Arrowhead Agility Test: Setup, Protocol, and Normative Times

A club coach clocks a winger's 30-meter sprint and walks away impressed, then watches the same player get turned inside out by a fullback with a slower straight-line number. Top speed and change-of-direction ability barely correlate, and that gap is exactly what the arrowhead agility test was designed to expose. Because most athletes have never seen the layout before their first score counts, it tends to measure genuine reactive cutting rather than a pattern they memorized after fifty reps of the pro-agility shuttle.

Lockie and Jalilvand (2017) tested the protocol on 20 NCAA Division I women's soccer players and reported test-retest reliability of ICC = 0.92-0.93 for both cutting directions — high enough that a single retest session is trustworthy rather than something you need to average across three visits to believe. This guide walks through the exact cone geometry, the four-trial protocol coaches actually use, where your number likely sits next to published soccer cohorts, and the three cutting errors that quietly add half a second to an otherwise sharp time.

What the Arrowhead Layout Catches That the T-Test Misses

Most change-of-direction tests run the same shape both ways — the T-test and Illinois test are mirror-symmetric, so a left-turn bias and a right-turn bias tend to cancel out in the final number. The arrowhead test deliberately runs two separate, non-mirrored trials to each side, which means it produces a left-side time and a right-side time that can be compared directly against each other rather than averaged into a single score.

That structural choice matters because limb dominance in cutting is common and usually invisible on symmetric tests. Rago et al. (2020) ran the arrowhead protocol through four separate sub-studies on soccer players (reliability sample n = 24; power-related sample n = 56; fatigue sample n = 20; competitive-level sample n = 264) and found the skillful-side and less-skillful-side times were both reliable independently (ICC = 0.80-0.83, CV = 1.25-2.21%). The one output measure that did not hold up was the asymmetry index itself — the calculated gap between sides was noisier than either side's raw time, a limitation worth knowing before you make a training decision off a single asymmetry number.

Course Setup: Exact Cone Geometry

You need six cones, a tape measure, and a flat non-slip surface — turf or a gym floor is more consistent trial to trial than grass, since footing variability shows up directly in the cut phases. Mark every point with tape or chalk before placing cones; eyeballing distances after the first trial is the fastest way to make your retest numbers meaningless.

A full trial covers roughly 35-37 m: the initial straight sprint to cone A, a diagonal cut out to the near-side cone, a second diagonal cut across to the apex cone B, and a straight run back through the finish line. That total distance is short enough to run four times in a session without the third and fourth trials being contaminated by accumulated fatigue, provided you rest properly between them.

Running the Protocol: Four Trials, Two Directions

The athlete starts 0.5 m behind the line, front foot loaded, and sprints on a self-start or coach signal. Four total trials are run — two cutting left through cone C, two cutting right through cone D — with roughly 2 minutes of rest between each, matching the recovery window used in Rago et al. (2020) to keep phosphocreatine resynthesis from becoming a confound. Record the best of the two attempts per side to the nearest 0.01 s using a stopwatch or, better, timing gates at the start/finish line, since hand-timing error on a 7-9 second test is proportionally larger than on a 15-second one.

Trial sequence

1. Initial sprint: Drive out of the start toward cone A, 10 m straight ahead. Treat this exactly like a short sprint start — no dancing, no chopping steps.
2. First cut: Plant the outside foot near cone A and angle diagonally toward the assigned side cone (C for a left trial, D for a right trial), rounding it without touching it.
3. Second cut: From the side cone, redirect diagonally toward the apex cone B. This is the sharpest angle in the course and the one most athletes underestimate on their first attempt.
4. Return sprint: Turn around cone B and sprint in a straight line back through the start/finish line. Coach the athlete to run through the line, not to it.

A trial is voided if the athlete steps over a cone rather than running around it. Once both sides are complete, you can calculate a simple asymmetry index — (slower side time − faster side time) / faster side time × 100 — but treat any single-session value with caution given the reliability limitation noted above. Track it across three or four testing sessions before flagging a real side-to-side deficit.

