Fair play · The line's record

The calibration ledger

Every Daily Drive line is frozen before entries open — the Predictor's AI line, or The Mark standing as the line when the Predictor has no honest number. This ledger is that line's own record against the final hammers: settled data only, the n beside every rate, and no player's calls anywhere in it. How marks work →

Scoring never changes mid-season

Scoring never changes mid-season — this ledger informs the next season, not this one.

Calibration is still building

Calibration is still building — n is small, treat rates as early reads. Every rate below carries its n; a rate without one would be marketing, not a record.

Lines issued — 8 settled

Split by the frozen line's source: Predictor (AI) 8 · The Mark 0 (n=8 settled lines). Higher/Lower correctness is the line's own directional record — the share of decisive lines (an exact-on-the-line hammer is a push, excluded) that a call on that side would have hit, judged by the game's one settlement rule.

Predictor (AI)

The Predictor's lines

Lines the AI published from the comp archive before the slate froze.

Lines issued (settled): 8
Match hit rate — hammer inside the ±5% band: 0% (n=8)
Higher-side correctness: 50% (n=8)
Lower-side correctness: 50% (n=8)
Predictor Average (mean absolute % error): ±42.2% (n=8)

The Predictor Average is the game's tracked skill line: the mean of |line − hammer| ÷ hammer × 100 over these settled Predictor lines. Anyone can recompute it from the frozen lines and the hammers above board — that is the point of publishing it here.

The Mark

The Mark as the line

Lines where The Mark itself stood as the line — the disclosed fallback, audited below.

Lines issued (settled): 0
Match hit rate — hammer inside the ±5% band: — (n=0)
Higher-side correctness: — (n=0)
Lower-side correctness: — (n=0)
Mean absolute % error: — (n=0)

Match audit — when does The Mark land?

For Mark-sourced lines only: Match hit rate broken down by the comp set behind the Mark. The question this answers with real numbers: do matches land more often when the comp set is deep, and when it is tight?

No Mark-sourced lines have settled yet

Every settled line so far came from the Predictor (n=8). The audit opens with the first settled line The Mark itself stood behind — with real comp-set numbers or not at all.

The Room, after settlement

For each fully settled game date: how the whole Room's calls split — Higher, Match the Mark, Lower, Pass — and what the hammers did. A date appears here only when every car on its slate has a final hammer. There is no live version of this table, and there never will be: the Room's percentages are a settled record, not a live signal.

The Room — 2026-10-24

Fully settled · 3 cars · n=3 decisions

HIGHER

0 of 3 decisions

MATCH THE MARK

0 of 3 decisions

LOWER

67%

2 of 3 decisions

PASS

33%

1 of 3 decisions

The Room leaned LOWER — hammers landed LOWER on 2 of 3 cars (67%, n=3 settled cars).

Hammers landed: Higher on 1 · Match on 0 · Lower on 2 (n=3 settled cars).

The Room — 2026-10-23

Fully settled · 5 cars · n=0 decisions

HIGHER

—

0 of 0 decisions

MATCH THE MARK

—

0 of 0 decisions

LOWER

—

0 of 0 decisions

PASS

—

0 of 0 decisions

The Room made no calls this day — the line's record stands on its own.

Hammers landed: Higher on 3 · Match on 0 · Lower on 2 (n=5 settled cars).

The calibration ledger

Scoring never changes mid-season

Scoring never changes mid-season — this ledger informs the next season, not this one.

Calibration is still building

Calibration is still building — n is small, treat rates as early reads. Every rate below carries its n; a rate without one would be marketing, not a record.

Lines issued — 8 settled

Predictor (AI)

The Predictor's lines

Lines the AI published from the comp archive before the slate froze.

Lines issued (settled): 8
Match hit rate — hammer inside the ±5% band: 0% (n=8)
Higher-side correctness: 50% (n=8)
Lower-side correctness: 50% (n=8)
Predictor Average (mean absolute % error): ±42.2% (n=8)

The Mark

The Mark as the line

Lines where The Mark itself stood as the line — the disclosed fallback, audited below.

Lines issued (settled): 0
Match hit rate — hammer inside the ±5% band: — (n=0)
Higher-side correctness: — (n=0)
Lower-side correctness: — (n=0)
Mean absolute % error: — (n=0)

Match audit — when does The Mark land?

No Mark-sourced lines have settled yet

Every settled line so far came from the Predictor (n=8). The audit opens with the first settled line The Mark itself stood behind — with real comp-set numbers or not at all.

The Room, after settlement

The Room — 2026-10-24

Fully settled · 3 cars · n=3 decisions

HIGHER

0 of 3 decisions

MATCH THE MARK

0 of 3 decisions

LOWER

67%

2 of 3 decisions

PASS

33%

1 of 3 decisions

The Room leaned LOWER — hammers landed LOWER on 2 of 3 cars (67%, n=3 settled cars).

Hammers landed: Higher on 1 · Match on 0 · Lower on 2 (n=3 settled cars).

The Room — 2026-10-23

Fully settled · 5 cars · n=0 decisions

HIGHER

—

0 of 0 decisions

MATCH THE MARK

—

0 of 0 decisions

LOWER

—

0 of 0 decisions

PASS

—

0 of 0 decisions

The Room made no calls this day — the line's record stands on its own.

Hammers landed: Higher on 3 · Match on 0 · Lower on 2 (n=5 settled cars).