In Part 1 of this post, we presented a simple “Game Speed Calculator” to help give coaches context to the intensity of skills-based training drills. The current post will discuss our second tool for your toolbox, the “TEI Calculator”. Following on from Part 1, Part 2 focuses on a new problem that arose after we began using this tool. We were pretty happy that our training was starting to reflect match intensity, and that our players were beginning to become accustomed to the intensity of the work. The idea of this intervention was that if the drills were intense enough, we wouldn’t have to prescribe traditional conditioning during the season, which players obviously loved. However, some of our staff were apprehensive that the drills were enough to make sure our players weren’t losing fitness, so we needed a solution.
The first option we had was to get our players to do the same fitness test (e.g. 30:15 IFT) that they had done in pre-season, but fitting a 20+ minute maximal fitness test into a 7-day turnaround was unlikely. A more practical alternative might be a submaximal test, such as the 4-minute YoYo, which only requires ~7 minutes of the session. These tests have been shown to be pretty strongly associated with their maximal counterparts, and are a more realistic substitute. However, to be accurate, these tests require some pretty controlled conditions, which got me thinking whether we could do better. The submaximal running tests are designed so that the activity profile of the test remains constant each time it is completed, but is that necessary? Our players always wear GPS each session, so is it really that important to control the running that they do within this test, or is the answer simply just to control for it? That way, every training session becomes a fitness test, and we can monitor athletes with no additional intervention required.
The concept of not having to change training in any way is one that really appealed to me, and definitely reflects my philosophy, which I’m not ashamed to say I’ve adapted from Kunu, a surf teacher in Hawaii (mainland name Chuck).
You might think it’s a bit weird to be drawing inspiration from a Paul Rudd quote in a stupid comedy movie, but it definitely applies. “The less you do, the more you do” – a simple idea that it’s all about getting bang for your buck. Sometimes you just have to “DO LESS”.
The Training Efficiency Index
The Training Efficiency Index (TEI) was created as a simple method of tracking changes in external work, controlling for changes in internal work, and is calculated using the formula:
Where x is a constant, derived as the average slope of the relationship between the log-transformed variables. Although there have been several attempts to establish integrate internal and external training loads in soccer and hurling, these papers were unfortunately let down by poor methodological and statistical approaches. As outlined by Dr. Dan Weaving and co., simple ratios are not sufficient, due to inappropriate scaling of variables and either end of the range of values. Similarly, these studies utilised a standardised test, which varied very little between trials (CV = ~5%), and therefore this ratio was simply reflective of the variation in internal load, compromising the use of a correlation analysis – not to mention that fact that if you standardise the movement profile, you might as well do a submaximal fitness test!
Getting back to the TEI, our research has found it to be pretty important which training load metrics are used in the development of this index. To detect subtle changes in the TEI, it is vital that the internal and external training load metrics are highly correlated, as it is deviations from what is expected that tells us whether the player is getting better or worse. For example, an increase in running output without a coinciding increase in internal load reflects a positive training outcome. In contrast, a higher internal load without a concurrent increase in external load is may be representative of a negative training effect.
Amongst our cohort (male rugby league players), we found that measures that incorporated speed, acceleration and body mass were more closely related to both session RPE training load (sRPE-TL), and heart rate loadings (r > 0.90). This made sense to us, given the acceleration-based activity profile of rugby league. This may differ for other sports, and I would imagine metrics such as PlayerLoad could be just as useful for indoor sports such as basketball. The attached “TEI Calculator” can be used to determine the best metrics to use for your athletes, and plots the TEI as a function of time. I recommend trying a few different combinations to see what works for your cohort.
It is also important that changes in TEI are considered within the context of the completed training load. As an example, data for the TEI and training load for two athletes over the first month of a pre-season training program are shown below, where the TEI is plotted as a Z-score. In this case, Athlete A appears to be responding well to the prescribed training load, resulting in an increase in training efficiency over this phase. In contrast, Athlete B’s response throughout this period is far less favourable, and it is in this case that an intervention might be required.
Does it work?
Although great in theory, if this metric doesn’t truly reflect training adaptation, then it’s pretty useless. In our initial paper, we found some pretty strong relationships between the TEI and changes in fitness, and some slightly smaller relationships with completed training load during a rugby league pre-season. These results are encouraging, and mean that we can track fitness between fitness tests, or in-season when fitness tests are not logistically practical. However, as you will see yourself when you plot your own data, the TEI is quite variable on a day-to-day basis. Originally we thought that this metric could be really sensitive to acute training status: that daily fluctuations could be because the player is feeling fatigued or sore. Unfortunately we have seen no association between daily TEI measures and perceived muscle soreness and fatigue. The day-to-day variability in HR-based TEI could be due to a number of factors such as hydration status, caffeine intake or ambient conditions, none of which we controlled for in our studies. From a practical perspective, we probably aren’t going to control for these anyway, so we might as well concentrate on what we know it is actually measuring – chronic adaptations to training load.
It is widely known that fluid losses resulting in hypohydration can negatively impact performance, and increase cardiovascular and thermoregulatory strain. The reduction in plasma volume associated with hypohydration results in an increased HR. A paper from 2014 attempted to quantify the actual impact of hypohydration on heart rate, suggesting that for every one percent body mass loss there is an average increase of 3 beats per minute. I vividly remember – because I nearly shat my pants – working on a simulated deployment operation in the middle of the desert with the UAE Armed Forces and having to explain the elevated HR data to officers and also discussing a similar thing with Arne Jaspers during the World Cup qualifiers on a hot, humid Bangkok evening. On both occasions it was not uncommon for me to see sweat losses of between 1 – 3 L per hour, obviously resulting in significant alterations to the HR readings. It is also known that the consumption of large dosages of caffeine (5 – 13 mg/kg) can result in physiological alterations, including a slightly elevated HR. While such high dosages are unlikely and unnecessary, it is important to consider its impact alongside that of hypohydration on HR data. When basing decisions using HR data, it is important that analysis methods used are robust to this variability.
When using the TEI, it is really important to be realistic about what we are tracking. Currently, using GPS, we don’t have the ability to accurately quantify the work required to complete non-locomotor activity such as kicking, jumping, tackling or wrestling. These actions are often high-intensity, and therefore contribute significantly to the load impose on an athlete, so any conclusions we draw of the TEI at the moment is limited to the running-based demands of training and competition. Secondly, when considering HR-based measures, it is vital that only “clean” data is used for derivation of the TEI. If the HR monitor drops out for any reason (I’m sure you all know how common this is, particularly in contact sports!), the GPS is still recording. Without reviewing the data, we may be drawing conclusions using 100% of the external loads of the session, but our HR monitor has only recorded for 60% of the time (see below), significantly compromising the results.
The sRPE-TL method is probably more robust for this reason, but is substantially less accurate. Your methods need to be robust to error – I have written R code that deals with bad data immediately, but if this is not possible for you it may be a case of manually checking HR traces before making decisions on the data.
– The TEI can be used to regularly monitor athletes’ training status.
– This technique is particularly useful for assessing athlete during the competitive phase, where fitness tests are impractical and a logistical burden.
– Changes in TEI must be considered in the context of the training load completed.
– Although indicative of chronic adaptations to training load, the TEI seems to be too variable to form the basis of daily training decisions.