19 Oct Who recovers the ball best in LaLiga? It depends on what you divide by
Same 380 matches. Same 20 teams. Three metrics. Three different leaders.
On the graphic, “efficiency” is shorthand for recoveries per 100 opposition possession sequences. It describes a ratio, not defensive quality, and a sequence is not necessarily an attack.
Ask who recovers the ball best in LaLiga 2025/26 and you can give three answers, all of them correct:
- Athletic Club made the most recoveries: 2,427.
- FC Barcelona recovers at the fastest rate: 3.18 per minute without possession.
- Real Madrid make the most recoveries relative to what the opposition builds: 69.1 per 100 opposition possession sequences.
Same events, different denominators. This post goes deeper than the graphic: why the leader changes, how far each version can be trusted, and what it says about reading any football metric.
Three denominators, three questions
Total recoveries asks how many times you won the ball back. Per minute without possession asks how often you win it back while you don’t have it. Per 100 opposition sequences asks how many recoveries you make for every 100 sequences the opponent builds.
All three are legitimate, but they answer different questions. Here is the full ranking, sorted by total recoveries as on the graphic. In brackets, places gained (▲) or lost (▼) versus the total ranking (1 = best).
| Team | Total | Per minute | Per 100 opp. sequences |
|---|---|---|---|
| Athletic Club | 1 | 5 (▼4) | 7 (▼6) |
| Rayo Vallecano | 2 | 2 (=) | 12 (▼10) |
| Real Betis | 3 | 12 (▼9) | 2 (▲1) |
| Alavés | 4 | 7 (▼3) | 14 (▼10) |
| Getafe | 5 | 9 (▼4) | 20 (▼15) |
| Atlético de Madrid | 6 | 8 (▼2) | 3 (▲3) |
| Sevilla | 7 | 6 (▲1) | 16 (▼9) |
| Real Sociedad | 8 | 10 (▼2) | 10 (▼2) |
| Osasuna | 9 | 13 (▼4) | 8 (▲1) |
| Mallorca | 10 | 14 (▼4) | 15 (▼5) |
| Celta | 11 | 17 (▼6) | 6 (▲5) |
| Villarreal | 12 | 19 (▼7) | 11 (▲1) |
| Elche | 13 | 3 (▲10) | 9 (▲4) |
| Girona | 14 | 15 (▼1) | 5 (▲9) |
| Valencia | 15 | 11 (▲4) | 13 (▲2) |
| Real Oviedo | 16 | 16 (=) | 19 (▼3) |
| Espanyol | 17 | 18 (▼1) | 17 (=) |
| FC Barcelona | 18 | 1 (▲17) | 4 (▲14) |
| Real Madrid | 19 | 4 (▲15) | 1 (▲18) |
| Levante | 20 | 20 (=) | 18 (▲2) |
Between total recoveries and recoveries per sequence, the rank correlation is practically nil (Spearman ρ = 0.07, n = 20). Between total and per minute it is 0.34, and between per minute and per sequence 0.34 as well. With 20 teams these are descriptive figures, not a formal test.
Why the picture moves
Volume reflects exposure. Athletic faced 3,721 opposition sequences over the season, Real Madrid 3,154, about 18% fewer. A team that spends more of the match defending has more chances to win the ball, and the total reflects that. Madrid are 19th in total recoveries and 1st per sequence.
The clock isn’t neutral either. Barcelona spent just 685 minutes without the ball across 38 matches. Villarreal spent 1,125, Espanyol 1,114. Match by match, recoveries per minute correlate at −0.88 with minutes without possession, while raw recoveries barely relate to those minutes (−0.11). The rate depends heavily on how long a team goes without the ball, not only on how many balls it wins back.
In seconds: Barcelona spend about 19 seconds without the ball per recovery, the league average is 26, and Villarreal, Levante and Espanyol sit around 30. That is time out of possession per recovery, not the time it takes to win the ball back after losing it. For that you need event data with timestamps.
Is it just noise? Splitting the season
To check whether the leaders hold, I split the season into matchdays 1 to 19 and 20 to 38.
- Barcelona lead the per-minute ranking in both halves.
- Athletic lead total recoveries in the first half, Rayo in the second (Athletic are 2nd).
- Madrid are 2nd per sequence in the first half, behind Betis, and 1st in the second.
The per-minute ranking is the most stable between halves (rank correlation 0.73). Totals (0.42) and per-sequence (0.38) move more. And one note on scale: 16 of the 20 teams sit between 62.5 and 66.5 recoveries per 100 sequences. Madrid, Betis and Atlético stand out at the top, but the order in the middle of the table says little.
The denominator that looks smart and isn’t: opponent losses
A natural idea is to divide by the number of times the opponent loses the ball, since that is when a recovery can happen. The two turn out to be almost the same thing: opponent losses average 60.6 per match and recoveries 59.3, with a match-by-match correlation of 0.98, and the ratio sits between 0.97 and 0.98 for every team. The result barely moves because the denominator is practically the numerator.
Not every denominator adds information, and checking is part of the job.
The same lens on the other side: losing the ball
Does the leader always change? I ran the same test on ball losses (fewer is better): total losses, losses per minute in possession and losses per 100 own sequences.
Here the top doesn’t move. Barcelona have the fewest losses on all three (2,133 in total, 1.59 per minute in possession, 62.5 per 100 sequences) and Real Madrid are 2nd on all three (2,155, 1.74, 62.6). Athletic are the opposite: 20th in total, 16th per minute, 12th per sequence. The rest of the table still reshuffles (rank correlations between the three versions: 0.29, 0.01 and 0.10).
That is a useful contrast. A robust finding survives any reasonable denominator, and a fragile one doesn’t. Barcelona and Madrid look after the ball however you cut it. Who recovers it best is another matter.
What none of this tells you
These metrics don’t say where a regain happens, how soon after the loss it comes, or what it leads to, whether a shot, a chance or a goal. Those are different layers of analysis that sit on top of this one. Possession sequences are also not full possessions, since one possession can contain several, and not every sequence offers the same chance to win the ball back.
This post isolates one thing: how much the denominator alone can change what a metric says, even for something as basic as a recovery.
Three questions for any metric
- What is in the denominator? Time, sequences, matches, minutes with the ball?
- What question does it answer? Frequency, volume, share of opportunities?
- Does the result survive a different denominator? If it does, you have something. If it flips, you have a framing, not a verdict.
Method: Opta data (LaLiga EA Sports 2025/26 export), 380 matches, 20 teams. Team figures aggregate the whole season, as the sum of recoveries divided by the sum of opposition sequences or minutes without possession, not an average of match ratios. Possession time is read as minutes.