피크 파워 vs 평균 파워: 차이와 활용 상황

If you have ever tested power output during a jump, squat, or bench press, you have likely encountered two numbers: peak power and mean power. At first glance they seem redundant — both are measured in watts, both describe how much power was produced, and both come from the same repetition. So why bother distinguishing them?

The answer lies in what each metric captures about the movement. Peak power reflects the single highest instantaneous power value during the concentric phase — the absolute ceiling of neuromuscular output at one point in time. Mean power, by contrast, is the average power sustained across the entire concentric phase. These two metrics can tell very different stories about an athlete's capabilities, fatigue status, and training adaptations. Choosing the wrong metric for your question leads to misleading conclusions, missed insights, and suboptimal programming decisions.

This guide breaks down exactly what peak power and mean power measure, when to use each one, how to interpret the ratio between them, and how to measure both accurately in the field.

Power in human movement is the product of force and velocity at any given instant: Power (W) = Force (N) × Velocity (m/s). During a concentric movement like a jump or squat, both force and velocity change continuously. The power curve — a plot of instantaneous power over time — rises, peaks, and falls. The two headline metrics are derived from this curve.

Peak Power (PP) is the single highest value on the power-time curve. It represents the absolute maximum rate of mechanical energy production during the movement. In a countermovement jump (CMJ), peak power typically occurs in the final 50-100 milliseconds before take-off, when both ground reaction force and center-of-mass velocity are simultaneously high. Peak power is sensitive to the athlete's ability to coordinate maximal force production with high velocity — a skill that distinguishes elite from sub-elite performers.

Mean Power (MP) is the total mechanical work performed during the concentric phase divided by the duration of that phase: MP = Work (J) ÷ Time (s). It captures the athlete's ability to sustain power output throughout the entire movement. Mean power smooths out moment-to-moment fluctuations and provides a more stable, reproducible measure of overall concentric performance.

To illustrate the difference numerically, consider two athletes performing a loaded jump squat at 60 kg:

Both athletes hit the same peak, but Athlete B sustains a much higher average output. Athlete B is completing more total work in less time, indicating superior overall concentric performance despite identical peak values. This example demonstrates why relying on a single metric can be misleading.

The distinction between peak and mean power is not merely academic — it has direct implications for how you design training programs, monitor fatigue, and evaluate athlete development. Research has consistently shown that these two metrics respond differently to training interventions, fatigue, and load changes.

Different training adaptations show up in different metrics. A study by Cormie, McGuigan, and Newton (2011) found that ballistic training (jump squats at 0-30% 1RM) preferentially improved peak power, while heavy strength training (75-90% 1RM) had a greater effect on mean power during loaded movements. This makes physiological sense: ballistic training develops the rate of force development and high-velocity force production that contribute to the peak, while heavy training improves the force throughout the range of motion that elevates the average.

Fatigue affects peak and mean power differently. Research by Gathercole et al. (2015) demonstrated that after a fatiguing training session, mean power in the CMJ declined by 8-12%, while peak power declined by only 4-7%. Mean power is more sensitive to fatigue because fatigue impairs the ability to sustain force throughout the movement, while the brief peak may be relatively preserved. This has important implications for readiness monitoring.

The peak-to-mean power ratio (PP:MP) is itself an informative metric. A high ratio indicates that the athlete produces a sharp spike of power but cannot sustain it — suggesting either a highly explosive but fatigable neuromuscular profile, or an inefficient movement pattern. A low ratio indicates a flatter, more sustained power output. Changes in this ratio over time can reveal:

• Improving movement efficiency — A decreasing ratio with stable peak power suggests the athlete is learning to apply force more effectively throughout the range of motion.
• Onset of fatigue — An increasing ratio with declining mean power suggests fatigue is impairing sustained force production while the brief peak is preserved.
• Training specificity effects — A period of heavy strength training may decrease the ratio by raising mean power disproportionately, while a plyometric block may increase it by raising peak power faster.

Understanding these dynamics allows coaches to make more nuanced programming decisions rather than relying on a single number.

Peak power is the preferred metric in several specific contexts where the instantaneous maximum output is what matters most for performance or decision-making.

1. Assessing explosive capacity in short-duration movements. For movements where the performance outcome is determined by the highest instantaneous output — such as the final push-off in a vertical jump, a shot put release, or a punch — peak power is more directly related to performance than mean power. Dowling and Vamos (1993) found that peak power during the CMJ correlated more strongly with jump height (r = 0.88) than mean power (r = 0.79) because jump height is ultimately determined by take-off velocity, which is most influenced by the peak of the power curve.

