Tactical & Realistic Fit

Tactical and realistic fit is the half of the platform that asks not how good is a player? but how well does he fit us — and could we actually sign him? It has three parts that share one view of how teams play: a model of each club's playing style, a per-player tactical-fit prediction (how a player would rate in a given system), and a realistic-fits ranker that turns fit into a signable shortlist. A fourth, sibling feature points the same machinery at your next opponent instead of a transfer target.

In one line: "fit" and "level" are different questions. Level is how good a player is on a global scale; fit is whether his game suits how a particular club plays. A possession side and a long-ball side want different players of the same level — so we model each club's style, predict how a player would rate in that style, and only then apply the realism — right tier, right work permit, plausibly available — that makes a recommendation actually actionable.

How a club plays

Everything starts with a model of how a club plays. For each club and season we measure eight style dimensions — how intensely it presses, how much it dominates the ball, how directly it builds, how wide it attacks, how vertically it progresses, how much it leans on set-pieces, how high it defends, and how fast it transitions. Each is normalised so that the league average is the reference point, which is what makes styles comparable across divisions: a club rated above the line does more of that thing than its peers, wherever it plays.

A club is then labelled with one of seven recognisable archetypes — gegenpress, possession-dominant, direct, low-block counter, wide-attacking, high-line, or a balanced mid-block — by a short, readable set of rules over those dimensions, not a black box you have to take on faith. The thresholds were tuned against reality: when a famously high-pressing, high-possession side was briefly mislabelled as a mid-block, we adjusted them until the labels matched what any analyst would say. Every profile also carries a confidence tier based on how many matches it rests on.

Predicting fit — player by formation

Tactical fit predicts how a player would rate in a given shape or style, formation by formation. The naive answer — his average rating in the games he happened to play a 4-3-3 — is unreliable over a handful of matches, so we use a more careful estimate that shrinks a small-sample average toward a sensible expectation: partly the player's own career baseline, partly how similar players fare in that shape. The more games we have, the more the estimate trusts what actually happened; the fewer, the more it leans on the expectation. Recent form counts for more than old form, on a smooth decay.

Every cell carries an honest confidence label — managers respect stated uncertainty — and when a player has barely played a given shape, the estimate falls back to the most similar players who did, surfaced separately as "similar players' fit" so you can see where the number came from. Where the exact formation isn't known for a match, the club's overall archetype stands in as a coarser tactical setting.

From fit to a signable shortlist

Realistic Fits is the per-club ranked board of who to sign. It is fit-led: the candidate pool is players at clubs whose style matches yours, within a realistic band of one tier up or down, eligible to play for you, and above a minimum fit. The ranking starts from how well a player would suit your system and then folds in the things that decide whether a move is real:

• Tier realism. A sideways move, or a step up from the tier below, is always plausible. A step down is only shown for a player who isn't a starter where he is — a squad player who'd come for minutes. Anything more than a tier away is excluded as fantasy.
• Level and fit, kept separate. Fit asks does his game suit how you play? A second, independent figure asks is he actually better than who you currently start? — his globally comparable ability set against your league's typical starter at the position. A player can suit your style yet be below your level, or be a clear upgrade who doesn't fit your shape — so we show both rather than blending them into one misleading number.
• Age, scaled to your tier. A top-flight club can sign prime players; a lower-league club, which has to care about resale and trajectory, gets a progressively stronger bias toward youth and upside, using the trajectory forecast.
• Signability, softly. Players who look hard to sign — unavailable, or tied to a long contract — are nudged down, never dropped, and missing information counts as neutral, so that thinly-covered lower-league players aren't unfairly buried.

This is a different question from the main recommendations shortlist, which starts from your squad's needs. Realistic Fits starts from who suits how you play and is realistically signable. They share inputs but answer different questions — a full recruitment view uses both.

The same machinery, aimed at your next opponent

Pointed at the next opponent instead of a signing, the same style engine composes a pre-match profile from data the platform already holds: who's available, the probable starting eleven and shape, the main threats, how the side's style has been drifting lately, when in a match they tend to score and concede, their substitution patterns, and where their goals come from. It is read-only and on demand, and it states its own coverage caveats rather than papering over them.

How we keep it honest

We are candid about what is and isn't formally tested here:

• Style is rule-based and inspectable by design — there is no labelled "ground truth" corpus of playing styles to score against, so we made it readable and tuned it against sides whose style everyone agrees on.
• Fit uses standard, well-understood small-sample techniques, and every cell can be traced back to the matches that informed it — but a formal backtest that fit predicts a player's actual rating after a move is ongoing work, not something we claim is done. The re-ranking has been validated by working through real before-and-after club cases rather than on a held-out outcome set.
• The opponent profile is descriptive: it composes already-verified facts, so its honesty is about data coverage, which it shows on the card.

What it can't do

• Formation coverage is partial. The exact shape isn't recorded for every match, so some shape-specific fit leans on the club's overall archetype as a coarser stand-in.
• Fit is not level. The shortlist is fit-led; a high-fit player can still be below your squad's level. The upgrade figure is shown alongside precisely so this isn't missed.
• Style needs games. A club with too few matches gets an "insufficient" style profile and drops out of the matched pool — thinly-covered leagues are weaker here.
• Keepers are confounded. Fit is built on the match rating, which for goalkeepers reflects the team's defence as much as the keeper himself — the same caveat as the rating.
• Signability data is thin in lower leagues, so the availability and contract nudges are deliberately soft — they tilt the order, they don't decide it.

The research behind it

The methods are the standard toolkit for small-sample estimation and style characterisation.

• Efron, B. & Morris, C. (1975). Data analysis using Stein's estimator. — The shrinkage behind the fit estimate: pull a noisy small-sample average toward a pooled expectation.
• Cover, T. & Hart, P. (1967). Nearest neighbor pattern classification. — The similar-players fall-back for players with too few matches in a shape.
• Fernández-Navarro, J. et al. (2016). Attacking and defensive styles of play in soccer. Journal of Sports Sciences. — Characterising team style from normalised performance indicators; the precedent for the eight style dimensions.
• Decroos, T. & Davis, J. (2020). Player Vectors: Characterizing soccer players' playing style. — Representing style as a vector of normalised actions; the conceptual cousin of the archetype dimensions.