How you should be using GPS (Part 1): GAME SPEED

Throughout all of my blogs to date you will have probably noticed a bit of a trend – that I am pretty critical about the use of GPS in team sports. It’s all well and good for me to sit back and talk about all the things we do wrong, but without suggesting any better alternatives, I’m just another negative voice in the crowd. In my opinion, the use of GPS as a performance tool is extremely under-utilised, and this two-part post will talk about ways that we can actually influence training for the better.

Keeping in mind that we are here to help coaches and performance staff (not in spite of them!), the following is an example of a problem I was faced with early on in my career, the answer of which actually formed the bulk of my PhD work. I was working for a team that was struggling with salary cap issues and coaching movements, and we were left with a young and extremely inexperienced squad. I would regularly hear comments along the lines of “we are too slow”, or “we can’t match it with the fast teams”. Probably without even realising it, these people were referring to the concept of “game speed”.

Blog #7 - Figure 1

In my context, game speed referred to the ability to generate a high physical output during competition, without the significant compromise of the technical, tactical or psychological aspects of match play. To develop this capacity, training needs to reflect the intensity of competition, so that when faced with physically demanding periods during a match, the players are prepared and therefore these situations will be less catastrophic. Whether you want to term this as an aspect of “tactical periodisation”, or simply training specificity, basically we are referring to the use of game data to accurately prescribe, monitor and program training drills and sessions. Eddie Jones has been a big proponent for the application of this in team sports, and with the help of strength and conditioning coach Dean Benton, has had a great deal of success in applying these methods.

Working together with Dr Grant Duthie, who was working as a strength and conditioning coach with us at the time, we decided that our first step in helping coaches change the way we train was to identify what we needed to actually change it to. Commonly coaches would refer to “game speed” as ~ 90-100 m∙min-1, the speed that was generally maintained over the duration of a match. However, we know that fluctuations in intensity exist throughout a match, due to points being scored, stoppages for referee intervention or sometimes injury. This led us to a moving average analysis, to identify the most intense periods of a match (below).

Blog #7 - Figure 2

The figure above shows a 5-min moving average, which is really useful for comparing to drills lasting 5 minutes, but not much help for everything else. As a result, we quantified the peak running intensity of competition for a range of durations, between 1 and 10 minutes. Initially this involved analysing speed data for rugby league, but was later extended to a range of other variables including our average acceleration metric. We were able to repeat this process for a bunch of sports, including AFL, rugby union and finally soccer, where we established a framework for prescribing and monitoring training drills based on the most intense periods of a game. For each sport we would end up with a table that looked something like this:

Blog #7 - Table #1

As you will see from this table, the speed values are much higher than the 90-100 m∙min-1 I referred to previously, but when you think about it, is 180 m∙min-1 really that fast? When converted to m∙s-1, this equates to just 3 m∙s-1, well below top-level players’ maximal aerobic speed. When using GPS, we are only quantifying the movement profile of the activity, so the limiting factors in these methods are always going to include the contribution of non-locomotor activities such as tackling, wrestling, kicking or jumping. However, following on from my last post regarding acceleration in team sports, I really do believe that it is the constant stopping and starting that is important to accurately quantify – for time-dependent training drills such as small-sided games or high-intensity skills drills, the average acceleration metric does a great job.

Blog #7 - Figure 3

Admittedly having only recently been introduced to this concept I’m still getting to grips with it. However, looking at the data points coupled with an understanding of exercise physiology you could hypothesise that the rapid depletion of the immediate phosphagen stores of energy (ATP and PCr) during the initial periods of exercise and the necessary shift to slower metabolic pathways are responsible for the marked drop in intensity from a 1-min window to a 2-min window. Limited amounts of ATP (~223 mmol) and PCr (~446 mmol) are stored in the muscle and are used up rapidly at the onset of intense exercise. ATP initiates the initial movement through direct actions on myosin; PCr then rapidly donates a phosphoryl group to resynthesize ATP at a rate of ~73.3 mmol/s. When PCr stores run dry there is a shift to glycolysis and a concomitant decline in exercise intensity as ATP production drops to a rate of ~39.1 mmol/s, slowed by the enzymatic reactions required of glycolysis. Although the larger stores of glycogen (~6700 mmol) allow for maintenance of ATP production over a slightly longer period. You could postulate that this presents itself in Jace’s data, I could be wrong. Either way this concept has certainly helped us in developing and optimising our model of nutritional periodization as we can estimate the specific demands and fuel sources on each day based on the data.

