When Should a Soccer Manager Insert His Subs?

Nifty article in today’s Journal about a nifty study by Bret Myers of Villanova:

The pace and flow of soccer generally make it difficult for managers to affect the outcome of a match once it begins. Since soccer has almost no stoppages for coaches to draw on clipboards or strategize with their players, a manager’s most critical in-game decision may be choosing when to utilize their three substitutions. …

Myers analyzed the substitutions and ensuing results of every game played during the 2009-10 season in the top English, Spanish, Italian and German professional leagues, as well as the 2010 Major League Soccer season and the 2010 World Cup. He concluded that if their team is behind, managers should make the first substitution prior to the 58th minute, the second substitution prior to the 73rd minute and the third prior to the 79th minute. Teams that follow these guidelines improve-score at least one goal-roughly 36% of the time. Teams that don’t follow the rule improve about 18.5% of the time.

I don’t know if there’s much of an empirical literature about substitution for other team sports (this baseball study, e.g., is more about optimizing matchups). It does strike me that as much as “conditioning” is appreciated in sports, the role of fatigue — mental and physical — is perhaps underappreciated. Yet another reason to think about using an “opener” in baseball?

At the top of this column is a (poor) photo I snapped in Barcelona in December, of Lionel Messi getting subbed in the 54th minute in a Copa del Rey match against Bilbao. (Below is a photo of Messi in action in that same match.) The other subs: David Villa at 63 minutes and Adriano at 79 minutes — just right, according to Myers’s research. But it didn’t work out for Barca. Neither Messi nor the others provided much of a spark as a sub, and the match was a 0-0 draw. He was nevertheless wonderful to watch.


Jeffrey

If Barça wasn't behind, how is this related to the study at all?

(Side note: Barça are pretty much never behind!)

Newalgier

Can wel say selection bias? A stronger bench means earlier, and better substitutions. So the study really says that better teams are more likely to win. Yawn.

Greg Z

Due to its pace and flow soccer is really difficult to quantify. However, I have a possible (of course partial) explanation for the observations of Myers. The earlier a substitution occured, the more obvious it has been to the manager that it would probably be useful. The opportunity cost of a substitution at 50' is much higher than at 80', so the expected advantages of an early substitution should have been higher. I cannot explain why this has no effect on non-losing teams - so maybe I'm wrong.

Rick

A perfect example why soccer will never catch on in America. Soccer is not played by the book. There are no managers on the sidelines going through stats and books to decide their next move. Managers coach by gut feeling and their heart. They know their players and know what they want from them when they insert them into the game. No manager will ever read this study and use them in his coaching. Nor should they.

May Americans be kept far, far away from this beautiful game before they ruin it, like they have done with every major American sports, which are nothing more than TV commercials interrupted by some sporting event. The beauty of soccer is once the players are on the field they pretty much are on their own for each 45 min half. That is the way it should be. Imagine an NFL game with no coaches, no headsets to send signals and no interruptions in the game. That is were the boys would get separated from the men and the true geniuses from the poseurs.

Read more...

TysonF

Interesting but bookings (similar to foul problems in basketball), fitness levels and competitions (knock-out, group or league formats) probably play a bigger role.

I'd love to see some thorough stat breakdowns on football (soccer) strategies. Stats are historically a no-go area at least in the UK/ROI region. Managing is considered more an art form than equations

CMB

Oh yeah, Rick. Watching two middle of the table EPL teams keep each other from scoring for 90 minutes is more beautiful than a field of wheat waving in a breeze or a clipper ship with a bone in her teeth bound for the spice islands. On the other hand, Barcelona.

Vinay

Apart from when to substitute, it also matters whom do you substitute. E.g. do you substitute a striker with a mid-fielder or vice-versa. Does the study indicate such a possibility. Of course, if you are behind, you want to score as many goals as possible. But then, how do you go about substituting? Introduce a striker first, followed by a mid-fielder and followed by a striker again? Or what else is the way you substitute?

Spooner

O.K. Now do another study of the previous season and see if it confirms your results. Then, if it does, make predictions about the next season. If they turn out to be accurate, you may be on to something.

IrishMan

If your team is drawing or losing by 1 goal, then it's usually put on a striker or attacking midfielder for a more defensive midfielder. If it's 2 or more goals, then it's a striker or attacking midfielder for a defender.

There wasn't anything said about making multiple substitutions. Why send on one sub when you can send on 2 together? Two strikers for a midfielder and another striker? A striker and a midfielder for a midfielder and defender. There are lots of possibilities.

Mike B

It is bizarre that Association Football continues to underutilize its coaching staff to the extent it does. The main problem is that its chief controlling bodies are not the for-profit leagues, but non-profit, amateur originated governing bodies that put more emphasis on tradition than on making the sport a more exciting product.

The lacks of clock stoppages do more than make it difficult for fans to refresh their drinks or the league to generate revenue through TV adverts. The final 20 or so minutes of the games are little more than a farce as exhausted players struggle to move about the field.

The Gridiron Football, Hockey and Basketball have all realized that players need rest to be able to put on the best show of performance for the fans. Moreover players will play harder if they know an injury won't put their team at a permanent man disadvantage.

More opportunities for coaching strategy opens up a whole new dynamic to the game that helps further involve fans and provides for something that can be studied academically. Association Football would be much improved with the addition of more substitutions, time outs and a video challenge system.

I pray that one day FIFA's monopoly on Football will be broken by a USFL style competitor that tweaks the rules to produce genuinely more exciting games. Much like Baseball evolved in the 1920's with rules to make hitting easier and basketball evolved with the shot clock and 3 point line, Association Football stands poised for a revolution that will make everything before as the "boring ball era".

Read more...

