The Revolution of the Measurement of Metabolic Power in Soccer

Couple of days ago a close friend of mine and head strength and conditioning coach for FC PARTIZAN Miša Filipović gave me a link to one article from Roberto Sassi’s website. The name of the article is La rivoluzione della misurazione della potenza metabolica nel calcio edited by Roberto Colli, Emanuele Marra, Cristian Savoia, and Vito Azzone. 

Miša suggested me to use Google Translate which I did, and with some effort I have read the article and I had one of the rare ‘A-HA’ moments lately. I have always found that velocity-based time-motion analysis was lacking something. For example, suppose that athlete accelerate for the 5m distance. His speed reached would be around 18km/h or less (someone corrects me if I am wrong) and thus would not be calculated in the High Intensity (Sprint) category although the power-output of that movement is tremendous and very important in soccer. In my opinion, power output is more important than speed, both external/mechanical and internal/metabolic. Thus we need to take all three into consideration: speed, acceleration and calculated power output. This can give us a lot of new and more useful data. I am contemplating about this and trying to develop KPIs (Key Performance Indicators) that could be measured using simple GPS device and utilized to evaluate game/drill performance, load and specificity from physical standpoint. 

The La rivoluzione della misurazione della potenza metabolica nel calcio edited by Roberto Colli, Emanuele Marra, Cristian Savoia, and Vito Azzone  gave me  a lot of new ideas and insights, so I decide to contact coach Roberto Sassi and ask him for a permission to publish it on my blog. I have used  Google Translate to translate the document form Italian and although hard to read it is manageable. Tables are pictures, thus they are not translated, but within the context of the text they are pretty straightforward to understand.

When you read the translated article, also check the following one that deals with the methodology behind this approach: 

Med Sci Sports Exerc. 2010 Jan;42(1):170-8.
Energy cost and metabolic power in elite soccer: a new match analysis approach.
Osgnach C, Poser S, Bernardini R, Rinaldo R, di Prampero PE.
Source
School of Sport Sciences, University of Udine, Udine, Italy. cristian.osgnach@gmail.com

Abstract
PURPOSE:
Video match analysis is used for the assessment of physical performances of professional soccer players, particularly for the identification of "high intensities" considered as "high running speeds." However, accelerations are also essential elements setting metabolic loads, even when speed is low. We propose a more detailed assessment of soccer players' metabolic demands by video match analysis with the aim of also taking into account accelerations.
METHODS:
A recent study showed that accelerated running on a flat terrain is equivalent to running uphill at constant speed, the incline being dictated by the acceleration. Because the energy cost of running uphill is known, this makes it possible to estimate the instantaneous energy cost of accelerated running, the corresponding instantaneous metabolic power, and the overall energy expenditure, provided that the speed (and acceleration) is known. Furthermore, the introduction of individual parameters makes it possible to customize performance profiles, especially as it concerns energy expenditure derived from anaerobic sources. Data from 399 "Serie-A" players (mean +/- SD; age = 27 +/- 4 yr, mass = 75.8 +/- 5.0 kg, stature = 1.80 +/- 0.06 m) were collected during the 2007-2008 season.
RESULTS:
Mean match distance was 10,950 +/- 1044 m, and average energy expenditure was 61.12 +/- 6.57 kJ x kg(-1). Total distance covered at high power (>20 W x kg(-1)) amounted to 26% and corresponding energy expenditure to approximately 42% of the total. "High intensities" expressed as high-power output are two to three times larger than those based only on running speed.
CONCLUSIONS:
The present approach for the assessment of top-level soccer players match performance through video analysis allowed us to assess instantaneous metabolic power, thus redefining the concept of "high intensity" on the basis of actual metabolic power rather than on speed alone.

I would like to thank Roberto Sassi for giving me a permission to publish this article on my blog and also thank to Miša Filipović for pointing me to it. The article gave me a lot of ideas of how to develop a system that can be used to analyze and compare various training drills and track training load imposed on the athletes. Anyone interested in the ideas can contact me anytime. Maybe I can write an article but I don’t want to give much away. Anyway, enjoy the article!

The Revolution of the Measurement of Metabolic Power in Soccer.

In number 8 and 9 of the magazine Science & Sport is a bomb: prof. Of Prampero and his collaborators have developed a method which, though still with a margin of error, finally allows operators to go into detail of what happens during a football game (from now on for brevity we will call DP system). We all know by now that in a game we have a lot of acceleration and deceleration, followed or preceded by phases of walking or running at medium-low speed, but we had never been able to quantify indirectly the power delivered by these variations. For years and even today, continue to refer to the mileage or how far does the player when it goes over 22 km / h. Even these indexes have been subject to evaluation of the shape of the player himself: in practice do more meters at that speed, the more you are in shape.

