The register was sealed before the season. This page reads it back against what has actually
happened, week by week, and does not grade anything. The bars below were fixed on 2026-08-28 and are quoted
from the sealed file; the realised numbers come from Understat, the source
addendum-01 designated for grading after FBref removed every xG statistic.
Nothing here is a verdict. Juventus have played 2 of 38 league games
(5% of the season, 180 available minutes so far).
Every registered player is below the register's 900-minute floor, so every row reads
Pending — which means not passing, not "on course". The register
grades at the end of 2026/27, in May 2027, and again after 2027/28.
The headline prediction
Randal Kolo Muani, non-penalty xG per 90
Sealed text: “Kolo Muani 2026/27 Serie A npxG/90 < 0.35 on ≥1,500 min. If ≥0.40 on ≥1,500 min, the register says we were wrong.”
Serie A minutes so far
180 of 1,500 required
Non-penalty xG so far
0.647 in 2 apps
npxG/90 so far
0.323
Marcel projected, sealed
0.147
Point test |realised − projected|
not evaluable — under the 900-minute floor
Status of the prediction
PENDING
The prediction is conditional on 1,500 Serie A minutes. He has 180. Neither limb is testable yet.
The sealed text confirms the call below 0.35 and falsifies it at or above 0.40, both
conditional on 1,500 minutes. Between 0.35 and 0.40 the register says nothing —
that gap is in the sealed file and is not closed after the fact.
A rate computed on 180 minutes is not evidence about a 1,500-minute bar. It is shown
because hiding it until it looks decisive would be the worse choice, not because it means anything yet.
Every registered player against the pre-committed bars
Serie A 2026/27, as of 2026-08-30
Minutes bar fills toward 1,500; the amber tick is the 900-minute floor below which the
register records a graded MINUTES failure. Share is of the 180 Juventus minutes available so far.
Pace is minutes × 38 ÷ games played — plain arithmetic, not a projection, and suppressed until matchday
8 (a display choice, not a derived threshold). The last column is the exact Understat row the
name was joined to, so a wrong join is visible.
Player
Min
To 1,500
Share
Pace
npxG
npxG/90
Marcel
Gap
Status
Matched row
Randal Kolo Muani
in · permanent · attacker · age 27
180
100%
from MD8
0.647
0.323
0.147
+0.176
PENDING
Randal Kolo Muani / Juventus
Guglielmo Vicario
in · loan · goalkeeper · age 29
180
100%
from MD8
0.000
0.000
—
—
PENDING
Guglielmo Vicario / Juventus
Zeki Çelik
in · free · defender · age 29
162
90%
from MD8
0.000
0.000
0.199
-0.199
PENDING
Mehmet Zeki Çelik / Juventus
Jhon Lucumí
in · permanent · defender · age 28
79
44%
from MD8
0.000
0.000
0.043
-0.043
PENDING
Jhon Lucumí / Juventus
Jonathan David
incumbent · n/a · attacker · age 26
61
34%
from MD8
0.000
0.000
0.483
-0.483
PENDING
Jonathan Christian David / Juventus
Kerim Alajbegović
in · permanent · attacker · age 18
29
16%
from MD8
0.000
0.000
—
—
PENDING
Kerim-Sam Alajbegovic / Juventus
Jeff Ekhator
in · permanent · attacker · age 19
0
0%
from MD8
0.000
—
0.216
—
PENDING
no Understat row — zero league minutes recorded
Kamil Grabara
in · loan · goalkeeper · age 27
0
0%
from MD8
0.000
—
—
—
PENDING
no Understat row — zero league minutes recorded
Loïs Openda
out · loan · attacker · age 26
90
—
—
—
1.258
1.258
0.281
—
NOT GRADED
Loïs Openda / Lyon
Dušan Vlahović
out · free · attacker · age 26
—
—
—
—
—
—
0.554
—
NOT GRADED
no Understat coverage — Süper Lig is outside Understat's six leagues
Dušan Vlahović is not on the graded rows: Beşiktaş play in the Süper Lig, which Understat does not cover, so
there is no free npxG for him anywhere — the register's OUT side has no measurable rate and that is a known,
unhedged gap rather than an omission. Loïs Openda is at Lyon; the sealed grading unit is Serie A minutes
only, so his Ligue 1 numbers are context and are not graded here.
The zero-skill row
34.3% of comparable signings clear a frozen bar. Two thirds fail.
Measured in P-1 on 1,365 fee-bearing big-5 signings, 2017/18–2022/23
(experiments/p-1-results-02-battery.json, commit
20b94a8), and attached to the register as
addendum-02 before any grading, because the sealed file requires it.
Full P-1 readout →
By position family
Centre-forward
27.9% n=244
Winger
27.0% n=167
Attacking midfield
24.3% n=74
Centre-back
37.4% n=281
By fee band · by destination
Fee ≥ €20.0m (Q4)
45.5% n=352
Fee < €4.0m (Q1)
24.9% n=342
Top-tier destination
37.5% n=467
Low-tier destination
27.3% n=466
This is not the probability Kolo Muani clears the bar above. P-1's bar is goals+assists per 90
against a moving league median, over two seasons, from Transfermarkt counting stats. The register's bar is
npxG/90 ≥ 0.30 from Understat, over one season. Different metric, source and window. What
34.3% gives you is the denominator for the ordinary sentence “the signing worked out”.
Kolo Muani's €41.2m places him in fee band Q4 (45.5%) at a top-tier destination
(37.5%) in the CF family (27.9%). The joint cell
(CF × top × Q4) is 41.7% on
n=48 — thin, and not pre-registered;
it is shown so a reader does not multiply the marginals themselves, which would be worse.
The market is not uninformative: Q4 clears at 45.5% against Q1's
24.9%. That gradient is the thing any model has to beat, and in P-1 it mostly was not
beaten out of sample.
What the minutes condition is actually worth
Reaching 1,500 league minutes in year one is close to a coin flip
Same P-1 universe, attacker signings only, year-one
domestic-league minutes at the acquiring club. Descriptive: no test, no threshold.
All attackers reaching 1,500 min
43.6% n=472
Reaching 900 min
66.1%
Median year-one minutes
1,312
CF into a top-tier club, ≥ 1,500 min
42.7% n=75
… and below 900 min
29.3%
… median year-one minutes
1,307
Roughly three in ten comparable centre-forwards do not reach 900 minutes in year one. The register's most
likely single outcome is therefore a minutes verdict, not a rate verdict — and the sealed rule already
says that counts as a graded failure, not an excuse.
The bars, verbatim, and the file chain
Success bar (attackers). realised npxG/90 ≥ 0.30 AND ≥ 1,500 Serie A minutes in 2026/27 [chosen, labelled].
Minutes rule. Serie A minutes only. <900 min = MINUTES failure, graded, not excluded (F4).
Point test. |realised − projected| vs Marcel validated RMSE 0.077 (≈ ±0.10 at 80%).
Zero-skill row. Base rate from P-1 on the 2017–2022 universe, attached before grading.
When. End of Serie A 2026/27 (May 2027); then 2027/28.
The sealed register has never been edited. Its SHA-256 is checked against the hash recorded in
addendum-01 every time this page is rebuilt; a mismatch aborts the build rather than publishing.
Realised numbers: Understat getLeagueData/Serie_A/2026, fetched
2026-08-30T21:34:01Z —
361 player rows, 18 of 380 fixtures played.
Understat is a single free source with no redundancy; addendum-01 records that exposure.
This page is generated by scripts/site/build_register_page.py from the register
and the live Understat pull. It is static: it is as current as its last rebuild, dated above.