2. Evaluating rate of force development (RFD) improvements. Peak power is sensitive to improvements in RFD — the speed at which force rises during the initial phase of contraction. Athletes who improve their neural drive, motor unit recruitment speed, and rate coding will show peak power increases even if mean power remains stable. This makes peak power a useful marker for tracking the effects of contrast training, post-activation potentiation protocols, and neural-intensive plyometric programs.

3. Identifying the optimal power load (P-max). When constructing a power-load profile to find the load that maximizes power output, peak power is conventionally used. The optimal load for peak power (typically 0-30% 1RM for jump squats, 30-50% for bench throws) identifies the external resistance at which the force-velocity combination produces the highest instantaneous output. Training at or near P-max is a well-validated strategy for power development (Kawamori and Haff, 2004).

4. Comparing athletes for selection purposes. In talent identification and squad selection, peak power relative to body mass (W/kg) is one of the strongest predictors of performance in power-dependent sports. Normative databases and draft combine benchmarks typically report peak power values:

• NFL Combine vertical jump: Elite prospects produce 55-70+ W/kg peak power
• NBA pre-draft: Guards typically exceed 55 W/kg, centers 45-50 W/kg
• Rugby union forwards: International level averages 50-58 W/kg
• Track and field sprinters: Elite level exceeds 65 W/kg

Caution with peak power: Because peak power is a single-sample value, it is inherently more variable than mean power. Between-session coefficient of variation (CV) for CMJ peak power is typically 3-6%, compared to 2-4% for mean power. This means changes of less than 6-8% in peak power may fall within normal measurement noise and should not be over-interpreted.

Mean power is the preferred metric in contexts where sustained output, reliability, and longitudinal tracking are priorities. In many practical applications, mean power provides more actionable information than peak power.

1. Longitudinal monitoring of training adaptations. When tracking an athlete's development over weeks and months, mean power offers lower between-session variability (CV 2-4%) than peak power. This means you can detect smaller real changes with greater confidence. A 4% improvement in mean power is likely a genuine adaptation, while a 4% improvement in peak power may be within measurement noise. For strength and conditioning coaches managing large squads with periodic testing, mean power provides cleaner trend lines and fewer false signals.

2. Fatigue and readiness monitoring. As discussed earlier, mean power declines more rapidly and consistently in response to fatigue than peak power. This makes it a more sensitive readiness indicator. Practical implementation: measure CMJ mean power during the warm-up before each session. Compare to the athlete's rolling 14-day baseline. If mean power is suppressed by more than 8-10%, consider modifying the session. Research by Watkins et al. (2017) demonstrated that daily CMJ mean power tracked accumulated training load more closely than any other single jump metric.

3. Evaluating performance during sustained-effort exercises. For exercises with longer concentric phases — heavy squats, deadlifts, Olympic lift pulls — mean power is more representative of the overall performance than peak power. The peak may occur during a brief acceleration phase, while the rest of the lift is performed at lower power. Mean power captures the athlete's ability to drive through the entire range of motion against the external load, which is the quality that ultimately determines whether the lift is completed.

4. Velocity-based training (VBT) autoregulation. When using mean power to autoregulate training loads, practitioners set minimum mean power thresholds for each set. If mean power drops below the threshold (e.g., below 90% of the first set's value), the set is terminated or the load is reduced. This approach, validated by Jovanovic and Flanagan (2014), prevents junk volume — reps performed at such low power output that they provide minimal training stimulus while accumulating fatigue.

5. Inter-limb and bilateral comparisons. When comparing power output between limbs (e.g., single-leg CMJ or single-leg press), mean power provides a more stable asymmetry index than peak power. Asymmetries greater than 10-15% in mean power have been associated with increased injury risk in team sport athletes (Bishop et al., 2018), making this a valuable screening metric.

Accurate measurement of both peak and mean power requires attention to sensor quality, sampling rate, and testing protocol standardization. Here is a practical guide to getting reliable data.

Sampling rate matters — especially for peak power. Peak power is a single-sample maximum, so it is directly affected by how many data points are captured per second. A sensor sampling at 50 Hz captures one data point every 20 milliseconds. If the true peak occurs between samples, it will be underestimated. At 200 Hz (5 ms intervals), the error is reduced but still meaningful. At 800 Hz (1.25 ms intervals), the true peak is captured with minimal temporal aliasing. Research by Street et al. (2001) demonstrated that sampling rates below 200 Hz underestimate peak power by 5-17% compared to gold-standard 1,000 Hz systems. Mean power is less affected by sampling rate because averaging smooths out inter-sample gaps.

