trial.
2. Run it on one exercise for six to eight weeks. Bench press is the safer choice, given the squat lost more speed and involves a bar on your back while plates fall off it.
3. Keep the lifting weight identical to your normal sets. The comparison only works if the only thing that changed is the lowering load. Log it plainly: Bench 80kg 3x5 (AEL 100kg ecc, rep 1).
4. Track reps at a fixed weight, not how the set felt. Heavy negatives feel productive. That is not evidence. Watch whether your reps at the same load climb faster than they did in the block before.
5. Give yourself an exit. Weight releasers need a spotter and a rack you can reset quickly. If a set now takes three minutes to set up, the one rep max calculator can tell you whether all that faff moved your estimated max at all.

Where does Gym Note Plus fit?

Running a six-week test on yourself needs one thing: a readable record of what you lifted before you started. Most people have that data sitting in their phone's notes app, in a form nobody can read a trend out of.

Gym Note Plus takes the plain text you already write and turns it into per-exercise charts. You keep your own shorthand, notes on technique included, and the app handles the reading back.

• Type or paste a session in plain English, get structured sets and reps
• Weight and rep trends per exercise, so six weeks of a technique shows up as a line
• Estimated one-rep max tracked from your working sets
• Your whole history stays searchable, so you can find the block before the experiment

Paste a session into the free notes to workout translator to see what it makes of your shorthand.

Eight studies agreed that the bar does not fly after a heavy negative. Whether it builds anything over months is a question only your own log can answer.

Frequently Asked Questions

What did the 2026 meta-analysis on accentuated eccentric loading find?

The 2026 meta-analysis by Yang and colleagues in Sports Medicine Open pooled eight crossover studies covering 106 strength-trained lifters. Accentuated eccentric loading made the repetition immediately after it slightly slower, not faster, with an effect size of -0.25. Bar speed returned to normal by the second repetition.

Do heavy negatives make your next rep faster?

No. Across the eight studies pooled by Yang and colleagues in 2026, mean bar speed on the repetition right after the extra eccentric load was removed was slightly lower than on a normal set. The studies agreed almost perfectly on this. Earlier reports of a speed boost measured peak velocity rather than average velocity.

Does accentuated eccentric loading build more muscle?

The Yang 2026 meta-analysis cannot answer that. It measured barbell velocity on the first three repetitions of a set, an acute outcome, not muscle size or strength after weeks of training. Whether accentuated eccentric loading beats normal loading for hypertrophy over a training block remains a separate and open question.

Should you use heavy negatives on the squat or the bench press?

If you use accentuated eccentric loading at all, the bench press is the less costly choice. In Yang and colleagues' 2026 analysis, the back squat lost roughly three times as much bar speed as the bench press on the first repetition (-0.33 against -0.11), a gap unlikely to be chance (p = 0.022).

How heavy should the eccentric load be for accentuated eccentric loading?

The 2026 meta-analysis found no benefit to going heavier. Eccentric loads above 100% of the one-rep max produced the same effect on bar speed as loads at exactly 100% (p = 0.908), and the size of the gap between lowering and lifting weights made no difference either. Extra eccentric load bought nothing measurable here.

Final Takeaways

• The bar was slower, not faster. Eight studies, 106 lifters, and a pooled effect size of -0.25 on the repetition right after the extra load came off.
• The effect vanished immediately. Reps two and three matched normal loading, so this is not a technique that primes a whole set.
• Loading heavier changed nothing. 110% and 120% of the one-rep max performed no differently from 100%.
• The squat paid more than the bench. Roughly three times the speed loss, likely from re-stabilising a bar that just got 40 kg lighter mid-rep.
• This was about bar speed only. Long-term muscle and strength effects were not tested here, and training to failure or plain progressive overload remain the better-evidenced places to spend your effort.