Rereading the 2024 J1 Scoring Charts: What Goals per 90 Changes

Léo Ceará won the golden boot with 21 goals. Sort the same seven players by goals per 90 and he falls to fifth. This is not a trick of the arithmetic — it is what a scoring chart was never designed to tell you.

What the totals column actually measures

A season goal total is a product of two things: how often a player scores when he is on the pitch, and how much of the season he spent on the pitch. Those two inputs answer different scouting questions, and a totals-only chart collapses them into one number.

For a recruitment department the distinction is not academic. Minutes played reflects fitness record, squad status, coach preference, rotation policy, and in some cases nothing more than the absence of a competitor in the same position. None of those transfer cleanly to a new club. Scoring rate, by contrast, is at least a property of the player — noisy, sample-dependent, but his.

The 2024 J1 season provides a clean illustration.

The outlier, and why shot volume matters more than the rate

Rafael Elias appears near the bottom of any totals-sorted list and at the top of a rate-sorted one. Eleven goals from 1,129 minutes across 15 league appearances is 0.88 per 90 — the best figure here by a wide margin.

The immediate objection is correct: short samples inflate. A striker who converts an unusual share of his chances over fifteen matches will regress, and the smaller the denominator the more likely the rate is describing variance rather than ability. This is the single most common way a per-90 screen produces a bad recommendation.

Which is why the more informative number in Elias's line is not the goals figure at all. He also led this sample on shots, at 3.75 per 90. That combination — high conversion and high volume — narrows the interpretation considerably. A player scoring at an elevated rate on low shot volume is a finishing-variance candidate. A player scoring at an elevated rate while also taking the most shots in the sample is at minimum occupying the positions where shots come from, which is a repeatable behaviour in a way that conversion percentage is not.

That does not make the 0.88 durable. It makes it worth a live viewing. The distinction between those two statements is most of what a screening layer is for.

Inversions inside full-season samples

The small-minutes case is the obvious one. Less obvious is that the ordering changes even among players with comparable, substantial samples — where the regression argument does not apply.

Ryo Germain finished second on total goals with 19 but ahead of Ceará on rate, 0.61 to 0.59, having played roughly 400 fewer minutes. Marcelo Ryan scored 14 in 2,174 minutes, a 0.58 rate that sits within a rounding error of the golden boot winner's, on about two-thirds of the minutes. Neither of these is a small-sample artefact. They are cases where the totals column ranked availability and the rate column ranked output.

At the other end, Yoshinori Muto's 13 goals came from 3,091 minutes — 0.38 per 90, the lowest rate here despite a higher total than Elias and Usami. A club reading the totals chart sees a double-figure scorer. A club reading the rate chart sees a player whose total is substantially a function of how much he played.

What this table does not tell you

Per-90 conversion fixes one distortion. It leaves several others untouched, and a screening method that does not state them is worse than no method at all.

Penalties are not separated. This dataset does not split penalty from open-play goals. A regular penalty taker's rate is inflated relative to a non-taker's by an amount this table cannot show.

No shot-quality measure. There is no expected-goals figure here. Shot volume is a partial substitute for chance quality, not an equivalent one.

Role and service are invisible. A striker in a team that generates high-volume chances and one in a team that does not are being compared on the same axis. Team context is doing work the number does not attribute.

League level is not adjusted. A J1 goals-per-90 figure is not comparable to a top-five-league figure without a conversion factor, and any such factor carries its own error. We do not apply one here, and no reader should apply one implicitly.

A further limitation is structural: this sample is the top of the scoring chart, which means it already excludes every player who scored at a strong rate in limited minutes without accumulating enough goals to appear. The most valuable names for a recruitment department are, by construction, the ones this table cannot contain.

A usable screening sequence

Read in the order below, a scoring chart stops being a ranking and becomes the first filter in a pipeline.

1. Convert before sorting. Never rank on totals. Rank on rate, then look at the totals column to understand why the rate exists.

2. Apply a minutes floor and keep it visible. We report below 900 minutes only with an explicit small-sample flag. Elias clears that floor at 1,129 and still warrants one — a floor is a reporting rule, not a validation.

3. Check volume before believing the rate. Shots per 90 separates a player finding positions from a player finishing well in a short run. The second regresses faster.

4. Resolve the rest by viewing, not by data. Penalty share, chance quality, service, and defensive contribution are questions for footage and for the club's own analysts. The screen exists to decide who is worth that time.