Deceleration Demands Part 3: Quantifying Braking Load
- Jo Clubb
- 1 day ago
- 6 min read
This article is the third part of a mini-series exploring deceleration demands in team sports.
In Part 1, we discussed why deceleration matters for both performance and injury risk. In Part 2, the focus shifted to how we might assess deceleration capacity through field-based testing. In this article, the emphasis moves again, this time towards monitoring decelerations.
Monitoring deceleration load is about quantifying exposure to braking demands over time. This is distinct from testing capacity, which aims to assess what an athlete is capable of. Both are important, but they answer different questions and come with different limitations.
What Do We Mean by Monitoring Deceleration Load?
As with any form of load monitoring, the aim is to quantify how often, and how intensely, athletes are exposed to a given demand. In the case of deceleration, we are trying to capture how frequently athletes are required to brake, and at what intensity, across both training and competition.
In applied settings, this is done using player tracking derived acceleration data, most commonly from Global Positioning Systems (GPS) technology. Decelerations are identified from the negative portion of the acceleration signal, with efforts counted when acceleration falls below a defined threshold for a minimum duration.
While this approach is widely used, it is important to recognise that it involves several assumptions and compromises.
Deceleration Data Is Continuous, Not Discrete
Acceleration and deceleration data are inherently continuous. Velocity and acceleration fluctuate constantly as athletes move, accelerate, decelerate, and change direction.
To monitor deceleration load, practitioners seek to convert continuous data into discrete “events”, although as we'll see this approach is inherently limited. It requires decisions about where to draw thresholds, and how long a signal must exceed a threshold to be counted as a distinct effort.
This process of discretisation is not unique to deceleration - we use it with high speed/sprint efforts (i.e. velocity) and accelerations - but it perhaps presents a particular challenge given how brief and variable many deceleration efforts are.
Absolute Thresholds and the Problem of Bucketing Data
A common approach is to define absolute acceleration and deceleration thresholds, for example ±2, ±3, or ±4 m·s⁻², and count how many efforts fall within each band.
The issue with this approach is that it imposes arbitrary cut-offs onto a continuous dataset.
Data points that are very similar may be classified into different categories, while data points that are quite different may be grouped together simply because they fall within the same band. The figure below, taken from Miguens et al (2024), illustrates just how spread these efforts are in reality.
Dwell Time and Missed Deceleration Efforts
In addition to threshold magnitude, practitioners must also define a minimum effort duration (MED), often referred to as dwell time. This setting determines how long acceleration or deceleration must remain above a threshold to be counted as an effort, and we've previously explored it on the blog here.
Dwell time is intended to prevent oscillations around a threshold being counted as multiple efforts. However, research has shown that commonly used dwell times may exclude a large proportion of acceleration and deceleration efforts, particularly those of short duration.
Silva et al., 2023 found a high proportion of acceleration and deceleration efforts occured at lower MED. In fact, if a MED of 0.2 or 0.3 seconds was applied to this dataset (as is commonly used), only 36% of accelerations and 34% of decelerations would be reported.
This presents a trade-off. If dwell time is set too high, many genuine deceleration efforts may be missed. If it is set too low, noise and signal variability may inflate counts.
Symmetrical Thresholds for Acceleration and Deceleration
In many monitoring systems, acceleration and deceleration thresholds are symmetrical. For example, +3 m·s⁻² may be used for acceleration and −3 m·s⁻² for deceleration.
However, evidence consistently shows that athletes reach higher maximum values in deceleration than in acceleration across many team sports. Applying symmetrical thresholds therefore risks overestimating deceleration load relative to acceleration load.
Recent work by Pimenta and colleagues (2026) illustrates this clearly. When the same absolute thresholds are applied to acceleration and deceleration, a greater number of high-intensity deceleration efforts are often recorded, because the threshold is low relative to their deceleration capacity.
In their illustration above, we can observe 5 acceleration efforts above the highest threshold (+5 m·s⁻²), but as many as 16 deceleration efforts in the coupled threshold, highlighting that perhaps that threshold is too low relative to the maximum capacity in deceleration.
Individualised Thresholds: A Promising but Imperfect Step
Absolute thresholds also fail to account for differences in individual capacity. Athletes with a higher maximum deceleration may accumulate fewer “high-intensity” decelerations relative to their capacity, while athletes with lower deceleration capacity may be exposed to a higher relative load that is underrepresented in the data.