Just to think that to reach that speed from a standing start a player takes almost 2 seconds and develops its maximum power between 10 and 16 km / h. So if for some tactical reason he arrived at 21 km / h, he developed metabolic power for two seconds more than 50 watts / kg or about 3 times its VO _2max, but no one would notice because it did not exceed 22 km / h. The absurdity of this situation is that the so-called "science" that has been accepted and is part of the validation study. Ergo all the studies that have used this method should be thrown in the trash and it would be also interesting that the referees did a self-publish such studies, but we know that will not go well.

Since the scientific world to us when it comes to physical education and training there has ever convinced of all, we took the work of prof. Of Prampero et al. and we tried to understand first of all the steps necessary to get to the power calculation, and then if with GPS systems in our possession we would have obtained the same results as providing out as a model, we also investigated whether some simple training methods often used in football would be states corresponding to the model or what would be deferred.

Please note that in order to read this article must necessarily read the article in Italian, signed C. Osgnach etc.. in the journal Science Sports & n 8 and 9, to which we refer to us as the nomenclature, and we assume that both known and studied with care in particular.

The first step was to analyze whether the data proposed by prof. Of Prampero in his 2005 study on the energy cost of running uphill and to what extent they were applicable to GPS systems. In practice, we got some 'of young players in the Spring, we have equipped with GPS and we did go through a series of sprints of 10 and 20 meters in line and shuttle, in addition we asked him to do well on the same distances activities submaximal. In order not to make too heavy this popular article we propose some tracks of those races that we believe are most enlightening: 

Figure 1

 

As we see from Figure 1, the data we have seen on the single player during a sprint show that tends to have the same acceleration as measured by Di Prampero (in the key: DP) on the sprinters, but after 10 meters is quite unable to increase the speed , presumably because they are rarely in the game reaches such high peaks as its acceleration takes place on short space and his pace is dictated by the position of the ball and not the clock. This in a very similar has happened to all the guys we've tested. In practice, the acceleration produced, this guy could reach a top speed over 30 km / h but in fact is not trained to develop coordination and muscularly. 

Another interesting fact seen from the graph shows more variability with the acceleration of the sprint compared to the GPS instead of Prampero is perfectly parabolic. The latter course is the result of a regression and can be read as if the guy ran on a track, while we know that we have a period of flourishing in the race where we speed (when we have your foot on the ground) and decelerated due to a phase where the component gravity during the flight. No wonder then if we see a 5 Hz sampling the "ups and downs" of acceleration. In order to compare the data with that of Professor. Of Prampero we have to do that we develop a regression parabolic "cancel" all the ups and downs inherent in the movement of the race.

We believe, however, it is very interesting to see the live performance of the power that tends to rise in the first 5 meters reaching a peak of about 70 watts / kg and then back down when the guy speeds up to a much reduced compared to the first meters. From here we can assess how much more expensive race in the first 10 meters (52 W / kg average) when the speed is still medium-low, than when the boy is now launched in the next 10 m (38 W / kg average). 

Beyond the scope of this article we have studied this phenomenon, but also with GPS at 15 Hz, but due to the amplification of this phenomenon accelerative we preferred to use a 5 Hz sampling, having comparatively evaluated at the end that the data were extrapolated total very similar. This thing but you can not do with the GPS 1 Hz, where in fact you can not properly evaluate the acceleration.

Figure 2

 

If we analyze the GPS is also a shuttle 20 +20 meters (Figure 2) made ​​by the players we see some other interesting thing: the slowdown happens in a very short space ± 2-3 meters, however, preceded by a speed control, so basically the subject tends to maintain speed after about 10 meters will have to stop because he knows that and therefore not advisable to go too fast. If you offer this tutorial to a sprinter you will notice that they always try to reach high speeds, with disastrous results in braking and restarting that occurs in a clumsy and slow.

So the eccentric braking is much faster (and cheaper) positive acceleration. This is very important for several reasons, one of which, as indicated by the same authors, is that the system they used to calculate the power only works up to a negative acceleration of about -5 m / s ^2. As we shall see later, braking at this intensity are not so many in football, the brakes at the prevailing average intensity and then the error of this evaluation is bearable.

Of course at this point we realized that the GPS can help us a lot of situations during the evaluation, because this way we can also evaluate whether or not necessarily linear with change of sense, measures speed, acceleration and power on winding roads with CDD (changes of direction).

