Running Power Explained: Training with Watts

In cycling, power is a direct physical measurement: a strain gauge in the crank or pedal hub measures the torque you apply, multiplied by angular velocity, yielding watts of mechanical work. It is one of the most precise metrics in sports science, accurate to within 1-2%. Running power, by contrast, is an estimation. No commercially available running power device directly measures the forces between your foot and the ground in real time during normal running. Instead, every running power meter uses a mathematical model — fed by accelerometer data, barometric pressure changes, GPS speed, and your body weight — to estimate the metabolic or mechanical cost of your running.

The concept was first explored by researchers like C.T.M. Davies in the 1980s, who modeled the energetic cost of running as the sum of work done against gravity, air resistance, and kinetic energy changes. The 2016 launch of Stryd brought running power to the consumer market, followed by Garmin's Running Power Connect IQ app (2017), COROS's native implementation (2019), and Apple Watch's estimated running power (2023). Each vendor adapted the underlying physics differently, which is why the same run produces different watt values across devices.

What running power attempts to capture is the intensity of your effort — a single number that accounts for speed, gradient, wind, and terrain simultaneously. On a flat road at constant pace, power and pace tell you roughly the same thing. But on a hilly course, into a headwind, or on a technical trail, pace becomes unreliable as an effort indicator while power (in theory) stays consistent. This is the core value proposition: power as a universal effort metric that works where pace does not.

Running power models generally estimate the rate of mechanical work by summing the energy costs of several components. While each vendor's exact algorithm is proprietary, the general framework follows principles established in the biomechanics literature, particularly the work of Minetti (2002) and Kram & Taylor (1990).

The Component Model

The critical difference between vendors lies in which components they include, how they weight them, and what sensors they use. Stryd's foot pod measures ground-level acceleration with high precision and includes a wind sensor. Garmin uses wrist-based accelerometry and optionally chest-strap data from the HRM-Pro, which provides running dynamics but no wind measurement. COROS relies on wrist accelerometry with their EvoLab algorithm. Apple Watch uses a combination of accelerometer and GPS data with a machine learning model.

A fundamental limitation of all running power models is that they estimate external mechanical work but cannot directly measure the internal metabolic cost. Two runners with different running economy producing the same mechanical work will have different oxygen consumption — the power number cannot distinguish between them. This is a key difference from cycling power, where mechanical output at the pedal is tightly coupled to metabolic demand regardless of pedaling technique.

Choosing a running power device means choosing an ecosystem. The numbers are not interchangeable between platforms, so the decision locks you into a particular set of reference values. Here is how the major options compare.

Stryd: The Dedicated Power Meter

Stryd is the only device purpose-built for running power measurement. The foot pod sits on your shoe, placing the accelerometer at ground level where it captures foot-ground interaction more accurately than a wrist sensor. The built-in wind sensor (anemometer) allows Stryd to adjust power for headwinds and tailwinds — a capability no wrist-based device offers. Stryd's Critical Power algorithm auto-calculates training zones from your run history, and their PowerCenter web platform provides detailed analytics including power duration curves and race power planning. Independent validation by Cerezuela-Espejo et al. (2020) found Stryd power correlated well (r = 0.911) with metabolic rate during level running. The main limitation is cost and the requirement to carry an extra device.

Garmin Running Power

Garmin offers running power either through the Connect IQ app (older approach) or natively on newer watches like the Forerunner 265/965 and Fenix 7+. When paired with the HRM-Pro chest strap, Garmin captures running dynamics (ground contact time, vertical oscillation, stride length) that feed into the power calculation. Without the chest strap, it relies solely on wrist-based accelerometry and GPS. Garmin's numbers tend to read 10-30% higher than Stryd for the same effort because their model includes a larger estimated metabolic overhead. Garmin power is best used as a trend metric within the Garmin ecosystem rather than compared to any other device's numbers.

COROS EvoLab Power

COROS integrates running power natively across their watch lineup (PACE 3, VERTIX 2, APEX 2). Their EvoLab algorithm uses wrist-based motion data and GPS to estimate power and automatically detects your threshold power from training data. COROS power values typically fall between Stryd and Garmin values. The platform provides training load analysis, fitness and fatigue tracking, and race predictions based on power data. Like Garmin, the absence of a wind sensor means COROS power does not account for aerodynamic conditions.

Apple Watch

Apple introduced running power estimation in watchOS 17, available on Apple Watch Series 8 and later. It uses the built-in accelerometer, gyroscope, and GPS with a machine learning model trained on metabolic cart data. Apple's implementation is the newest and least independently validated. Power data appears in the Workout app and Apple Health but currently lacks the detailed analytics ecosystem (power duration curves, auto-zones, CP tracking) that Stryd, Garmin, and COROS provide. For runners already in the Apple ecosystem who want a basic power reference, it is a zero-cost addition.

Power-based training requires anchor metrics to set meaningful zones. The two primary concepts are Critical Power (CP) and running Functional Threshold Power (rFTPw). Though related, they come from different theoretical frameworks.

Critical Power (CP)

Critical Power is a mathematically derived threshold from the power-duration relationship, first described by Monod & Scherrer in 1965. It represents the highest power output that can theoretically be sustained indefinitely — the asymptote of your power-duration curve. In practice, CP corresponds to roughly a 30-40 minute all-out effort, close to but not identical to lactate threshold. The mathematics behind CP is elegant: the relationship between power (P) and time to exhaustion (t) follows the hyperbolic model: (P - CP) x t = W', where W' (pronounced "W-prime") is the fixed amount of work you can perform above CP before exhaustion. W' represents your anaerobic work capacity — your "battery" above threshold.

