A player ranking is only as honest as the weights behind it. This page lets you set those weights yourself â goals, assists, minutes, defensive contribution and discipline â and watch an eleven-player squad reorder in real time. Every score is arithmetic you can inspect, and the table exports as CSV so you can defend the selection in a meeting instead of gesturing at a black box.
How the score is calculated
Each player has five raw inputs. Before weighting, every input is normalised against the squad maximum for that input, so a striker's 18 goals and a defender's 9 tackles are both expressed on a 0â1 scale. The weighted score is then:
Dividing by the total weight keeps the score on a stable 0â100 scale no matter how you distribute the sliders.
That single decision â normalise, then divide by total weight â is what makes the tool useful for comparison. If you only multiplied weights and summed, moving one slider would rescale every score and you could not tell whether a player improved or the whole scale shifted.
What each input means here
- Minutes rewards availability. A brilliant player who has played a third of the season should not outrank a reliable starter on identical per-90 output.
- Goals and assists are raw counts, not per-90 rates. That is deliberate: this model measures contribution to date, not projected talent.
- Defense is a single supplied number â tackles, interceptions and blocks combined by whoever entered the data. It is not a proprietary metric.
- Discipline is inverted: fewer cards and fouls score higher, so a clean record adds to the total rather than subtracting from it.
Choosing weights that match your question
The default 30/20/15/25/10 split is a balanced starting point, not a recommendation. Different questions need different weights, and the presets in the toolbar are shortcuts to two common ones.
| Question | Suggested emphasis | Why |
|---|---|---|
| Who starts up front? | Goals 45, Assists 25, Minutes 15, Defense 5, Discipline 10 | Output dominates, but availability still filters out part-season samples. |
| Who anchors the back line? | Defense 45, Minutes 25, Discipline 15, Goals 5, Assists 10 | Defensive work leads; discipline matters because a suspended defender costs the team. |
| Who is the most complete squad member? | Roughly even across all five | Balanced weights surface players with no weak dimension rather than specialists. |
| Who should we rest? | Minutes 50, everything else low | Load management is a minutes question, not a quality question. |
Reading the result honestly
A ranking produced by five numbers is a summary, not a verdict. Three limits are worth stating plainly:
- It cannot see context. Opposition strength, role, system and the quality of teammates are absent. A defender in a low block accumulates different numbers from one in a high line.
- It rewards volume. Because goals and assists are raw counts, a player with more appearances has more chances to accumulate them. Minutes is a partial correction, not a full per-90 adjustment.
- It is only as good as the supplied data. The squad here is an illustrative example. Replace it with your own numbers before drawing any conclusion about real players.
Use the sensitivity check that matters most: move one slider and see whether the top three change. If a small weight change reshuffles the whole table, the ranking is fragile and you should say so rather than presenting a single order as fact.
Exporting and sharing the model
The CSV download contains the current weights, every player's raw inputs, their normalised values and the final score, in rank order. That means a colleague can open the file, see exactly which weights produced the order, and disagree with the weights rather than with an unexplained number. The file is generated in your browser from the same state that drives the table and pitch â there is no separate calculation path that could drift.
Frequently asked questions
Is this a prediction of future performance?
No. It is a weighted summary of supplied season-to-date numbers. It does not forecast goals, injuries or form, and it contains no machine learning. Calling it a predictor would overstate what the arithmetic does.
Why normalise against the squad maximum instead of using z-scores?
Squad maximum keeps every value on a 0â1 scale that is easy to read and explain in a selection meeting. Z-scores would centre values around the squad mean, which makes a below-average player's contribution negative and harder to interpret alongside a simple weight slider.
Can I use my own squad data?
The current build ships with a fixed illustrative squad so the interaction stays focused on the weighting model. The CSV export gives you the exact schema â player, position, minutes, goals, assists, defense, discipline â if you want to reproduce the calculation in a spreadsheet with your own numbers.
Does anything get uploaded?
No. All calculation and CSV generation happen locally in your browser. There is no account, no server call for the ranking, and no storage of your slider positions beyond the current page session.
What does the pitch view add?
It shows the same ranking spatially, with circle size proportional to score and colour banding by third of the table. It is a second reading of one dataset, not a separate model, and it makes it obvious at a glance when a weighting choice pushes an entire line of the team to the bottom.