We invite all those locations to film-test, because it is much easier than the problem of a lack of speed is to be ascribed to a wrong posture which increases the difficulties in coordination rather than muscle or worse metabolic deficiencies.

We hope that the price of GPS is considerably lower, especially in addition to providing the right software for reading the metabolic power, the parameter is still unknown to those who trade in this area.

The second step was to analyze some games with GPS systems to see if the data obtained were similar to those obtained with video analysis. We know perfectly well that the GPS can be worn only in friendly competition and non-official, so we believe that the best way to analyze the game is definitely the video analysis, while the cheapest way (although it seems strange, the video analysis is at all ' now very expensive) to control the training appears to be the GPS technology, that after the initial expense, it costs nothing if not hard to analyze the data.

Table 1 below shows the comparison between our data collected on some batches of a category Spring (30 surveys) are very similar to those obtained from the study of prof. Of Prampero, with a slight difference of greater commitment in the middle of our players.

We continue to gather data on friendly matches also to follow the evolution of pre-season and then this data is subject to further (but we think small) changes. 

Table 1

This comforted us and then we drew very similar conclusions about the race, quite comparable to those of the study carried out with the video analysis. We are pleased to point out, however, some key points of this work, which shows that the above based on high-speed turns out to be a huge mistake for any metabolic evaluation of the commitment of the players.

Following the proposals of the DP, we have not done is rebuild the table from him with the proposal getting the GPS data from Table 2 below:

Table 2

• If we evaluate only the erroneously speeds higher than the MPA, or MAS (> 16 km / h) those dealing with "only" 4.3% of the total time;
• the player develops the 14.3% of the time to a power higher than the metabolic MPA (maximal aerobic power set 20 W / kg approximately 57 ml × kg ^-1 × min ^-1 O _2);
• energy expenditure (EE = Energy expenditure) depends on the actions by 42.4% higher than the MPA, compared with a time over the MPA of 14.3%;
• average metabolic power during the game is 12 W / kg corresponding to 34.3 ml × kg ^-1 × min ^-1 oxygen equity;
• energy expenditure (EE) of the game is 65 kJ / kg for a player of 70 kg is 4550 kJ approximately 1100 kcal;
• for over 50 minutes, the player tends to walk (up to 6 km / h) but only for 2-3 minutes is completely stopped.

Table 3

 

Having now the DP system that allows us to calculate the power, we see how futile the attempts of manufacturers, software and video analysis of the "scientists" to give an indication of intensity, with the speed.

In Table 3 we have simply converted the various areas of the power of a player as if it ran at constant speed when up to 16 km / h there is a certain similarity between the data of speed and power (except for 0-6 km / h), we realize that more than 16 km / h ratio is likely to differ completely, so that with power above 35 W / kg there is no speed that justifies it. Therefore, we must think that what has been done to date using the speed beyond a certain threshold varies from company to company and "scientist" to "scientist", is worth very little to indicate the effort made ​​by the player. 

If in the 70s because of the distance traveled in the game, corresponding to about 10-12 km with about 1000 m over 20 km / h has helped us (not at all actually!) To understand that football is played variable speed, now these data are no longer needed anything, moreover obtained with very high costs.

We have set ourselves some trouble by coaches such as: how intense action takes place? How long and is distributed primarily as the recovery? 

We then performed on these data, further study is believed to be very useful as well as the original. 

Table 4

It seems interesting to note (Table 4) as 62% of the shares over the MPA (> 20 W / kg) is used up within 2 seconds, after these 2 "the player moves in the most" aerobic "of his performance, catching effort done, but very briefly. In fact, we note that the switch between aerobic and anaerobic action occurs with some frequency and with an average recovery of 10 ", although it will be more useful to see the distribution of the timing of the recovery as the standard deviation appears to us highly variable. Only 6.6% of the shares for a period lasting more than 6 "and it is interesting to note that when stress is prolonged beyond about 2" average power output increases and tends to also increase the recovery time.

To make it even clearer, between the maximum duration of these actions over the MPA, only 0.8% lasted more than 10 "and the maximum found was 14 seconds.

Figure 3

Entering the details of recoveries (Figure 3), we find that the actions over the MPA are repeated every second within 5 "and this is an extremely important exercise. It is plain to everyone that a player, as long as action is concerned, tends to occupy the space and then doing intense actions separated by short breaks up to avoid excessive reliance on anaerobic glycolysis and greater use of creatine phosphate (CP ) that can be restored in seconds. Then when the ball goes into another area where it is less interested, the player takes her longer breaks. 

