Scouting & recruitment
The scoring rate that changes when penalties are removed
A worked recruitment example shows why a headline ranking is only the beginning of a player assessment.
Removing penalties changes this shortlist. In the worked example below, the attackers’ different goals-per-90 figures become equal non-penalty scoring rates. The arithmetic is straightforward; the recruitment implication needs care. It reveals how much the original comparison depended on penalty responsibility. It does not tell us which player can perform the role a Nordic club needs.
All goals and non-penalty goals give different comparisons
| Player | Minutes | Goals | Penalty goals |
|---|---|---|---|
| Player A | 900 | 8 | 3 |
| Player B | 1,800 | 10 | 0 |
Goals per 90 minutes. Shared scale from 0 to 1.
Rate = goals × 90 ÷ minutes. Non-penalty goals exclude the supplied penalty count.
Keep the football question inside the calculation
If the club wants an attacker to finish chances created in open play, penalties deserve separate treatment. A player’s penalty responsibility can be valuable, but that is a different part of the recruitment brief. Non-penalty goals can still include free kicks and other set pieces, so this adjustment does not isolate open-play finishing. Mixing those contributions makes it difficult to judge the ability the club is actually buying.
The example assumes the supplied minutes and goal classifications are accurate. It does not adjust for team strength, opposition, shot quality, or a player’s role. Equal non-penalty rates therefore narrow one comparison rather than settling the decision. A rate based on limited minutes also remains sensitive to what happens in the next few matches.
A league move changes the situations to be observed
For a Ghanaian attacker considered by a Nordic team, the next step is to inspect the situations behind the goals: starting positions, service into the box, time available for the finish, and work when possession is lost. Observe the intended role in the receiving team before deciding which of those situations transfers. A winger asked to receive wide against a set defence faces a different task from a forward running into space behind it.
Passing totals become more revealing when separated into the actions the receiving role needs: passes through pressure, switches, and delivery into the final third. Count the situations available as well as those completed. A team circulating possession may create different opportunities from one attacking quickly after a regain. Retain that difference instead of treating every pass as evidence of the same ability.
A forecast deserves its assumptions beside it
Oliver Craven’s conference presentation explores how player changes affect a modelled team. He also discusses limitations in its defensive modelling. That is useful recruitment practice: an attractive predicted effect should travel with the situations the model represents poorly. A model output is a reason to investigate a player, not permission to overlook contradictory footage.
The most informative shortlist is therefore a set of testable judgements. Keep the headline rate, the adjustment that changed it, representative clips, and unresolved questions together. For African talent with uneven data coverage, make the missing information explicit and seek another form of evidence where possible. A thin export and a weak footballer are different conclusions.
Interrogate the shortlist
- Separate the contribution the club needs from other goal types.
- Inspect the playing situations behind the rate, including unsuccessful actions.
- Keep model assumptions and missing competition coverage beside the recommendation.