Research · Testing & Assessment

Half the Squad Fails the Y-Balance Test. That Is the First Problem.

By Muhammad Dzulhisham · MSc Sports Coaching, NSCA-CSCS

Pre-season testing day. Thirty players take turns standing on one leg on a plastic platform, pushing a reach indicator forward, back and to the side. The numbers go into a spreadsheet, the spreadsheet does the arithmetic, and by the end of the afternoon a third of the squad is highlighted in red as "at risk".

Then the season happens, and the players who get hurt turn out to be spread fairly evenly across the red rows and the white ones.

That is not bad luck. It is roughly what the predictive evidence for the Y-Balance Test says will happen when you take a cut-off from one population and apply it to another. The test is useful. Forecasting injury is the one job it does poorly.

Where the reputation came from

The Y-Balance Test (lower quarter) is a standardised version of the Star Excursion Balance Test. The athlete balances on one leg and reaches as far as possible in three directions: anterior, posteromedial and posterolateral. Reach is normalised to leg length. Coaches usually look at two things: the composite score, and the left–right difference in each direction.

Its injury-prediction reputation rests mainly on early prospective findings that are still quoted in coaching courses. High-school basketball players with a composite reach of 94% of leg length or less were reported as 6.5 times more likely to sustain a lower-extremity injury, and collegiate American footballers at or below 89.6% were 3.5 times more likely to sustain a non-contact injury. An anterior asymmetry of more than 4 cm was also linked to injury.² Those numbers are where "under 94% is a red flag" and "4 cm is the line" come from.

Each of those figures came from one sport, one age group and one study. The question is what happened when people went looking for the same effect elsewhere.

What sixteen predictive studies show

A 2021 systematic review and meta-analysis searched ten databases up to October 2019 and included 57 studies of the lower-quarter Y-Balance Test. Sixteen of them asked whether test scores predicted future injury.¹ The authors did not pool those sixteen. The cut-offs, populations and injury definitions varied too much for a pooled estimate to mean anything, and they report that only two predictive studies used homogeneous methods and reporting.

The composite score is the number most coaches look at first, and it is the weakest predictor. The single positive composite study was in American football, with a cut-off of 89.6%. There it produced an odds ratio of 3.5 (95% CI 2.4–5.3), with 100% sensitivity and 71.7% specificity.¹ Two other cohorts that used composite cut-offs in the 89–94% range found likelihood ratios that were not significant.

The review's conclusion is cautious but clear. The test is a reliable measure of dynamic neuromuscular control, scores differ by sex and by sport, and with general cut points that do not account for those differences, the test "may not be predictive of injury."¹

Two details matter for anyone testing children. First, reliability is lower in adolescents. A score that shifts noticeably between sessions with the same tester is a shaky basis for labelling a fourteen-year-old. Second, the norms differ. Composite reach averaged 97.6% of leg length in soccer players and 92.8% in basketball players.¹ Apply the 94% threshold to that basketball group and you flag its average player.

What happens in young athletes

Of the three youth cohorts below, the largest is the clearest test of whether published cut-offs transfer. Researchers tested 636 male Gaelic footballers and hurlers before the season: 293 adolescents with a mean age of 15.7 and 343 collegiate players. They then recorded injuries at least weekly across the season, logging 222 lower-extremity injuries.²

Read the first two numbers together. Depending on which published threshold was used, up to 56.9% of the squad was flagged before a ball was kicked, and fewer than half of the players who went on to be injured were among them. The authors tried to derive their own Gaelic-specific cut-offs instead. They could not, because no score cleared an area under the curve of 0.6.²

Their multivariable models did reach statistical significance, but look at how they reached it. The model for contact lower-limb injury had a sensitivity of 11.9% and a specificity of 98.7%. The model for non-contact ankle injury had a sensitivity of 22.2% and a specificity of 100%.² In plain terms, the test was very good at clearing players who would not get hurt and very poor at finding the ones who would. One result actually ran the wrong way: a smaller posterolateral asymmetry was associated with contact ankle injury.

