BLOG: A new way to assess aerial performance

Article by Todd Kingston

The 20th edition of Major League Soccer (2015) will be remembered for a number of reasons, one of which being Kei Kamara’s aerial dominance. Whether it was getting on the end of a cross or releasing pressure for his team in midfield, Kamara was instrumental in helping Columbus Crew SC to their most successful season since winning MLS Cup in 2008. By the end of the playoffs, Kamara had scored 12 headed goals, breaking the previous single season MLS record (10 by Eddie Johnson for the Seattle Sounders in 2012). 

Kamara’s strength in the air is easily recognisable; he is big, strong, has great mobility, leaping ability and timing. Needless to say, he is a difficult matchup in the air for most defenders. Quantifying this can be difficult, however.

Quantifying aerial ability

Kamara competed in a remarkable 242 aerials in MLS 2015 during the regular season, 48 more than any other player. While he won more aerials than anyone (137), he also lost 105; more than all but four players.

His success rate of 57% ranked 72nd in the league amongst players to have competed in at least 35 aerial duels. This simply doesn’t capture his ability well enough. As a result, OptaPro analyst Johannes Harkins and I attempted to better quantify aerial performance for players in MLS.

A different way of evaluating aerial strength

Not all aerials are created equally. First and foremost, quality of opponent in an aerial can vary substantially.

One way to compare players’ skill level in head-to-head competition is through the Elo rating system. Each player’s rating will increase or decrease depending on outcomes of aerials throughout the season. The difference between two competing players’ ratings will serve as a predictor of the outcome of the aerial that will determine the number of points added or lost in each aerial. As such, these ratings will help account for difficulty of opponent, but they also allow for additional context to be added in the formula.

Evaluating aerial strength: top 20 players in MLS 2015

Rank Player Elo rating Aerials Won Rate
1 Kendall Watson 1624.0 158 119 75.3%
2 Chad Marshall 1596.9 126 92 73.0%
3 Kei Kamara 1595.7 242 137 56.6%
4 David Horst 1592.9 160 112 70.0%
5 Steve Birnbaum 1590.6 148 106 71.6%
6 Omar Gonzalez 1583.6 159 109 68.6%
7 Atiba Harris 1576.0 93 67 72.0%
8 Clarence Goodson 1575.7 151 101 66.9%
9 Liam Ridgewell 1572.0 94 69 73.4%
10 Ryan Hollingshead 1572.0 99 66 66.7%
11 Matt Miazga 1571.3 166 111 66.9%
12 Damien Perrinelle 1568.0 131 90 68.7%
13 Aurélien Collin 1567.6 123 84 68.3%
14 Blas Pérez 1567.6 119 70 58.8%
15 Axel Sjoberg 1565.8 86 63 73.3%
16 Alan Gordon 1562.2 150 78 52.0%
17 Kevin Doyle 1556.4 138 72 52.2%
18 Andrew Jacobson 1556.2 101 66 65.3%
19 Juan Agudelo 1554.1 123 64 52.0%
20 Seb Hines 1552.6 97 63 64.9%

The 2014 regular season was used to establish initial ratings to use at the start of the 2015 season. All players were assigned an Elo rating of 1500 for their first aerial attempted during this two year stretch. 


Kamara fares well in this analysis for a few reasons; he is likely competing against the opposition’s best aerial player frequently. The average rating of his opponent in aerial duels was 1520, above average and impressive considering his volume of aerials attempted. Unsurprisingly the majority of his aerials were attempted in areas of the field that have lower success rates on average as well.

All graphics in this post show teams attacking left to right.

The size of the hexagons relates to the volume of aerials attempted. This graphic unsurprisingly illustrates that aerials are won more frequently in the defensive half. There are several reasons why this may be the case. Teams in the defensive half are likely to have more players in the area which allows better pre-aerial positioning and those players should also be goal-side, allowing them to see the flight path of the ball as well as their potential opponent.

As the primary goal within this article is to better assess aerial performance, I’ve accounted for field location in the formula’s expected outcomes.

I also found there was a significance to aerials resulting from a goal-kick or goalkeeper distribution (launches or punts). Teams playing the ball in this scenario tend to lose aerials more frequently than when defending the opposite scenario. An adjustment for this was included in the model as well.

Kamara wins a greater percentage than average in most of these difficult zones. On top of this, roughly 15% of his aerials were following Crew distribution from the back. His 57% success rate looks much better with this additional context.

While Kamara’s case reinforces this particular approach to evaluating aerial duels, perhaps more importantly, the Elo formula is a stronger predictor of aerial outcomes compared to aerial success rates or logistic regression using only our adjustment factors.

Caveats to consider and the contextual case of Bradley Wright-Phillips

Importantly, there are still some caveats to these ratings. We are only able to account for aerials actually attempted; which excludes players shying away from challenging for balls in the air they deem unwinnable. Also, we cannot account for the starting position of players, where stronger aerial players may attract more attention from the opposition.

One example where these ratings appear removed from intuition is Bradley Wright-Phillips, whose lack of aerial success appears to be contextual (Wright-Phillips ranks last in these ratings, scoring 1422.8). He has lost 264 aerials since the start of MLS 2014 while winning just 111 (29.6%).

Many of his aerials were the result of New York playing safe out of the back but perhaps it highlights an area of additional focus for the Red Bulls in 2016. For instance, looking at the wide midfield areas Wright-Phillips has won 52% of his 25 aerials attempted comapred to the central midfield areas in which he has won just 20% of 116 attempted. Wright-Phillips faced an average Elo rating of 1507 wide compared to 1512 centrally.

Next steps

A next step to further analyse and evaluate aerial ability would be to consider what happens after the initial aerial, establishing whether certain players are able to not only win aerial challenges, but also help regain or retain possession for their team.

The model can also be focused to bring additional depth by analysing set piece efficiency for instance, or can be expanded to evaluate teams based on their players’ weighted averages.

Although by no means a definitive answer, this approach certainly brings an additional required context to evaluating players’ aerial ability.

Back to analysis