Where Your Time Lands: Reference Ranges

The arrowhead test doesn't have the multi-decade normative database that the Illinois test or T-test have accumulated, so treat the ranges below as a starting reference rather than a fixed scale — your own athlete's baseline, tracked over time, matters more than where they fall on a chart built from other populations. What the literature does establish clearly is the direction of the differences: Rago et al. (2020) found national-level players outperformed regional-level players across every age group tested (effect sizes -1.97 to -0.36), and that senior and U18 players outperformed U16 players (effect sizes -0.84 to -2.33) with no meaningful gap between senior and U18 themselves.

Rago et al. (2020) also reported that a 15-meter linear sprint explained a significant but partial share of arrowhead performance (R² = 0.42), meaning raw speed accounts for less than half the variance — the remaining 58% is deceleration ability, cutting mechanics, and reactive strength, which is exactly the part a straight-line sprint test will never reveal about an athlete.

Three Cutting Errors That Add Time

1. Rounding the apex cone too wide. Because cone B sits at the sharpest angle in the course, most athletes drift 30-50 cm wide of it on their first attempt to avoid decelerating hard. That wider arc adds distance and, counterintuitively, doesn't actually let them keep more speed — it just costs 0.1-0.2 s per trial for nothing. Mark a small chalk radius around cone B and coach athletes to clip it tight.
2. Coasting through the finish line. Athletes who can see the finish line often ease off their last two strides, the same anticipatory deceleration seen on the 505 and Illinois tests. Time the last 5 m separately on video once and most coaches are surprised how much speed drops before the line — coach the athlete to accelerate through it, not to it.
3. Inconsistent start-foot position across trials. If the front foot lands 5-10 cm closer to or farther from the line on different attempts, the first ground contact changes enough to shift within-session times by 0.05-0.1 s — small on its own, but large enough to swamp a genuine training effect if you're chasing a change smaller than that.

Turning a Slow Time Into a Training Plan

A single arrowhead time tells you an athlete is slow at changing direction; it doesn't tell you why. Pair the test result with what you actually watched happen on the course before writing a program.

• Slow initial 10 m to cone A: The limiter is likely acceleration, not agility. Hip-extensor rate of force development work — trap bar jumps, resisted sled sprints at 10-20% body mass — trains the exact quality being tested in that first phase.
• Wide, slow first cut but a sharp second cut: Usually a deceleration problem specific to shallower angles. Reverse lunges into a stick, and eccentric-focused single-leg work, build the braking force needed to redirect without bleeding speed into the arc.
• Consistent one-sided deficit across three or more sessions: Now the asymmetry is a real signal, not testing noise. Bias unilateral plyometric volume — single-leg bounds, lateral hops — toward the weaker side for 4-6 weeks and retest with the same protocol.

Re-test on a 4-6 week cycle. Given the smallest worthwhile change reported by Rago et al. (2020) sits around 0.06-0.12 s, don't over-interpret week-to-week swings smaller than that — they're closer to measurement noise than adaptation.

Tracking Asymmetry Session to Session with PoinT GO

Because Rago et al. (2020) found the calculated asymmetry index unreliable within a single session, the practical fix is logging every session instead of trusting any one number. The PoinT GO app lets you save each trial's time alongside a quick countermovement jump reading before testing days — if CMJ height drops more than 5% below an athlete's rolling average, that's a readiness flag worth noting before you trust the arrowhead result at all, since a fatigued neuromuscular state will inflate cutting times independent of true change-of-direction ability.

Over a full training block, the app builds a session-by-session left-right trend line rather than a single snapshot, which is the only way to responsibly use an asymmetry measure this noisy — one bad session doesn't mean anything, but four sessions pointing the same direction usually does. Visit poin-t-go.com to see how it fits into a field-testing day alongside sprint and jump assessments.