New from smarterscout: Coaching profiles
For roughly a decade now, football players have been proifled using data. Dashboards, radars, and their like are common currency in professional scouting and social media. Yet the same kind of analysis hasn't been available for coaches – until now.
We've put together a profile that highlights some of the most distinctive traits of the playing styles that head coaches prefer and our own measures of their effectiveness. Here's an example, for Ole Gunnar Solskjaer's first full season as permanent manager of Manchester United:
Notice that the charts on the top and in the middle are split by home and away matches. Let's start in the top-left corner and work our way down:
Formations. Solskjaer favored a 4-2-3-1 formation but occasionally used three at the back, both home and away.
Average seconds to recover possession in open play. We measured the time between possessions in open play to see which clubs got the ball back the most quickly. Solskjaer's teams regained possession in less time than the majority of Premier League teams, especially in away matches.
Attacking style. We measure directness by looking at the total distance the ball travels during a possession versus the shortest distance to goal. A club team that passes and dribbles 100m to get 50m closer to goal is less direct than one that uses just 80m to make the same progress. We also look at the frequency of aerials in attack. Solskjaer's side was among the least likely to play in the air and was just a touch below average in terms of directness.
Defending style. We look at what happens when a team has the ball in its own half and then compare how likely they are to play out of the back versus clearing the ball or going long. We also measure the height of their press by checking how far up the pitch they tend to recover (or try to recover) balls when out of possession. Solskjaer's squad was among the most likely to play out and pressed higher at home than away.
Game flow and average substitution times. We track when teams generate expected goals in both of our models – shot creation and ball progression – and then average them minute-by-minute across all matches. Manchester United's attacking threat tended to peak around 65' under Solskjaer, and their defending was softest on either side of the break. Their attacking was more dangerous at home as a rule. Solskjaer's substitution times were similar home and away.
Average xGD. We compare expected goal difference across both halves to see how a coach uses his resources over the course of matches. Solskjaer's teams performed better in second halfs, notably in away matches, suggesting he was able to make adjustments and/or motivate them to finish strongly.
Average points. We want to see whether coaches can maintain and/or turn around results using the players at their disposal. Despite their superior expected goal difference in second halves, Manchester United still gave away more points than they gained in the closing stages of matches. The net cost was about four points over the course of the 2019-20 season, but they may have been rather unlucky in this respect.
FB touches. Manchester United's fullbacks rarely came inside to the middle of the pitch under Solskjaer, and his RBs were more adventurous in attack than his LBs.
W/WM touches. By contrast, Solskjaer's wingers and/or wide midfielders did come inside quite often in the attacking half. His right-sided wide players tracked back rather more than the left-sided ones.
Shot locations in open play. Manchester United's shots were spread fairly broadly across the penalty area and even beyond, with a lot of shots taken from wide angles, particularly on the left side. These shots usually have a low chance of scoring.
xGF effect per match. As explained in our article on black coaches in global football, we predict how well a club should perform based on the previous output of its players, then compare our predictions with actual results to see whether the head coach may be having a beneficial effect. Solskjaer's teams seemed less effective in attack than we predicted – about 0.15 xG less per game across our models – so we assigned a low probability of a beneficial effect.
xGA effect per match. By contrast, we estimated that Solskjaer's teams conceded 0.23 xG less than expected per match. This was virtually certain to signify a beneficial effect in comparison to the levels of performance that we expected from his players.
So, data-driven profiles aren't just for players anymore! Of course, we don't claim that these metrics and charts can tell the whole story about a coach, just as we wouldn't for a player. Yet they can start a discussion and help us to ask some of the right questions.
Overall, we were a little surprised to see that a former striker like Solskjaer would firm up Manchester United's defending more than he helped their attacking. But it's undeniable that he adopted a high-pressing style that helped to recover possession efficiently, giving his young attackers a chance to use their speed and hit back at the opponent from advantageous positions. And that youth and speed – which often came on in the second half via players like Mason Greenwood and, later in the season, Dan James – may help to explain why the Red Devils' attack seemed to peak later in matches.
We'll continue to refine the profiles, which are available for every head coach in the 48+ leagues we cover, going back to the 2017 or 2017-18 season. If you'd like to purchase one, please contact us using the link below. Thanks, and we hope you enjoy.