Understanding Tracking Systems: GPS, Optical Tracking, and Accelerometers in Football

Prerequisites: This article assumes familiarity with the distinction between external and internal training load and the basic logic of load monitoring. If any of these topics are new to you, start with:

External Load and Internal Load: Concepts, Differences, and Monitoring Strategies

Learning Objectives

After reading this article, you will be able to:

Distinguish the operating principles, strengths, and limitations of GPS, optical tracking (OT), RFID/UWB, and inertial measurement units (IMU).
Explain that validity and reliability of tracking systems exist on a continuum that varies by variable and context.
Understand the difference between positional differentiation and Doppler-shift methods for GPS-derived speed and distance, and the error compounding issue in acceleration measurement.
Explain the limitations of comparing data across different tracking systems and the need for calibration equations.
Evaluate the derivation principles, validity limitations, and appropriate application contexts of accelerometer-based metrics such as Player Load and metabolic power.

How Players Are Tracked: System Classification and Evolution

Player-tracking technologies fall into two broad categories. The first is positional tracking systems, which determine where a player is on the pitch. This category includes optical tracking (OT), radio frequency identification (RFID), and the Global Positioning System (GPS). The second is wearable microsensors, built around inertial measurement units (IMU) that detect how a player moves regardless of pitch location (Clubb & Murray, 2022).

The history of objective movement analysis in football dates back to 1976, when Tom Reilly used manual coding to record player movements during matches. Camera-based tracking emerged in the late 1990s, enabling simultaneous tracking of all players, referees, and the ball across a full match for the first time. GPS devices entered professional sport in the early 2000s and have since become the most widely adopted wearable technology for training monitoring. Today, elite clubs routinely combine multiple systems within the same environment to capture both positional and inertial data (Clubb & Murray, 2022).

The choice of system depends on venue availability, budget, and the type of information required. OT systems are typically installed in home stadiums for match analysis. GPS units are portable and used for training. RFID and ultra-wideband (UWB) solutions offer high-frequency indoor tracking but require dedicated infrastructure. Understanding these trade-offs is the starting point for building a monitoring workflow that fits a specific environment.

From Satellites to the Device on Your Back: How GPS Works

GPS is a satellite-based navigation network that calculates a receiver’s position by comparing the arrival times of low-power radio signals transmitted from orbiting satellites with onboard atomic clocks. A minimum of four satellites must be visible for a three-dimensional position fix. The broader term Global Navigation Satellite System (GNSS) encompasses both the US GPS constellation (24 satellites) and the Russian GLONASS constellation (24 satellites), among others. Modern devices that access multiple constellations are referred to as multi-GNSS receivers (Clubb & Murray, 2022).

Two methods exist for deriving speed and distance from GPS signals. Positional differentiation calculates distance by connecting successive position coordinates over time. The Doppler-shift method measures the frequency change of the satellite signal caused by the receiver’s movement to estimate instantaneous velocity. In practice, most team-sport GPS devices use positional differentiation for distance and the Doppler-shift method for speed, because the latter produces higher precision at the point of measurement (Clubb & Murray, 2022).

Acceleration is the first derivative of velocity and the second derivative of displacement. When acceleration is derived from GPS position data, measurement error at each stage compounds, making GPS-based acceleration values inherently noisier than speed or distance values. High-frequency low-pass filters such as Kalman or Butterworth filters are routinely applied to smooth this noise (Murray & Clubb, 2022).

Sampling frequency matters. Early GPS devices operated at 1 Hz; current best practice recommends 10 Hz or above. Some manufacturers interpolate a lower native sampling rate upward using onboard accelerometer data, so practitioners should verify the true GPS sampling frequency of any device before purchase (Clubb & Murray, 2022).

Signal quality is affected by weather, urban canyons, tree cover, and stadium roofs. Horizontal Dilution of Precision (HDoP) quantifies satellite geometry quality on a continuous scale: a value of 1 is ideal, below 2 is excellent, and above 20 indicates poor signal quality (Murray & Clubb, 2022).

