“… three days of being called out during the middle of the night for a wrestling match with half a dozen angry cattle.”

With the amount of “metrics” available to us today it’s all too easy to get led by the numbers, and while this huge amount of available data is extremely useful to cyclists and their coaches, there are hidden dangers of relying too heavily on some of these numbers. Many people like to see things in black and white, they want to know, by just how much, they are getting fitter, faster, stronger.

The obvious place to look for un biased answers to these questions is their training and competition performance data, because after all, the numbers don’t lie, right? Well not exactly, actually SOME numbers do lie and it’s therefore really important we know which data is valuable and which should be at best taken with a pinch of salt.

The majority of this false data comes from when calculations have been made using only a fraction of the information required to produce an accurate result. Let’s take the Performance Management Chart (PMC) in the Training Peaks software as an example. This particular chart is intended to plot three key pieces of information: Fitness Level, Fatigue Level and Form. The chart aims to do this using just the data from your cycle computer. Unfortunately, with such limited information the chart is very unlikely to produce an accurate picture using just that data alone, at best it gives a rough guide and in many cases, not even that. This is not a criticism of Training Peaks I use the software to manage the training diaries of all the people I work with. My point is just to highlight an example of a chart that shouldn’t be relied on as being accurate, without factoring in a lot more variables. Additionally, it is a specific chart which people have frequently asked me about.

To accurately understand our levels of fatigue, we would need to monitor our training data, sleep quality and quantity, nutrition, hydration, heart rate variability, levels of oxidative stress, levels of muscular inflammation and our daily life activity. This list is not exhaustive, but it’s clear that the data from our cycle computer captured during a training session is a long way from the actual data needed to calculate our fatigue levels. It is possible to plot most of these metrics on the performance management chart but you would need to be recording all that data (some of which would require blood tests) on a daily basis to create a truly accurate picture.

Realistically, such an accurate picture is just not required for over 99% of us but that said, if you are going to use a chart to form the basis of your training decisions (as I know some people do), then you need to be looking at more than just predicted training stress scores calculated by your cycle computer.

To give a real-world example, the chart assumes that for every day a training session is not uploaded, we have rested, therefore, fatigue goes down and form increases. Two of the cyclists I work with are farm vets, they have extremely physical jobs which often involves being called out during the night. Three days of no training sessions could mean three days of being called out during the middle of the night multiple times for a wrestling match with half a dozen angry cattle and getting very little sleep in-between bouts. Not surprisingly by the end of three days like this, our farm vets are feeling absolutely knackered! Yet without factoring in the additional life stress information, the performance management chart counts the same 3-day period as 3 days of “Rest”. You can see how this would massively impact the fatigue and form scores and make them unusable.

The PMC also plots the cumulative effect of the training stress caused by each training session. Whether using heart rate or power to calculate the levels of training stress caused by each workout, most cycle computers don’t know the difference between whether you are feeling good and taking it easy, or actually feeling completely knackered but still working really hard. The computer tends to treat a low value, whether that be a power or heart rate reading, as you are taking it easy, when that may not be the case at all.

Other problems with training stress calculations stem from using a specific threshold value to calculate them, the value used may or may not be relevant to the session in question. I explain these issues in my previous post here.

I realise this post may come across as being heavily critical of Training Peaks, it’s not intended to be, as a coach, I use the software on a daily basis and will continue to do so. The software has a lot of very useful features. This post simply serves as a good example to explain the limitations of looking at data in isolation. I’ve had number of people who have come to me for coaching support who had been misinterpreting some of the numbers provided whilst planning their own training.

I hope you’ve found this useful, you can sign up for free to receive my future posts by entering your email address in the "Connect with Us" box at the bottom of the page.

Dan Small, Mountain Goat Coaching