Stryd calculates CP automatically from your training history using the power-duration curve. A well-trained male distance runner might have a Stryd CP of 230-280W, while a well-trained female runner might be 180-230W. These values are highly individual and depend on body weight, fitness, and running economy. Jones et al. (2019) validated the CP concept for running, confirming that it closely approximates the maximal lactate steady state when properly modeled.

Running Functional Threshold Power (rFTPw)

rFTPw is the running equivalent of cycling's FTP — the power you can sustain for approximately one hour. While CP is a mathematical model parameter, rFTPw is typically estimated from a 20-30 minute time trial (then taking 95% of the average power for 20 minutes, or 100% for 30 minutes). Garmin and COROS use threshold power concepts similar to rFTPw for their zone calculations. For most runners, CP and rFTPw are within 3-5% of each other, with CP slightly higher because it represents a shorter sustainable duration.

Power Zones

Most power-based training systems use 5-7 zones derived from CP or rFTPw. Below is a typical zone structure used by Stryd and similar platforms.

Testing your CP or rFTPw is straightforward: perform a 3-minute and a 9-minute all-out time trial (for CP calculation) or a single 20-30 minute time trial (for rFTPw). Stryd can also estimate CP from accumulated training data without a dedicated test — its algorithm analyzes your best efforts at various durations from normal training to construct the power-duration curve. Whichever method you use, retest every 4-8 weeks during focused training to track fitness changes.

Running power and pace are not competing metrics — they are complementary tools suited to different situations. Understanding when each excels prevents you from over-complicating simple workouts or under-utilizing power where it genuinely helps.

The practical summary: if your running is predominantly flat road running and track work, pace does the job well and power adds unnecessary complexity. If you regularly run hills, trails, treadmills, or in variable weather, power provides a more consistent effort metric. Many experienced runners use both — pace for flat racing and intervals, power for hilly long runs and trail running. The transition from "pace-only" to "power-aware" does not need to be all-or-nothing.

The Grade Adjusted Pace Alternative

Grade Adjusted Pace (GAP), offered by Garmin and Strava, attempts to solve the same problem as power on hills by estimating what your pace would be on flat terrain. GAP has the advantage of being intuitive — it is still expressed in min/km or min/mile. However, GAP relies on elevation data quality (which can be poor with GPS-derived altitude) and does not account for wind, surface, or vertical oscillation. For runners who find power numbers unintuitive, GAP is a reasonable intermediate tool. For a detailed analysis of GAP, see our Grade Adjusted Pace article.

Once you have established your CP or rFTPw and set zones, power becomes a practical tool for several training applications.

Pacing with Power: Even Effort Strategy

The most impactful application of running power is pacing on variable terrain. The principle is simple: instead of targeting a constant pace (which means surging on downhills and crawling on uphills), target a constant power output. This produces an even physiological effort — your heart rate, oxygen consumption, and lactate production stay more consistent throughout the run. Research by Poole et al. (1988) on cycling and confirmed in running by Denadai et al. (2006) showed that even pacing strategies optimize glycogen utilization and delay fatigue compared to variable pacing at the same average speed.

For a hilly marathon, this means planning a target power rather than a target pace. If your CP is 260W and you plan to race at 90% CP, your target is 234W — hold that number regardless of whether you are going uphill (where your pace will slow) or downhill (where your pace will quicken). Your body does not care about your GPS pace; it cares about the rate of energy expenditure.

Monitoring Training Load and Fatigue

Power enables quantitative training load tracking through the concept of Training Stress Score (TSS) or its running equivalents (rTSS, Stryd's RSS). The formula integrates the intensity and duration of each run: a 60-minute easy run at 75% CP produces far less training stress than a 60-minute tempo at 95% CP, even though both lasted an hour. Tracking weekly TSS helps prevent the two most common training errors: under-recovery (chronic TSS too high) and stagnation (chronic TSS plateau).

Detecting Fatigue: Power-Heart Rate Decoupling

One of the most insightful power applications is monitoring the relationship between power and heart rate over time. In a fresh, well-rested state, your power-to-heart-rate ratio is relatively stable. As fatigue accumulates — either within a single long run or across a training block — heart rate rises while power output at the same perceived effort declines. This "decoupling" is a quantitative fatigue signal. A decoupling of greater than 5% during a long run suggests your aerobic fitness at that duration is still developing. Tracking this metric over months reveals whether your aerobic base is genuinely improving.

Tracking Fitness via CP/rFTPw Trends

Your Critical Power or rFTPw over time is a straightforward fitness indicator. Unlike VO2 Max estimates from watches (which depend on HR-pace algorithms and can be confounded by temperature, fatigue, and cardiac drift), CP is derived directly from your best actual performances at various durations. If your CP rises from 240W to 255W over a 12-week training block, that is an unambiguous fitness improvement — assuming consistent body weight. Stryd's Power Duration Curve provides a visual representation of this: the entire curve shifting upward means you are producing more power at every duration.

Running power is a useful tool, but it is important to understand what it cannot do and where its claims outrun the science.

None of these limitations make running power useless. They mean it should be used as one tool among several — complementing pace, heart rate, and perceived exertion — rather than as a single source of truth. The runners who benefit most from power are those who understand what it measures and what it does not, and who use it in the specific situations where it adds genuine value.

If you have decided to incorporate running power into your training, here is a practical roadmap for getting started without overcomplicating things.

The runners who get the most from power are those who adopt it gradually, use it where it adds clear value, and resist the temptation to let a new number dictate their entire training philosophy. Power is a lens, not a prescription.