Cameras and Radio Waves: Stadium-Based Tracking Systems

Optical tracking (OT) systems reconstruct two-dimensional player movement by inferring x-y coordinates from multiple cameras installed around the stadium. Machine learning and computer vision algorithms identify and track each player across frames. OT can track the ball as well, providing tactical information that wearable devices cannot capture. The main limitations are cost, lack of portability, susceptibility to occlusion and lighting changes, and the inability to quantify vertical movements such as jumps (Clubb & Murray, 2022).

A validation study of the TRACAB optical tracking system compared two generations (Gen4 and Gen5) against a VICON motion-capture criterion. Gen5 recorded a position RMSE of 0.08 m and a speed RMSE of 0.08 m/s. At the level of key performance indicators, Gen5 showed only trivial deviations from the criterion across all metrics, while Gen4 displayed small-to-moderate deviations in low-speed and high-speed distance categories (Linke et al., 2020).

Radio Frequency Identification (RFID) systems calculate position from the synchronised signal reception times between anchor antennas around the pitch and microchip transmitters worn by players. Some RFID systems sample at 100 Hz or above and can integrate heart-rate data and video synchronisation. Ultra-Wideband (UWB) technology is a higher-precision variant of RFID that uses radio signals with a bandwidth exceeding 500 MHz, offering low power consumption, strong obstacle penetration, and high interference resistance (Clubb & Murray, 2022).

RFID and UWB can outperform GPS and OT in instantaneous speed and acceleration accuracy, and they function indoors where satellite signals are unavailable. The trade-off is significant infrastructure cost and limited portability, which restricts deployment to venues with permanent installations (Murray & Clubb, 2022).

System	Sampling Rate	Portability	Indoor Use	Ball Tracking	Key Limitation
GPS/GNSS	10–18 Hz	High	No	No	Signal affected by environment.
Optical Tracking	25 Hz	None	Limited	Yes	Cost, occlusion, no vertical data.
RFID/UWB	Up to 100+ Hz	None	Yes	Only if ball is tagged.	Infrastructure cost, limited portability.

Sensors on the Body: The World of IMU and Accelerometers

An Inertial Measurement Unit (IMU) integrates three types of sensors within a Microelectromechanical System (MEMS): an accelerometer measuring linear acceleration across three axes, a gyroscope measuring angular velocity, and a magnetometer measuring orientation relative to the earth’s magnetic field. Three-axis accelerometers in modern devices sample at 1,000 Hz or above, capturing high-frequency movement detail that positional systems cannot resolve (Clubb & Murray, 2022).

GPS manufacturers use the resultant vector magnitude of three-axis accelerometer data to quantify whole-body mechanical stress. This metric is labelled Player Load (Catapult), Body Load, or Accumulation Load depending on the manufacturer. It represents a scaled instantaneous rate of change in acceleration and is reported in arbitrary units rather than standard SI units. The proprietary nature of these algorithms means that the exact calculation process is opaque, and independent researchers have reported discrepancies between published formulas and actual software outputs (Staunton et al., 2022).

The accelerometer is worn between the scapulae, not at the body’s centre of mass. Player posture influences the signal: a field hockey player who spends 89% of match time in a forward-flexed position produces different accelerometer profiles than a football player running upright. This positional dependency must be acknowledged when interpreting accelerometer-derived data (Murray & Clubb, 2022).

Buchheit & Simpson (2017) proposed a three-level classification of tracking metrics. Level 1 variables are speed-zone distances derivable from any positional technology. Level 2 variables involve acceleration, deceleration, and change of direction, requiring higher-frequency data. Level 3 variables are derived exclusively from inertial sensors and include stride characteristics, Force Load, and left-right asymmetry. Because Level 3 variables are independent of tactical activity on the pitch, they hold particular promise for monitoring fitness and fatigue longitudinally.

How Much Can You Trust Tracking Data: Validity and Reliability

Validity describes how accurately a device measures what it intends to measure. Reliability describes how consistently it reproduces the same result under the same conditions. A critical principle for practitioners is that validity and reliability are not binary labels but exist on a continuum that shifts with the variable being measured and the movement context (Murray & Clubb, 2022).

Across all tracking technologies, measurement precision deteriorates as the rate of velocity change increases. Distance and average speed are measured with relatively low error, but acceleration, deceleration, and change of direction push every system toward its accuracy limits. A comparison of three 10 Hz GPS manufacturers found that the coefficient of variation (CV) for deceleration ranged from 2.5% to 72.8%, while distance and speed stayed between 0.2% and 5.5% (Murray & Clubb, 2022).