Matt Jones

I think it’s important to consider that while the decline in running intensity for short values of time are likely due to physiological fatigue and depletion of short-term energy sources, this is not necessarily the case for longer values of time. During team sports, all players are limited by the same stoppages (points scored, referee intervention etc.) in the same way, which is why we see a pretty similar rate of decline in running intensity across positions and players.

This table and graph presented above are pretty useful as a fast reference point to see generally how each training drill fared, but logistically was a bit annoying to constantly refer back to, particularly when differentiated by position. After a quick consultation with a mathematician, we found that using a simple power law analysis, we could quantify the relationship between running intensity and duration, revealing a simple formula for determining game speed for any value of time:


Blog #7 - Equation #1


Where c is the intercept of the linear relationship between the two log-transformed variables (intensity and duration), t is the duration of the drill or period in question, and n is the slope of the model, or the rate of decline in running intensity per unit of time. The details of these calculations can be found in our paper, but the constants used for estimating game speed will depend on your sport, the playing level of the team in question, and the manufacturer of the GPS devices in use. We presented some basic guidelines for the four Australian football codes (rugby league, rugby union, AFL and soccer) at a conference a while back, but I would recommend establishing your own thresholds using your own data.

The attached spreadsheet, the “Game Speed Calculator”, does this for you at an individual level, where you can see how real data can be used to form this equation. By inputting peak values derived from competition (that can be calculated simply in excel at this level), you can now estimate game speed for any value of time. It doesn’t matter what the variable is you choose, whether it is speed, acceleration, high-speed running, metabolic power or PlayerLoad – it makes no difference. This calculator will give you a bit of an idea of what the most intense period of competition looks like, for any duration. As such, the above formula can be built into any GPS database to give coaches and practitioners fast feedback on how the running intensity of training compares to that of games. However, if individual or positional thresholds are going to be developed, I highly recommend establishing a more comprehensive profile of an entire season’s (or multiple seasons’) worth of data to be confident in the results. Such analyses are far outside the realm of the Game Speed Calculator I have attached and require a little more know-how with some more advanced data processing techniques.

There are a couple of things to note when using this methodology. Firstly, to generate game speed in training, you need to have a playing group and coaching staff that have bought in to the concept. For example, if you have programmed an up-tempo session aiming for 70% of the session at or above game speed, your goals will be quickly compromised if a coach decides it’s time to intervene and spend several minutes lecturing the players. This buy-in hinges on regular feedback to players and coaches, but when players understand that they can maintain fitness in-season without necessarily running conditioning drills every week, you should have no problems.

Secondly, training at game speed is specific to the peak intensities of competition, but if you need to improve a certain physiological quality (running endurance, speed, change-of-direction ability etc.), you will likely need to isolate that quality so it can be overloaded. Manipulation of the training drill (field size, duration, number of players) can help to direct that drill in one of these directions, but traditional conditioning techniques are likely to achieve these improvements more quickly. Once these capacities have been developed, we can bring them together in game speed drills. If a player cannot sustain the required intensity during such drills due to fitness, they may benefit from further individual development to “build the engine” so to speak, as to ensure they are not taking away from the quality of the drill overall.

Lastly, I would suggest that these high-intensity sessions cannot be delivered day-in day-out for an entire program – coaches will need time to coach and teach initially, before players are able to perform the skills at high intensity. Over time, as the team becomes more component and able to complete what is being asked of them from a coaching perspective, these drills can form a larger component of the team’s overall training program.

2 thoughts on “How you should be using GPS (Part 1): GAME SPEED

  1. What an insightful & practically useful article. Thank you. I am very interested to understand the mathematics as well as apply the principle to the training & match data of the Premier League Academy cohort I work with. Please email me when you get a chance to discuss this topic further. AJP

  2. Ross Brosnan

    Hi Jace,
    Thanks very much for the really informative post.
    I am head of a State AFL Academy at the minute and would love to know more about the calculator – how best to dissect the drills and how best to communicate it with players and coaches.
    Since the start of the season we have been breaking up the drills into three different categories for load moitoring purposes and to give the coaches a gauge on where their drills are at from a m.min, hsr%, max vel, and td perspective.
    Would love to chat more if you ever had the time.
    Thanks again for the great information

Comments are closed.