SCIENCE OF CYCLING: Mastering Load – Part 3. Training Load and Injury – A Cycling Coach Perspective.

19.May.21 | Bikefit, Physiotherapy

In the first blog of this series it was mentioned that when it comes to fit there has been more of a focus on performance than injury prevention. This is also true for the load modeling that occurs in the sport of cycling; training load is managed more to optimize performance than prevent non-traumatic injury. In this blog we will investigate the potential that properly managed training load has the serendipitous effect of also helping to prevent non-traumatic and overuse injuries in cyclists.

Performance load modeling was first developed by Banister et al in 1975. Their model was based on heart rate (HR) during training bouts. Later, training load models that used power as a metric to determine training stress were developed and used Banister’s model as a base. This method of modeling training load for cyclists via power data vs. HR became viable mainly due to two factors:

1) HR correlates with the power output for an individual at a given training status and

2) the advent of portable power meters that cyclists could utilize to track power output during training and racing scenarios- outside of a lab setting.

Power based training load models have attributes that could be argued advantageous in modeling for injury prevention. 1) While not scientifically validated in most cases, these training load models are commercially available for cyclists and coaches to use and can be easily accessed via the internet. Couple this with the automated data upload that many cycling computers offer and data management becomes very easy. 2) Modeling training load with power better represents overall mechanical load vs. using HR which more a measure of internal stress.

Of the commercial training load models, Performance Management Chart (PMC) developed by Dr. Andrew Coggan and marketed via Peaksware is one of the most popular.

Figure 1. shows the training load modeling for an athlete over the time span of one year.

Figure 1- Training load modeling for a cyclist for one year. Each red dot represents a workout and its corresponding training stress. The blue line represents fitness, the purple line represents fatigue, and the yellow represents freshness.

The Coggan method of modeling training load relies on some basic sport science ideas that have rebranded. The first idea is the idea of “threshold”. With the Coggan method “threshold” is defined as the maximum power output an athlete can maintain for 60 minutes. It is officially named Functional Threshold Power (FTP). Another concept unique to the Coggan model is the idea of normalized power. Normalized power (NP) was developed mainly because the average power (and kjs of work) for a ride or race doesn’t necessarily correlate to its physiological stress. This is especially true with rides/races that have high amounts of intermittent bursts of power. NP and FTP, along with workout duration are used to calculate an athlete’s training stress score (TSS) for a particular workout/day. TSS is the underlying data point that is compiled and utilized in the Coggan model and used to calculate Fitness, Freshness, and Fatigue for an athlete.

Fitness, know as Chronic Training Load (CTL), is calculated by an exponentially weighted rolling average of the last 42 days’ TSS. It shows up as a blue line in PMC. An athlete’s fatigue is known as Acute Training Load (ATL) and is again calculated with an exponentially weighted rolling average, but instead it defaults to the last 7 days. ATL is a purple line in PMC.  Freshness is calculated by subtracting ATL from CTL. This is called Training Stress Balance (TSB) and is illustrated with a yellow line in PMC.

Training load modeling has allowed for coaches and athletes to break away from classical and block periodization and still see gains. When these schemes are often employed an athlete can go a whole or multiple microcycles of training (e.g. a week or more) without returning to a “fresh” state. One could argue this might lead to a higher chance of overtraining syndrome and illness. On the other hand, cyclists utilizing a functional non-linear periodization scheme (as seen above in Figure 1) coupled with training load modeling, can potentially be in a “fresh” state once or twice a microcycle and still see gains. This helps to ensure that stress from training load is not too high too soon or for too long.  This is especially important for cyclists who are building back fitness after coming back from time off of the bike (e.g. after traumatic injury, health issues, seasonal breaks).

In Figure 1 notice that there is a quick and drastic decline in CTL in the middle of the graph. Also note that there are a group of TSS data points clustered around zero on the x-axis. In reality this athlete had a bad crash in a race that forced them off of the bike. You can see their CTL dropped to the levels they had in January. Interestingly, their interval power numbers matched the numbers they were putting out during that time. You see the athlete’s return to fitness over the course of the next month or so. One thing to also point out about this particular incident is that the athlete’s FTP was not retested/reentered into PMC after the crash/loss of fitness. This would have decreased the accuracy of the model. But, as many practitioners know, best practice for an athlete does not always include testing them. In many cases it is better for the overall scenario if testing can be avoided.

In rugby, cricket (Gabbett 2016) and athletics (Raysmith 2016), an increase in injury risk has been shown with spikes of increased load, troughs of unloading (Drew 2015), or unloading due to an initial injury, and inadequate chronic load (pre-season training). It has been suggested that the chronic load (long-term): acute load (short-term) relationship is extremely important in training load injury management.

In conclusion, at this time the idea that training load modeling can help prevent cycling injury is merely a hypothesis based on evidence from other sports, anecdote and reasoning (and some may say it is actually more conjecture). We do not believe this is problematic for the obvious reason that anecdote drives research questions and therefore our understanding of sport.

Cycling coaches are in the unique position to have the data required to monitor training load at their fingertips, so in recording data there is no cost or burden to the cyclist above and beyond something they would normally do to increase performance. If injury prevention is an unexpected side effect of training load modeling for performance then we might already have the answer to mastering load in cyclists.

Jason Boynton, M.S.
Postgraduate Student– Exercise and Health Science
School of Exercise and Health Sciences
Edith Cowan University

Want to know more?

Join us at our Science of Cycling course on 28 & 29 May 2021 for our LIVE Online lectures, or simply watch all the lectures and workshops online.

Email for a registration form.