When GPS and optical tracking outputs were compared during official matches, all variables showed ICC values above 0.90. However, the CV for sprinting distance reached 13.5–14.9%, whereas total distance and low-speed zones remained below 5% (Pons et al., 2019). This pattern reinforces a core paradox: the variables that practitioners consider most important for monitoring — high-speed running distance, acceleration and deceleration counts, metabolic power — tend to carry the lowest validity and reliability (Buchheit & Simpson, 2017).

GPS inter-unit variability can reach up to 50% even within the same brand. Best practice requires assigning the same unit to the same player across all sessions and applying a larger threshold for meaningful change when comparing data between players. Additionally, firmware and software updates can alter data outputs substantially. One documented case showed acceleration counts dropping from 251 to 177 after a single software update (Varley et al., 2022). Validation is therefore not a one-time event but an ongoing process.

Training on GPS, Matches on Camera: Can You Merge the Data?

Professional clubs frequently use GPS during training and OT during matches due to league regulations or commercial contracts. This creates a practical challenge: aggregated metrics from different systems cannot be directly compared without calibration (Murray & Clubb, 2022).

Regression equations have been developed to translate outputs between systems. When GPS (Wimu Pro, 10 Hz) and OT (Mediacoach, 25 Hz) were compared across 42 official matches, researchers derived conversion formulas for each variable — for example, total distance: $y = 0.971x + 29.653$ m (Pons et al., 2019). These equations showed almost perfect agreement (ICC .999) for total distance but wider error bands for high-speed zones.

However, calibration equations are specific to the manufacturer, model, chipset, and software version used at the time of derivation. A firmware update or a switch to a different GPS chip can invalidate previously established equations. The recommended approach is for each organisation to conduct its own calibration protocol by running both systems simultaneously over a representative sample of sessions and deriving environment-specific equations (Murray & Clubb, 2022).

Consideration	Recommendation
Same system for training and match	Ideal scenario; direct comparison valid.
Different systems, calibration done	Compare with caution; update equations after any firmware change.
Different systems, no calibration	Do not compare aggregated metrics directly.

Beyond the Numbers: Limitations and Principles for Using Tracking Metrics

Metabolic power attempts to estimate energy expenditure by combining GPS-derived speed and acceleration data, expressed in W/kg. The model was developed from straight-line running energetics, and four independent research groups have consistently reported that GPS-based metabolic power underestimates actual metabolic demand during team-sport-specific movements involving multi-directional effort and static high-intensity actions (Buchheit & Simpson, 2017). The gap between estimated and measured values is large enough to limit the metric’s utility as a standalone training load indicator (Rice et al., 2023).

Displacement-based variables assume that a player’s direction and loading are constant, an assumption that breaks down in the multi-directional, contested environment of a football match. A player covering 100 m in a straight line and a player covering 100 m through repeated changes of direction accumulate the same distance but vastly different physiological and mechanical demands (Clubb & Murray, 2022).

Threshold definitions for speed zones and acceleration bands vary considerably across studies and manufacturers. A systematic review found that the most frequently reported external load metrics were speed-zone distance (50 studies) and total distance (47 studies), yet definitions of speed zones, units, and terminology differed substantially between publications (Miguel et al., 2021). Absolute acceleration thresholds (e.g., >3 m/s²) do not account for individual capacity differences, and GPS inter-device variability of up to 56% amplifies the misclassification of high-intensity efforts when fixed thresholds are applied (Pimenta et al., 2026).

Tracking systems measure external load exclusively. External load data interpreted in isolation provides an incomplete picture because the same external stimulus produces different internal responses depending on a player’s fitness, fatigue, nutrition, and psychological state. Integrating external and internal load measures is conceptually necessary for understanding the training process (Impellizzeri et al., 2019).

Before adopting any new technology or metric, practitioners should evaluate its usefulness through a cost-benefit lens: cost, ease of use, portability, and impact on the training programme. The question “what can we measure?” should always be preceded by “what do we need to know?” (Torres Ronda, 2022). Not everything that can be measured is worth measuring. The success of a monitoring system rests not on the technology itself, but on the practitioner’s understanding of each variable’s limitations and their ability to communicate findings in a way that supports coaching decisions (Buchheit & Simpson, 2017).