Many other things we've noticed are extremely useful to address the training with knowledge of the facts now and not when cloning methods from other sports or have references only physiological. The data that still further surprised us is that even for a while that 'we believed to be closer to the model performance as the RSA (repeated sprint ability), is totally bogus as a construct for football. In fact:

• actions up to 6 "are very rare and especially are never as straight;
• in the first repetition of the RSA, the aerobic system is reaching a value around the baseline standard of the game around 2 ^to -3 ^to rip.: this leads to further high anaerobic energy production by a broad and without reason (for model) early production of lactate;
• in the test from rest while almost all the actions in the accelerative kick start at average speeds of 5-12 km / h;
• in the recovery phase of the RSA provides breaks in football still while you walk or run medium-low speed and actions are also high and medium intensity.

It is therefore clear that as the assessment systems we have to find something that is more relevant to the effort even from a biomechanical point of view, and above all, that does not lead to exhaustion on a football player (who looks good in the race from achieving it), making production large shares of lactic acid that can never be achieved in the game.

For many years (2001) had pointed the finger at the video detection systems that were not even very expensive more than confirm the studies by Thomas Reilly of the '70s, giving only the values ​​of displacement and speed. We also published an article on Science & Sports where he indicated the importance of the detection rate of 5 Hz GPS to display the correct acceleration.

With the availability of these new games, however, we tried acceleration and deceleration sifted because they also provide an important reference model can then be compared with a workouts.

First we consider a very important theoretical and practical aspect that comes from Figure 4:

Figure 4

 

The acceleration is not absolute because if the subject is very low speed (0-8 km / h) speeds up very well at 5-7 m / s ^2, but if it is at 18-22 km / h will be its acceleration only 2-3 m / s ^2, although the ceiling. We must be able to read during the game, and together the two values ​​to determine whether the player is speeding up with two very different rates.

The chart tells us that even if it increases the maximum speed reached by the subject, the slope of the line between speed and acceleration does not change (the equation of speed acceleration is always to Prampero PE).

That's why we used this index in the game maximal acceleration dependent on the speed and we got this table (5): 

Acc <50% max	80.1%
Acc> 50% max	19.9%

Table 5

Virtually in the game grow 20% acceleration of more than 50% of the maximum possible at any speed, while all others are moderate acceleration of action. Many tests are also possible on the brakes, but for now we believe that this simple chart (6) can help us understand what happens during a game: only 5% of the player maximum braking deceleration (<-5 m / s ²⁾ and many more (37%) are the average deceleration important but not ceilings. 

Please note that the concept of the relationship between speed and acceleration (Figure 4) there is in the braking, meaning that when the subject decides to stop in the shortest possible time, he can apply even at high speeds (≥ 20 km / h) significant deceleration order -10/-12m × s ^-2 (see Chart 2).

In fact, braking occurs at the same speed much faster acceleration (eg braking from 20 km / h can be done in 500 ms and accelerate up to speeds of 20 km / h in ≈ 1 "5 -2") . 

-5 <Dec <-2 m / s 2	37%
-12 <Dec <-5 m / s 2	5%

 Table 6

Another interesting parameter concerns the changes of direction (CDD), their size and relationship with the power and speed. The CDD (Table 7) are present to a very high (about 1000 with angles above 30 °) and over 800 are at angles> 30 ° but also developed in powers higher than 20 watts / kg. In practice, we develop a CDD> 30 ° and W / kg> 20 every 17-18 ". Those who foolishly thought to satisfy the training by making a road race should meditate at length. 

Table 7

We believe it important to report, the CDD has the power (since we can now calculate it) that with speed.

Table 8 shows that CDD fairly open angles (30 °) there are about 17% of shares over the MPA which decrease below 10% if the angle of the CDD is more closed than 90 ° and therefore more challenging.

Table 8

 This means that we must not neglect the CDD with small amplitude because very often it is an action of high intensity and, above all at high speed.

In this case we find solace in Table 9 in relation to the speed of CDD: CDD when we do a more than 90 ° usually the speed is very low (55.6% within 4 km / h) because obviously we have to stop to do this action.

Table 9

 

With more open angles up to about 60 ° we can do some CDD also at higher speeds, but still up 15%, while the speed with which they often carry out the CDD at these angles is between 4 and 8 km / h.

 

COMPARISON OF MODEL WITH SOME TRAINING EXERCISES

 

We firmly believe that the final issue of the use of GPS is to control what happens during training and as we move away from the model in different exercises, this is to identify and classify, especially for exercises with the ball, which have a greater correlation not only metabolic but also and above all, coordinative and muscular with the game. 

