The principle
The best comparison is the one that answers the recruitment question.
Player A versus Player B is rarely a neutral question. The choice of role, cohort, time period and metrics determines what the comparison can reveal. Make those decisions explicit before ranking either player.
Define the comparison question
Begin with the decision the club needs to make. Are you comparing replacement candidates, different tactical options, emerging players or the current squad benchmark? The purpose changes which evidence matters.
Which midfielder is better?
Which candidate best fits a possession-dominant number-six role requiring secure progression and transition defending?
Record essential traits, useful supporting traits and constraints such as age, availability, likely fee and competition readiness. This prevents the analysis drifting toward whichever player has the most impressive headline number.
Compare compatible players
Place both players inside the same defensible comparison cohort. Use the same season, position logic, minimum-minutes threshold and metric definitions. A percentile calculated against different peer groups is not a direct comparison.
Keep the observation window aligned.
Rank both players against identical peers.
Use identical definitions and directions.
Show minutes and valid observations.
When players operate in different leagues or tactical environments, the comparison becomes a screening exercise rather than a definitive ranking. State that limitation.
Select role-specific evidence
Choose a compact metric set linked to the actions in the brief. Include both volume and efficiency where the relationship matters, and avoid double-counting several highly correlated measures as separate proof.
Progressive passes, final-third entries, carries and completion rates.
Expected assists, shot assists, key passes and penalty-area delivery.
Duels, success rates, interceptions and pressure-adjusted actions.
Expected goals, shots, touches in the box and finishing output.
Keep contextual fields beside performance
Age, minutes, team, position and market value do not belong on the same performance radar, but they are essential to the recruitment decision. Display them alongside the comparison.
Use several comparison methods
No single visual or score should carry the decision. Combine methods that answer different questions.
Raw values
Show the actual output and the size of the gap.
Percentile ranks
Show relative standing across metrics with different units.
Radar charts
Summarise profile shape using identical axes and cohort.
Ranked bars and scatter plots
Show distribution, trade-offs and whether either player is a genuine outlier.
Understand player percentiles → · Build a useful radar chart →
Challenge the numbers
Ask what could explain the statistical difference other than player quality. Team possession, territory, game state, league strength, set-piece duty, tactical instruction and teammate behaviour all shape opportunity.
Search for counter-evidence
- Does the stronger per-90 output come from a much smaller sample?
- Does one player operate in a team that dominates possession?
- Are success rates supported by meaningful action volume?
- Does video confirm the action quality suggested by the data?
- What weaknesses would make the player a poor tactical fit?
Run sensitivity checks by changing a sensible filter or metric. A conclusion that survives reasonable alternatives is more credible than one dependent on a single setup.
Build a transparent decision record
Finish with a concise comparison that separates evidence from judgement. A useful decision record should be understandable by someone who did not build the analysis.
- Brief: the role, constraints and reason for comparison.
- Cohort: positions, minutes, age, competitions and period.
- Evidence: the most relevant raw values, ranks and visuals.
- Interpretation: strengths, trade-offs and contextual explanations.
- Risks: missing evidence, sample concerns and counter-arguments.
- Next action: video themes, live-scouting questions or data requests.
Talent DNA lets you filter one cohort, compare profiles, rank metrics, create scatter plots and assemble those visuals into a shareable PDF report.
Compare players in Talent DNA →