Phase detection is critical. Both metrics require accurate identification of the concentric phase — the portion of the movement where the load or body is moving in the intended direction against gravity. If the phase start or end is misidentified by even 50 milliseconds, mean power calculations can shift significantly because the time denominator changes. High-quality systems use acceleration zero-crossings or velocity thresholds to identify phase transitions. Verify that your measurement device clearly defines how it identifies the concentric phase and that the method is consistent.

Standardized testing protocol for both metrics:

1. Warm-up: 5-minute general warm-up, followed by 3-5 progressive practice trials at 50%, 70%, and 90% effort. Rest 30 seconds between warm-up trials.
2. Rest before testing: 2 minutes after the last warm-up trial.
3. Test trials: Perform 3-5 maximal-effort trials with 60-90 seconds rest between trials for jumps, or 2-3 minutes for loaded barbell exercises.
4. Data selection: For peak power, report the best single trial. For mean power, report the average of the best 3 trials for greater stability. Always report which selection method was used.
5. Environmental controls: Test at the same time of day (±1 hour), with the same footwear, on the same surface, after the same warm-up. Provide consistent verbal encouragement (or none at all).

Common measurement errors to avoid:

• Inconsistent countermovement depth — Deeper countermovement increases peak power but may decrease mean power. Standardize depth with a target (e.g., 90-degree knee angle) or use self-selected depth consistently.
• Arm swing variation — Arm swing adds 8-12% to peak power. Either allow it consistently or eliminate it by placing hands on hips.
• Sensor placement drift — For IMU-based devices, ensure consistent sensor placement on the body or barbell between sessions. Small positional changes can affect acceleration readings.
• Inadequate rest between trials — Insufficient rest causes progressive fatigue that disproportionately affects mean power in later trials, artificially inflating the PP:MP ratio.

To bring these concepts together, here are three practical scenarios demonstrating how the peak power vs mean power distinction drives real training decisions.

Scenario 1: Pre-season profiling of a basketball team. A strength and conditioning coach tests 15 players using CMJ with an 800 Hz IMU sensor. Player A shows peak power of 5,100 W (62 W/kg) and mean power of 2,400 W (29 W/kg) — a PP:MP ratio of 2.13. Player B shows peak power of 4,600 W (56 W/kg) and mean power of 2,800 W (34 W/kg) — a ratio of 1.64. Despite Player A's superior peak, Player B produces more total work per jump. Player A likely has excellent fast-twitch fiber recruitment but poor sustained force production — suggesting a need for heavy strength work to build the force platform under the peak. Player B would benefit from reactive plyometrics and contrast training to sharpen the peak.

Scenario 2: In-season fatigue monitoring for a rugby squad. The coach measures CMJ mean power every Monday morning before training. Over three consecutive weeks, a prop forward shows mean power values of 2,950 W, 2,780 W, and 2,610 W — a cumulative decline of 11.5%. Peak power values over the same period are 4,100 W, 4,020 W, and 3,890 W — only a 5.1% decline. The mean power trend reveals significant accumulated fatigue that the peak power values would have masked. The coach reduces the player's training volume for the following week, and mean power recovers to 2,880 W.

Scenario 3: Evaluating a 12-week training block. A sprinter completes a 12-week program combining heavy squats (4×4 at 85% 1RM) with loaded jump squats (5×3 at 30% 1RM). Pre-to-post testing results:

The disproportionate improvement in mean power and the declining PP:MP ratio indicate that the heavy squat component improved sustained force production throughout the movement. The combined effect — higher peak and substantially higher average — translated into meaningful jump height improvement. If only peak power had been tracked, the coach would have seen a solid 7.3% gain. But the mean power data reveals that the real driver of performance improvement was enhanced sustained power output — suggesting the heavy strength work was the primary contributor, not the ballistic work. This insight guides the next training block.

Summary decision framework:

• Use peak power for: talent identification, P-max profiling, evaluating explosive capacity, comparing to normative databases
• Use mean power for: longitudinal monitoring, fatigue detection, autoregulation, evaluating sustained-effort performance
• Use both for: comprehensive athlete profiling, training block evaluations, diagnosing force-velocity imbalances