Bundesliga Teams Could Be Using More Substitutions. But They Aren’t.

Compared to football, baseball and basketball, the “big three” American sports, soccer is a uniquely fluid game: no set possession length, a continuously running clock, a tiny number of actual scoring attempts and constant transitions that blur the line between attack and defense. That’s made it much more difficult to quantify. Lionel Messi is better than everyone else, and shots from close to the goal are better than attempts from farther away, but otherwise data-minded truths have been hard to come by. However, there are a few static aspects of the game that have proven much easier to pin down and resemble those of other major sports. One of them is substitutions.

Per Stats Perform data, starting field players in the German Bundesliga were averaging 0.15 goals per 90 minutes this season before play was paused in mid-March due to the coronavirus pandemic. Substitutes, meanwhile, were averaging 0.28 goals per 90 minutes. In other words, substitutes were nearly twice as likely to score a goal than starters were. Despite that, only four of the league’s 18 teams had used all of their available subs in every game, and not a single team was averaging one sub in the first two-thirds of regulation.

The benefits of aggressively using your limited subs are obvious. A player with fresh legs will have an advantage against defenders who have been running around for 45 minutes. Plus, if players know they’re only going to play half a match, they could max out their effort, rather than conserve energy for the second half.

Yet most managers remain stingy with their subs, holding onto them until the last half-hour — even when they’re losing. Colin Trainor conducted one of the original public analyses of substitution effects and patterns, and he likened the in-game conservatism to what we see every Sunday in the fall: “I would guess the reason is that doing this will result in them sometimes being left short if a player gets injured. If that did happen, I would imagine the manager would take serious flak for having the temerity to buck the trend and go against common conventional thinking. So by being prudent, as managers currently are, it means that rarely do teams end up with less than a full complement of players. It’s kinda like not going for it on fourth down in the NFL.”

One way to test Trainor’s theory: Give managers more subs, which is exactly what the International Football Association Board has done for leagues starting up again, as the Bundesliga did on May 16. Because of the increased injury risk from playing a concentrated slate of matches after an extended period of time off, Bundesliga teams are temporarily allowed to make five subs per match. They’re still allowed only three separate substitution periods, though, so they need to make multiple subs at once to use up all five.

“It‘s a little bit less risky to make some substitutions earlier, like at the half,” said Bayer Leverkusen manager Peter Bosz. “If you make a three-player change that moment, you still have two more possibilities later on.”

But despite the lowered risk and the larger margin for error, Bundesliga managers aren’t changing their behavior nearly as much as they could.

Managers are making more subs, to be sure: Before the pandemic, the league average was 2.92 subs per match, and since play resumed, that number has risen to 4.29 subs per match. But the share of available subs being used has dropped from 97 percent to 86 percent. And perhaps more importantly, the increased personnel changes have been concentrated toward the end of the match. The average number of subs made per team within the first 60 minutes of a match has eked up by only a tiny bit: 0.62 to 0.88. Put another way, 21 percent of subs before play was paused came before the hour mark, which is slightly higher than the 20 percent of subs that have occurred in the first hour since play resumed.

Could it be that player quality declines so much as coaches reach deeper into their benches that the sub advantages no longer exist with your, say, 15th- or 16th-best players? Since mid-May, Bundesliga starters are averaging 0.15 goals per 90 minutes — same as before — while sub scoring rates have declined to 0.22. However, expected goals are more predictive of future performance, and subs in the five-changes era are averaging 0.25 xG per 90, which is still below their pre-stoppage rate of 0.32 xG per 90 but is a 67 percent increase on the baseline starter scoring rate. That’s still a ton of value to leave on the bench until the last 30 minutes. Rather than using the new rules as a potential strategic advantage, most managers seem to be using more subs just to prevent their players from getting hurt.

“The new option is necessary to avoid injuries because of the very intensive match schedule in the upcoming weeks,” said Bosz, who’s averaging four subs per match and even made a near-unprecedented three-man halftime sub in a 4-1 loss to Wolfsburg. “This really helps a lot. And this positive effect is the most important.”

The two best teams in the league per FiveThirtyEight’s Soccer Power Index ratings, Bayern Munich and RB Leipzig, are using only 3.75 subs per match since play resumed — the joint second-lowest rate in the league. Despite having two of the deepest benches in the country, neither one has decided to press the advantage. It’s worked out fine for Bayern Munich, which has won all four matches and now has a 99 percent chance of winning its eighth-straight title. But RB Leipzig has had a tougher go of it. Before play resumed, Leipzig manager Julian Nagelsmann hinted that his team would be approaching the post-pandemic matches differently: “These nine games are like the Euros, and we want to win them.” The team has drawn two of its four games, and it has used more than three subs in only two matches.

The worst substitution offender, though, is Wolfsburg manager Oliver Glasner. Before the pandemic, he was making 2.92 subs per match. Now that he has two extra strategic cards to play and a much higher risk of player injury to manage, Glasner has opted to start subbing … less often. He’s made a league-low 2.75 subs per match since play resumed — a rate that would have been second-worst in the league when teams were still only allowed to make three changes.

The anti-Glasner is Köln manager Markus Gisdol, who has used five subs in every game. Nearly half of them — nine of 20 — have come within the first 60 minutes. No other team has made more than seven first-hour subs. Köln hasn’t won a match since play resumed, but it has performed impressively: Per Stats Perform, Köln has the fifth-best expected-goal differential since the sport came back. Gisdol’s overall approach is working; his team has just been unlucky.

Beyond “Bayern Munich is gonna win again,” the return of the Bundesliga has proven one other thing: that we’re still a ways away from most top-flight coaches implementing earlier, aggressive substitution patterns. And we’re even further away from some kind of MLB-style opener strategy in which a player burns himself out over the first 30 minutes in search of an especially valuable first goal and then heads to the bench before halftime. The extra two subs haven’t changed much.

“We do not plan substitutions before a match,” Bosz said. “Substitutions are always reactive concerning the things which are happening in the game. There is no difference between three or five opportunities to our approach in this regard.”