This introduces bias when comparing athletes, and may obscure meaningful differences in how deceleration demands are experienced internally. In their example below, while the maximum acceleration exceeds the commonly used arbritary absolute threshold by 67% (i.e. +5 m·s⁻² vs +3 m·s⁻²), the equivalent deceleration capacity exceeds the threshold by as much as 133% (i.e. -7 m·s⁻² vs -3 m·s⁻²).
To address these issues, Pimenta and colleagues have proposed using individualised acceleration and deceleration thresholds, expressed as percentages of each athlete’s observed maximum, with
This approach is intuitively appealing. Normalising deceleration efforts relative to individual capacity should improve physiological relevance and reduce bias between athletes.
However, individualised thresholds are not without limitations. The proposed percentage bands are still arbitrary; they propose zero to 25%, 25 to 50%, 50 to 75%, and 75% and above of an individual's maximum deceleration as the thresholds. Those, of course, in themselves are arbitrary cutoffs, and they acknowledge this in their paper.
In addition, GPS measurement error tends to increase at higher acceleration and deceleration magnitudes, precisely where accuracy matters most.
As such, individualisation should be viewed as a step forward in interpretation, rather than a definitive solution. More research is needed to explore these approaches.
Alternative Metrics and the Limits of Simplification
Some tracking systems have introduced alternative metrics, such as acceleration load or acceleration density, which aim to quantify the overall burden of acceleration and deceleration in a more continuous manner, as I've previously discussed for Sportsmith.
While these approaches may help address some issues associated with threshold-based counting, they often still combine acceleration and deceleration into a single metric. This risks obscuring the distinct mechanical and fatigue costs associated with braking actions.
Reducing a complex, continuous signal into a single value also risks overlooking important nuance within training sessions.
Practical Interpretation of Deceleration Load
Despite these limitations, threshold-based monitoring of deceleration load remains the most commonly used approach in practice.
The key, therefore, is not to abandon deceleration monitoring, but to interpret it appropriately. Practitioners should:
be explicit about the thresholds and dwell times used
recognise how these choices shape reported load
avoid overconfidence in between-athlete comparisons
interpret deceleration load alongside other external and internal load measures
Deceleration metrics are best viewed as high-level indicators of exposure, not direct measures of tissue stress or injury risk.
Key Takeaways
Monitoring deceleration load presents unique challenges. Deceleration data are continuous, thresholds are inherently arbitrary, and load metrics are sensitive to individual capacity and methodological choices.
Recent work advocating for individualised thresholds offers promising direction, but does not eliminate the underlying complexities of deceleration monitoring.
For now, the most valuable approach is a reflective one: understanding what your deceleration metrics represent, how they are generated, and where their limitations lie.
In the context of a broader monitoring framework, deceleration load data can still provide useful insight, provided it is interpreted with appropriate caution.
Frequently Asked Questions (FAQs)
How is deceleration training load typically monitored?
Deceleration load is most commonly monitored by identifying negative acceleration events that exceed predefined thresholds for a minimum duration (dwell time). These events are then counted and often grouped into intensity bands to provide an estimate of braking exposure.
Why is monitoring deceleration load challenging?
Deceleration data are inherently continuous, but monitoring systems require practitioners to convert this data into discrete events. This process depends on arbitrary decisions around thresholds and dwell times, which can substantially influence reported deceleration load. In addition, measurement error with tracking technology can be at its greatest during high-intensity movements such as decelerations.
What are individualised deceleration thresholds?
Individualised thresholds express deceleration efforts relative to an athlete’s own maximum capacity, often as a percentage of their observed maximum deceleration. This approach aims to improve physiological relevance and reduce bias between athletes.
While individualisation may improve interpretation, it does not remove all limitations. Threshold cut-offs remain arbitrary, and measurement error in tracking systems tends to increase at higher acceleration and deceleration magnitudes.
Interested in learning more? Check out my course, in partnership with Sport Horizon, the Fundamentals of Load Monitoring. This unique course blends scientific theory with practical data viz application in Microsoft Power BI.
For more information on wellness questionnaires and athlete monitoring, check out the Global Performance Insights YouTube channel’s Load Monitoring playlist.
Found this content useful? Say thanks here 😊: Buy Me a Coffee