Football analysis: Altitude and goal scoring  //  The effect of altitude on the 2010 World Cup

05 Jul 2010 Posted by
Amazing to think that we’re now done with 60 out of 64 matches, and by this time next week, the World Cup will have left South Africa.  It’s been a month-long celebration here in SA, and while the disruption to work and traffic and general life will be over, we’re bracing ourselves for the mother of all hangovers down here!

Fortunately, the Tour de France will be into its second week, and so the withdrawal will be minimal, and here on The Science of Sport, we’ll keep forging ahead with the next big thing!

Today, a look back at Round 1 at some really interesting statistics, first regarding the altitude and then the goal-scoring.

The effect of altitude on physiological performance

Before the tournament began, I did a series of posts on this issue.  This level of competition in a team sport like football rarely takes place at altitude, and so the impact on the game was the subject of much conjecture, with FIFA doing their best to deny that it would have any impact.  Many people felt it would impact on players, and  I hypothesized a number of changes in matches caused by altitude.  The overall summary is that players would cover less distance and also sprint less.  The impact of altitude would be to slow the game down – the comparison I gave is that footballers in top-level competition (Italian Serie A) run and sprint more than footballers in lower level (Danish) leagues, with less rest relative to work.  The altitude would have the effect of bringing the level down.

Now, thanks to Castrol’s sponsorship of FIFA, we can actually evaluate that hypothesis, at least partly.  For those who don’t know, Castrol are one of FIFA’s eight ‘middle level’ sponsors, and they have been doing technical analysis of matches as part of their sponsorship activation strategy (I won’t go into the details now).

So every match is tracked using cameras and all kinds of statistics are produced.  How far players run, how many passes they attempt and complete, tackles made and missed, number of sprints attempted and so forth.  I must confess that I can’t vouch for the accuracy, and I’ve tried to email them to find out more about how the data are collected.  I’ve yet to hear back, but I think it’s important to stress that there probably will be some error in the data – there always is.   However, I think the data they’ve produced is quite compelling, and as you’ll see, the differences are large.

Altitude – less sprinting, less high intensity distance

The table below is a summary of the data on distances run and sprints attempted during the Round 1 matches.  There were 29 altitude matches and 19 sea-level matches, meaning that there are values for 38 teams at sea-level and 58 at altitude (I’ve treated Nelspruit as sea-level, incidentally, at 600m).

The distances covered by teams in matches are looked at for TOTAL DISTANCE, DISTANCE AT LOW INTENSITY (walking and very slow jogging – the average speed in this zone is 4.5 km/h, which is walking pace), and DISTANCE AT HIGH INTENSITY.  The average speed in this zone is 16 km/hour, but in some matches, were more high intensity running was done, it gets up above 20km/hour.

Altitude-and-distances-covered

So, the total distance covered per match is reduced by 4% at altitude.  What is more interesting is that the low intensity distance is only 0.2% lower (statistically it is the same), but the distance covered at high intensity is down by a pretty large 11.5%.  The distance of 25,106m at altitude is 2,750m lower than at sea-level, and assuming that the goal keeper contributes negligibly to the high intensity distance, this equates to about 275m less sprinting and high speed running per player per match.  Of course, some players will be affected more than others, and so it’s quite conceivable that certain players have covered a kilometer less at high speeds at altitude.  That is a massive difference – about 30%, and may impact on match outcomes.

This is further reflected in the number of sprints – down 11.5% from 1,059 to 949 sprints per match.  That’s about 10 sprints per player per match less, which makes up most of that 275m per player difference.  Note that I’ve cut out the medium intensity running distance from the table – the distances there were 16,890m for sea-level and 15,819m at altitude, so again, there is a reduction in distance at altitude.  I’m emphasizing the extremes though.

The point then is that altitude slows the game down at the “top end”, reducing the distances covered in sprinting as well as the number of sprints by about 11%.  That is a significant difference.

What the data doesn’t allow us to do, unfortunately, is to decipher whether the reduction at altitude is produced entirely in the second half as players fatigue, or whether they slow down right from the onset and pace themselves differently.  I would lean towards the latter, and suspect that the distances covered are reduced at altitude from the very first 15 minutes of matches.

Individual match variability

I must also point out that there is a chance of a false conclusion here because although the differences are large, there are many confounding factors too.  For example, the match between Switzerland and Spain in Durban produced the most running of any match in the tournament.  The Swiss covered 119,909m in that match, of which 39,256m was at high intensity, at an average speed of 22 km/hour, and consisting of 1,484 sprints!  This was the most of any team in any match in the tournament (for interest’s sake, the highest total distance was by Australia in their match against Serbia – 121,506m).

However, not far behind this was South Africa’s opener against Mexico, but this was in Johannesburg.  There, SA covered a total of 118,856m, and 37,312m was at high intensity, with 1,690 sprints.  So here we have two games with similar distances, but one is at altitude and the other at sea-level.  To emphasize this point even further, consider France in their match against Uruguay in Cape Town.  There, they covered only 101,499m in total, with 25,320m at higher speeds.  So, that sea-level match had 50% LESS high intensity running than an altitude match…the graph below summarizes these three teams’ physiological profiles.

Comparison-three-matches

The point I want to emphasize is that the tempo of the match, and the distance covered is not solely determined by the altitude – the strategy of the teams and also the nature of the match influences it as much.  The France v Uruguay match was particularly “narrow” with little movement because of the strategy of the two teams, and so less running was to be expected, compared to the Swiss who chased and worked tremendously hard off the ball against Spain.  However, on the balance of 58 altitude performances and 38 sea-level performances, altitude has exerted a significant effect, 11%, on the nature of the matches.  Perhaps another 50 matches would remove this ‘randomness’, but I’m quite convinced by the finding.

It’s especially compelling, as I mentioned, if you consider that some players will be affected more than others, and with the possibility of 30% reductions in sprinting distances, the altitude has certainly had an impact.

When are goals scored?  Could altitude affect goal-scoring?

The next question related to this is whether the timing of goals scored differs between altitude and sea-level?  The reason one might hypothesize this is two-fold:

  1. There is already plenty of evidence that more goals are scored in the final 15 minutes of matches than in any other fifteen minute period.  In fact, over 20% of a match’s goals are scored in the final 15 minutes, according to a study by Armatas et al.  This has been attributed to fatigue at the end of the match, since players certainly slow down as the game progresses.  I suspect there is more to it than this – concentration, the state of the game, the intent of players as the final whistle approaches are all factors, but certainly, the final 15 minutes produces more goals.
  2. Altitude will have a greater effect on fatigue (and thus potentially concentration), and so if the trend is observed at sea-level, the theory might be that altitude exaggerates it even more.  That is, one might suggest that more goals will be scored in the latter part of matches at altitude than at sea-level.

This is of course quite easy to determine – you simply have to count the number of goals scored in each 15 minute period.  The tricky part is knowing whether it’s just co-incidence, because it’s very easy to read too much into it and come up with ‘patterns’ where there are none!

I’ve done this, counted all the goals scored, and I have this result.  However, it is for another post, because this one has given out enough numbers for one day, and besides, there’s work to be done!

So next time, we’ll look at when goals are scored, and whether there’s a chance that altitude affects it!

Join us then!
Ross