A cohort of 257 elite adolescent male Australian footballers, followed through an 18-game season with time-to-event analysis, found no significant association between injury risk and any single Y-Balance parameter.³ The one signal was an interaction. Players with high posteromedial asymmetry who were also in the top 25% for agility had a hazard ratio of 3.26 (95% CI 1.01–10.54, p = 0.048). Depending on how previous injury history was handled, the same estimate fell to 2.69 (0.92–7.82, p = 0.069).³ The authors' own summary is that, in isolation, the test is not useful for identifying injury risk in junior Australian football.

That interaction is interesting rather than actionable. A lower confidence bound of 1.01 is a result sitting right on the line. It came from a subgroup, and it has not been replicated. One plausible reading, which is not something the study tested, is that asymmetry matters more in athletes who can generate the speed to load it.

The upper-quarter version has fared no better. In 133 sub-elite adolescent handball players with a mean age of 15.4, followed for roughly eight months, 57 (43%) sustained a time-loss injury.⁴ Pre-season reach, composite and asymmetry scores did not differ between injured and uninjured players. The only significant predictor in any model was inferolateral asymmetry of 7.75% of arm length or more, which predicted lower-body injury with a hazard ratio of 2.18 (95% CI 1.02–4.68, p = 0.045). That is an upper-limb test predicting leg injuries, at the edge of significance. The authors conclude the test has "limited value as a field-based screening tool."⁴

Why a reliable test can still predict badly

Reliability and predictive validity are different properties, and the Y-Balance Test shows the gap clearly. It measures something real and measures it consistently, at least in adults. But injury in field sport comes from many factors: collisions, fixture congestion, training load, previous injury and growth. A single slow, controlled reach on a platform captures one small part of that. The 2021 review makes the same point: the test does not reproduce the high-velocity mechanisms behind most sport injuries.¹

Base rates matter too. When a third to a half of a squad sits below a threshold, and injuries are common across the whole squad, even a genuine association produces a screen that flags a lot of players and misses a lot of injuries. This is the same pattern seen with functional movement screening, where single-test cut-offs predict poorly and combined models do better.

What this changes in practice

Stop treating 94% and 4 cm as universal lines. Those thresholds came from specific sports and age groups. In young Gaelic players they flagged up to 57% of the squad and found fewer than half of those later injured. A borrowed cut-off is not a risk rating.

Read a good score as more informative than a bad one. In the youth data, specificity ran far ahead of sensitivity. A clean test is modest evidence of lower risk. A poor test means "look more closely", not "this player will get hurt".

Test the same athlete against themselves. Norms differ by sex and sport, and adolescent reliability is lower. A meaningful change from an athlete's own baseline, such as before and after a growth spurt or an ankle sprain, tells you more than a comparison with someone else's population.

Use it to guide training, not to make injury predictions. Single-leg control is worth training regardless of what the test predicts, and the established neuromuscular warm-up programmes train it without needing anyone to be flagged first. Every player does the warm-up, so a screen is not needed to decide who gets it.

If you are going to call it risk, combine it with other information. Previous injury, training load, maturation status and physical qualities together say more than one reach score. The only youth signal worth noting here, asymmetry combined with high agility, came from exactly this kind of combination.

Where this evidence stops

The 2021 review could not pool its sixteen predictive studies, so its conclusion is a count of positive and negative findings, not an effect size. Its reach also ends in October 2019. Most of the included cohorts were collegiate or adult, not youth.

The youth evidence covers three sports, all male or mostly male, each followed for a single season. The Gaelic cohort mixes adolescent and collegiate players and does not report predictive results separately for the 293 adolescents. The Australian football signal rests on a subgroup with a confidence interval that barely clears 1. The handball study used the upper-quarter test and recorded neither injury mechanism nor training load.

None of these studies shows that dynamic balance is irrelevant to injury. None of them tested whether training an athlete's Y-Balance score changes their injury rate, either. What the evidence supports is narrower: the test is a reliable measure of single-leg control, and on its own, with off-the-shelf cut-offs, it is a weak predictor of which young athlete will be injured.