Phil Birnbaum

There is a huge literature on pitching subsitutions in baseball. The idea is that you want to use your best pitchers in the most critical situations, a concept that is called "leverage".

An early study of mine (.pdf) is on page 7 here:

www.philbirnbaum.com/btn2003-02.pdf

But much of the work is the product of Tom Tango - he has a whole section on the subject on his website here:

http://www.tangotiger.net

You can also search for "leverage" on the various sabermetric blogs, such as "Inside The Book" and "The Hardball Times".

Finally, you can google "reliever leverage" for lots more.

AP

Wonderful to watch? It was 0-0!

Now if someone would just call me an ignorant American.

Aaron

If 2 of your subs are Messi and Villa, I'd suggest the earlier you get them in the better!

Also, like #1 said, if the study was based on a team coming from behind, then your Barca game fell outside of the applicable test.

Tigernik

This is an interesting study, but it's purely a "best case average scenario".
However, if it began to delve a little deeper, it could reveal some much more appealing insights.

For instance, what is the success rate per manager,
Within games where a sub (or series of subs) is required?
And is there a clear difference between manager success?

An answer or indication to this would reveal which managers are more tactically astute and can enable change, in a very difficult environment and situation.

Could it be that a slightly different set of insights appear? Such as those managers that change strikers early on; or to a more attacking formation reap better results than those managers less eager to change?
Two managers for this comparison immediately spring to mind.... Alex Ferguson (Man Utd) and Rafa Benitez (ex Liverpool, Inter Milan)

I would be shocked if Benitez came out on top!

Michael

Most subs in soccer are made around the 60th minute, 75th minute and 80th minute.

So is the researcher saying that the managers who don't adhere to these norms are usually bad managers?

And "prior to" is awfully useless wording from anyone pretending to be scientific. If I make 3 subs at the start of the second half, a move that's normally a bad strategy, it would fulfill the requirement of "prior to."

scott

@Mike B
"The final 20 or so minutes of the games are little more than a farce as exhausted players struggle to move about the field. "

Umm... no. Often as legs get tired opportunities open up. Have you never watched a match where a goal was under siege for the final minutes? Fatigue is part of the romance of the game, and to watch a tired team battle has its own charm.

"Moreover players will play harder if they know an injury won't put their team at a permanent man disadvantage. "

Umm... no. That is completely eclipsed by the knowledge that an injury will *injure them* and put them out for days, weeks, months or perhaps their entire career. Players avoid breaking their legs because they don't want broken legs.

Jim

TIgernik:

http://tomkinstimes.com/2010/04/sub-standard-rafas-subs-record/

Sean

I usually like the studies done on the blend of sports and optimal strategies but I'm sorry, this one holds no water for me. There are so many more variables that effect an outcome of a soccer game (goals scored) than when to sub in players.

Besides the aforementioned variable of which kind of player to sub in for whom, you also have to take in part of the chemistry in general on a soccer team. Certain players are much more team orientated and will pass more often while some have a selfish style of play. In sports such as basketball, hockey, and baseball especially, one player can take the ball or puck and score for his team basically single handedly. In soccer the field is much too large and there are too many opposing players for one single player to carry a team on his/her back. So subbing the right type of player should easily trump at which time the player is subbed in when it comes to effecting the outcome of the game (goals scored).

Also, soccer is such a low scoring event in which I'll say a significant portion of the goals are scored based on luck as compared to skill. If scores were normally in the 20's or above, then I'd say that lucky goals would become more of an outlier stat (side note: curious to see how many goals are "own goals" as a percentage to the total if anybody has that). Powerhouse teams like Manchester United lose to bottom feeder Wolverhampton (Feb 5th) when they have significantly more talent on their roster.

When I first saw the question my initial thought was to rate every player on your team from 1-100 and find out how that number changes throughout an average game (or specific game style), then when a bench player's attributed number is higher than a player currently on the field, sub them. I'd still stick with that strategy for the most part but throw in the chemistry thing as well... so basically create "unit ratings", and sub accordingly.

For me though, tracking soccer data is just so shaky due to the low scoring and high luck factor. Until they remove the goalies from the game or shrink the field and number of players, the correlation significance of data will always seem too low for me for most soccer strategies.

Read more...

Jason

One of the fundamental assumptions of this analysis is that soccer teams play to win and employ certain aggressive strategies/tactic - hence the side discussion of who to be subbed in, a striker or an attacking midfielder -. However, this is not necessarily true because some teams play to draw instead in recognizing that a draw might be more beneficial than winning the game at all costs in the long run.

Referring to #3 comment, I do not understand why "The opportunity cost of a substitution at 50' is much higher than at 80'". A substitution, judging from a tactical perspective, could mean a change in tactics by the side as the manager has either identified a flaw in his team or a flaw in the opposing team in which the new player can exploit. It does not matter whether a weakened player is being subbed for a stronger player or a strong player subbed for an equally strong player as long as the overall chances of winning/not losing by a big margin is improved.

That said, as much as I appreciate the effort spent on this analysis, I must say that there are too many variable factors in the world of soccer. In 1999 UCL Finals, who would have thought that Manchester United can score two goals against Bayern Munich in the extra time to win with a 2-1 scoreline? Even I cannot believe it when I am a ManU fan myself! Thus, I believe this phenomenon that you have observed may well just be a statistic correlation rather than organized, theoretical explanation.

Read more...

Nikki

A game where players of Messi's and Villa's caliber are available but out of the starting lineup is a low-priority game, and substitutions generally take place when spectators start falling asleep (and then you end up with a 0-0).

FWIW, in 2006 Champions League final, where Barcelona was first one goal behind, then won 2-1, the substitutions were timed as prescribed here: 46', 61', 71', and it was the subs who made things work.