Key Takeaways

Player-tracking technologies fall into two categories — positional systems (OT, RFID, GPS) and wearable microsensors (IMU) — each with unique strengths and limitations.
In GPS devices, speed is typically derived via the Doppler-shift method and distance via positional differentiation, with the former showing higher precision.
Across all tracking systems, measurement precision deteriorates as the rate of velocity change increases — the variables deemed most important for monitoring tend to have the lowest validity and reliability.
Aggregated metrics from different tracking systems should not be directly compared; deriving calibration equations within your own environment is best practice.
Accelerometer-based metrics such as Player Load are expressed in arbitrary units, with differing proprietary calculation methods across manufacturers making standardised comparison difficult.
Validity and reliability exist on a continuum rather than as binary judgments, and validation is an ongoing process due to firmware and software updates.
The foundation of successful player monitoring is not the technology itself, but the understanding of each variable’s limitations and the ability to contextually interpret external and internal load together.

References

Buchheit, M. & Simpson, B. M. (2017). Player-Tracking Technology: Half-Full or Half-Empty Glass?. International Journal of Sports Physiology and Performance, 12(s2), S2-35-S2-41. https://doi.org/10.1123/ijspp.2016-0499
Clubb, J., & Murray, A. M. (2022). Characteristics of tracking systems and load monitoring. In D. N. French & L. Torres Ronda (Eds.), NSCA’s Essentials of Sport Science. Human Kinetics.
Impellizzeri, F. M., Marcora, S. M., & Coutts, A. J. (2019). Internal and External Training Load: 15 Years On. International Journal of Sports Physiology and Performance, 14(2), 270-273. https://doi.org/10.1123/ijspp.2018-0935
Linke, D., Link, D., & Lames, M. (2020). Football-specific validity of TRACAB’s optical video tracking systems. PLOS ONE, 15(3), e0230179. https://doi.org/10.1371/journal.pone.0230179
Miguel, M., Oliveira, R., Loureiro, N., García-Rubio, J., & Ibáñez, S. J. (2021). Load Measures in Training/Match Monitoring in Soccer: A Systematic Review. International Journal of Environmental Research and Public Health, 18(5), 2721. https://doi.org/10.3390/ijerph18052721
Murray, A. M., & Clubb, J. (2022). Analysis of tracking systems and load monitoring. In D. N. French & L. Torres Ronda (Eds.), NSCA’s Essentials of Sport Science. Human Kinetics.
Pimenta, R., Antunes, H., Silva, H., Ribeiro, J., & Nakamura, F. Y. (2026). The Need for GPS Data to be Normalized for Performance and Fatigue Monitoring in Soccer: Considerations for Accelerations and Decelerations. Strength & Conditioning Journal. https://doi.org/10.1519/ssc.0000000000000958
Pons, E., García-Calvo, T., Resta, R., Blanco, H., López Del Campo, R., Díaz García, J., & Pulido, J. J. (2019). A comparison of a GPS device and a multi-camera video technology during official soccer matches: Agreement between systems. PLOS ONE, 14(8), e0220729. https://doi.org/10.1371/journal.pone.0220729
Rice, J., Kovacevic, D., Calder, A., & Carter, J. (2023). Wearable technology. In A. Calder & A. Centofanti (Eds.), Peak performance for soccer: The elite coaching and training manual. Routledge.
Staunton, C. A., Abt, G., Weaving, D., & Wundersitz, D. W. (2022). Misuse of the term ‘load’ in sport and exercise science. Journal of Science and Medicine in Sport, 25(5), 439-444. https://doi.org/10.1016/j.jsams.2021.08.013
Torres Ronda, L. (2022). Technological implementation. In D. N. French & L. Torres Ronda (Eds.), NSCA’s Essentials of Sport Science. Human Kinetics.
Varley, M. C., Lovell, R., & Carey, D. (2022). Data hygiene. In D. N. French & L. Torres Ronda (Eds.), NSCA’s Essentials of Sport Science. Human Kinetics.

Log in