Angles
Every game has more than one truth.
Put the voices around one table, browse their records, compare their disagreements, or return to the angles you follow.
Arena standings · current standings · ranked by CLV, not wins
Every proven lens votes on every game; current registry records drive the ranking. Your own Roundtable record renders below, independent of the house table above; a crew row needs a crew-scoped graded ledger and isn't built yet. ATS break-even is 52.4% at -110.
No proven lens has a graded arena record yet.
The lens_registry query returned no PROVEN rows to rank by CLV.
Bounty board · angles the system found, nobody has claimed
turn one into an angle → GRADING → PROVEN = badge + attribution on Tonight
- Overs, total 42 or lower
below -110 break-even
1437-1424 ATS (50.2%)
NFL game total · full history · n=2,861 · as of Oct 18, 2026, 6:21 PM ET
- Home favorites by 7 or more
below -110 break-even
812-838 ATS (49.2%)
NFL ATS spread · full history · n=1,650 · as of Oct 18, 2026, 6:21 PM ET
- Home favorites at key number 3
below -110 break-even
268-328 ATS (45.0%)
NFL ATS spread · full history · n=596 · as of Oct 18, 2026, 6:21 PM ET
Held for evidence · not yet surfaced
- Unders, total 48 or higher 705-666 ATS (51.4%) · full history · n 1,371 · not significant after FDR correction (q=0.477) · as of Oct 18, 2026, 6:21 PM ET
- Home underdogs, weeks 1–9 621-614 ATS (50.3%) · full history · n 1,235 · not significant after FDR correction (q=0.168) · as of Oct 18, 2026, 6:21 PM ET
- Home underdogs, weeks 10–18 603-604 ATS (50.0%) · full history · n 1,207 · not significant after FDR correction (q=0.138) · as of Oct 18, 2026, 6:21 PM ET
no claims yet — claiming opens with angle authoring
-110 break-even is 52.4%; angles with a hit rate at or below it stay held.
Pundit leaderboard · real people as lenses
actual calls transcribed and graded (mentions → show · episode · timestamp · quote ≤ 15 words, linked) · style model predicts how they'd see any game · anyone can be a pundit, including the joke ones
Ranked by ATS hit rate. CLV — priced at call time versus the close — breaks a tie only when both entries in it have a measured value; it is never required to enter the ranked board.
CLV is unavailable for tie-breaking in this refresh — unmeasured CLV shows as — and never counts as zero. Below-floor sources stay in the not-ranked list below.
CLV population is unavailable in this refresh; no missing value is treated as zero.
No graded pundit calls yet
The pundit_leaderboard query returned zero explicit_hit_rate rows.
Linked quotes are source-level episode mentions, not row-level graded-call citations; the current source and grade relations expose no shared citation identifier.
How a pundit becomes a lens
- Tape LIVE
- podcast / show atomised into 264,454 mentions
- Actual-call lens LIVE
- 278 graded claims
- Style model NOT RUNNING
- no style model has produced a take yet — this stage is not running
- Leaderboard LIVE
- 27 ranked pundits, with n + CI
- Remix LIVE
- use them as a Feed, a Filter, or a duel opponent
Simulated takes are always labelled "simulated from N calls"; real quotes are ≤ 15 words, attributed and linked to the episode. Joke lenses are graded too — a serious lens that can't beat Coin Flip gets demoted, in public.
Remix tree · Rest-days edge
where it came from, what it became, what it's earning
Headline records are ATS W-L with percentage, window, n, and as-of; -110 break-even is 52.4%.
Remix tree could not load
remix_attribution returned a cyclic path for child 2026e68e-ddad-4d47-bc73-c5d405bff052.
credit: a descendant credits every ancestor
Remixes keep the parent's record beside theirs, on the same plays — a "better" remix has to beat the original on its own rows, never on a friendlier window.