The Best Team Is Not the Six Best Players

Selecting the strongest individual players can feel straightforward. A coach can compare throwing, catching, movement, experience, and general match output. Selecting the strongest lineup is a different problem. A dodgeball team is not six independent players whose ratings can simply be added together.

Players share the same balls, court positions, attacking windows, and tactical responsibilities. One player’s presence changes the opportunities available to the other five. A lineup containing the six highest-rated individuals may duplicate the same strengths, create competition for the same resources, or remove a quieter player who makes several teammates perform better.

Why combinations are harder to judge

A coach can watch only one lineup at a time. That lineup may face a stronger opponent, begin in a more difficult state, or play only a few sets. When another combination appears later, the surrounding conditions have changed. The outcome is noisy, and memory must compare performances that occurred at different times against different problems.

The number of possible combinations also grows quickly. A squad of twelve players contains 66 different pairs and 924 possible six-player lineups. Most will not be observed often enough for their quality to be obvious. Human observation processes those combinations in sequence. Analytics can organize them in parallel and compare recurring patterns across many matches.

Practical example

Imagine that Player A is the team’s best individual thrower. Player B has a modest hit percentage, while Player C ranks second behind A. If selection is based only on individual output, pairing A with C seems obvious.

Across many matches, however, A converts 50 percent of throws made alongside B but only 38 percent alongside C. The team also creates more event value when A and B are on court together than when either plays without the other.

Video provides a plausible explanation. B selects targets well, times the supporting ball, and covers A’s throws with convincing pump-fakes. C is individually stronger but needs many of the same balls and positions as A. B’s own statistics remain modest, yet their presence improves the output of the team’s best player. The stronger individual pair is not necessarily the stronger partnership.

Roles, resources, and fit

A balanced lineup may need primary throwing, catching, ball retention, retrieval, calling, defensive stability, and coordinated attack execution. Choosing six excellent primary attackers does not guarantee that those needs are covered. The players may compete for the same balls, wait for the same attacking role, or leave nobody responsible for organizing the possession. A perfectly synchronized two-ball attack from weak throwers may be more successful than an unsynchronized attack by two stronger throwers, which can approximate two single-ball attacks and result in a catch.

Complementarity is more than filling a checklist of positions. Two players in the same nominal role may combine very differently because of timing, communication, coordination, preferred targets, trust, or decision-making. Some players create value directly through their own hits and catches. Others create part of their value through better opportunities for the players around them.

What analytics can add

Combination analysis can begin with pairs. We can compare each player’s hit percentage when throwing alongside different teammates, the conversion of recurring throwing pairs, or the change in event points when two players are on court together rather than apart. These measures can reveal partnerships that consistently perform above or below what their individual statistics would suggest.

The same idea can be extended to larger lineups. A model can compare the team’s output with different combinations while accounting for opponent quality, player-count state, possession, and the difficulty of the situations they faced. The aim is to estimate whether a group creates more value together than would be expected from the players considered separately.

From a pattern to an explanation

A positive on-court difference does not automatically prove that one player caused the improvement. Coaches choose lineups deliberately, strong pairs may be used in favorable situations, and some combinations will have small samples. The analysis must therefore illustrate uncertainty and compare similar contexts wherever possible.

Even so, the pattern tells the coach where to look. Video can then reveal whether the advantage comes from target selection, ball distribution, synchronized timing, defensive cover, or another interaction that individual box scores miss. Analytics identifies the combination; coaching knowledge explains how it works.

Selection is not simply a ranking exercise

Individual ratings remain useful. A lineup still needs capable players, and no interaction model can rescue consistently poor performance. But rankings answer only part of the selection problem. They do not show whether several players require the same resources, whether a lower-rated specialist completes the unit, or whether one player raises the performance of everyone else.

Analytics should not choose the team automatically. It should help selectors compare combinations that are too numerous and noisy to judge reliably from memory alone, then return those findings to the court and the video for interpretation.