We begin to analyze a very dry year used by coaches and that is the stretch that we describe briefly:

• stretch carried out in line over distances ranging from 20-40-60-80 m "and go back" with speeds ranging from 100% to 60% of max;
• before and after the walk distance were made ​​5 m stretch of a slowdown;
• micropause 15 "-20";
• macropause 1'-1'30 ".

In this tutorial we have 30 surveys. It is clear from Table 10 that stretch well built, in addition to not meet the metabolic, suffer from a noticeable lack of situations with CDD and especially braking intense than the game. 

Table 10

The table (11) of the distribution effort in the areas of power, we note that the amount of time devoted to the actions of intensity above 35 W / kg is much higher than that race (10.5% vs. 4.3%), not due to acceleration, then the player tries to maintain high speeds for an extended time over 16 km / h, in contrast, we note that the intensity of mild (where walking or standing still), is considerably higher than the competition, so as to have almost 30% of the exercise while the player stopped playing this "phase" lasts more than 3-5% of total race time.

Table 11

 In terms of energy expenditure (EE) the area mean power is hardly affected by this exercise. In practice, there is an impressive work that will appeal to anaerobic glycolysis with lactate production as this broad activity is more prolonged than 2 "-3". Would therefore be more advisable instead to reach very high speeds over long distances, vary much more often the speed (and direction) on the line, in order to induce braking and CDD, also focusing on recovery situations where the player has not completely stopped (as in this tutorial) but, while walking, still make activities. With a few tricks (sticks, cones etc.. To carry out CDD) and braking situations faster and more intense at the end of the race, we can better approach the model, thus giving a stimulus to strength and coordination, not only the appearance metabolism. So, should be introduced in situations stretch of variations in acceleration and deceleration with CDD, but also runs at average speed would bring more dry this tutorial to model metabolic, muscular and coordinator of football; notwithstanding the lack of the ball!

Ball possession drills

• Have been proposed various kinds of ball possession with different technical and tactical objectives (goal, goal, etc.);
• we actually had 10 vs 10 in a space of 60 x 65 m;
• work consisted of 2-3 sets of 2'-4 ';
• the break between sets was 60. "

We can see that in ball possession drills we analyzed the metabolic is small, about 70% of the power race (Table 12). 

Moreover, the number of CDD> 30 ° and held above 20 W / kg is only 30% of what happens in the game. Even when it comes to muscle and coordinative efforts, intense accelerations are approximately 40% of the race and even fewer decelerations intense: about 25% of the race.

Table 12

The average distance traveled per minute by the player does not even reach the 100 m / min and this fact, combined with the low-intensive actions, we say that this exercise does not have the minimum requirements of the model and must be restored.

The way this is done exercise develops a EE by 38% over the MPA against 42.4% (race), but also greatly improves gait phase bland compared to the game and this is partly due to the lack of involvement by the player's ball is played as it almost ranks nearly complete on a small field, so these exercises by developing a commitment to lower metabolic (Table 13). 

Table 13

 

It is already likely that a simple increase in the area of the field, or conversely the decrease in the number of players, this could change in a positive exercise, but this must be verified.

In addition, the duration of the same number made ​​more intense, they are absolutely to be stretched even though we are convinced that during training, if you want to develop the model, we must be at least somewhere in VO ₂ above the competition. 

Exercise vs. 10 of 10 matches field 60 X 65 (door to the limits of the penalty area)

Almost all workouts end with a round of small field that probably meets the favor of the players and takes about 15-30 '. Very often it is free from technical constraints or limited by touches of the ball. This activity continues with up to 1-2 small breaks of 60 "to give indications of tactics.

Often people think that if we play during training, we get an attitudinal change, coordinating and metabolic by more players. This, if you can get the parameters of the race are partly met (Table 14).

Table 14

 

The trend is that the match instead of the end of training is often developed on surfaces much more modest, where every single player in that race itself should cover about 300 m ² field, sometimes it has to cover only 100-150 m ^2. This changes a lot as we see from these data (table 15), where we note that the metabolic is very low (similar to ball possession), even the muscle actions and coordinated as acceleration, deceleration, intense and CDD are not even half of those that are developed in the game.

To confirm this, we note that in these final training games the players tend to take breaks especially the recovery much longer than the race. 

Table 15

When the players receive the ball or the ball is on their side, have a commitment similar to the metabolic power above the competition, producing 39% of the MPA, however, is significantly penalized phase at medium speed, when the player lost the ball, must run to cover the space that separates him from fellow defense. Conversely, to support the attack (but if space is limited there), requires this same commitment.