Measurement workspace
Current public page built from staged chart artifacts and a static site manifest.
CourtVision evidence
Measurement, documentation, and the work behind them.
Browse the public chart artifacts first, then follow their source modules and documentation in the public repository. The market-efficiency finding and documented limitations remain part of the record.
Published snapshot
Pipeline ledger
Counts are read from the committed site manifest, not a live service.
- InputsBasketball, Baseball, Soccer, Tennis
- Processing77 documented modules
- Published artifacts56 chart-backed records
- Sports covered
- 4
- As-of range
- 2026-04-12 to 2026-09-17
Architecture narrative
A documented path from footage to research artifacts
- Documented inputNBA broadcast video and public sports data enter the repository's computer-vision and research pipeline.
- Documented processingTracking, feature work, validation, and calibration produce dated research artifacts under recorded constraints.
- Published evidenceThis gallery exposes the chart artifacts that were staged for the public site, with links back to their modules.
Capability map
What is published, documented, and proposed
CV and research pipeline
Repository architecture and limitations are documented in the linked public records.
Hosted live backend
Not represented here as an operating public service or live-status claim.
Validated new analytics
Future analytics require their own published evidence before they appear in this gallery.
Evidence gallery
Published charts, linked to their source
Static artifacts from the public manifest. This page does not report a live backend state.
56 published charts shown
PublishedAs of 2026-07-22
Agent Fleet History
Published chart; see the linked source module for its recorded method.
Partial documentationAs of 2026-09-17
Blowout Dynamics
Per game, order score ticks by (game-clock, ts); a margin threshold's point of no return is the first tick from which |home_score-away_score| stays >= threshold to the end. Reads only the score path, not model/market prob.
PublishedAs of 2026-09-16
Brier Skill Scores
mlb: BSS(model vs market) in [-0.113, -0.092] -- model does not beat the closing reference forecast; BSS(model vs climatology) in [+0.081, +0.621] (BSS>0 means it beats that baseline). soccer_intl: BSS(model vs market) in [-1.171, -0.469...
Partial documentationAs of 2026-09-16
Calibration By Market Type
Brier scored only for market types whose corpora carry resolved outcomes: mlb_moneyline: model Brier 0.1668 vs market 0.1510 (n=27351); soccer_match: model Brier 0.3325 vs market 0.1952 (n=4265). Unscorable market types (no resolved outc...
PublishedAs of 2026-09-16
Calibration Over Time
Published chart; see the linked source module for its recorded method.
Partial documentationAs of 2026-09-16
Calibration Stability
mlb/model: 3/10 eligible bins have a calibration-gap 95% CI that excludes 0; 7 within noise. mlb/market: 0/10 eligible bins have a calibration-gap 95% CI that excludes 0; 10 within noise. soccer_intl/model: 8/9 eligible bins have a calib...
PublishedAs of 2026-07-23
Cf Pace Variance
Pace is a real but MODEST variance lever: across the observed NBA pace range (~92-104) the same matchup's upset prob shifts by only a few points -- fewer possessions favor the underdog (sqrt(N) scaling), but the effect is small because r...
PublishedAs of 2024-25 (Elo end-of-season) x 2024_25 on/off slice
Cf Star Removal
Descriptive only
PublishedAs of 2026-09-17
Comeback Atlas
Published chart; see the linked source module for its recorded method.
Partial documentationAs of 2026-07-22
Cross Sport Scoreboard
Model beats market only where verdict starts MODEL_SHARPER (still PROVISIONAL pending more data). Most rows are UNDERPOWERED (CI spans zero) or MARKET_SHARPER -- reported as-is.
PublishedAs of 2025-26 regular season (through 2026-04-12)
Ctx Lineup Proxy
team win rate with player ACTIVE vs MISSED, inside the player's own tenure window; Wald 95% CI on the win-rate difference. RAPM-free proxy.
PublishedAs of 2026-05-21
Ctx Team States
halftime margin & second-half margin per team-game from running pbp scores; split each team's games by led/trailed at half
PublishedAs-of not stamped
Cv Tracking Stats
volume/coverage stats only -- no accuracy/quality claims (MOT metrics unbenchmarked, see JOB_EVIDENCE_PACKET s4)
PublishedAs-of not stamped
Dossier Completeness
Published chart; see the linked source module for its recorded method.
PublishedAs of 2026-07-11
Fwd Claim Scoreboard
Per claim family (sport x hypothesis) from the 4 validation ledgers: verdict history ordered by run_ts, current status = last entry. Families with >1 distinct status flagged flipped.
PublishedAs of 2026-07-22
Honesty Exhibit
nulls (351) outnumber confirms (168) 2.1x -- we publish our nulls
PublishedAs of 2026-09-16
Info Arrival Curve
Published chart; see the linked source module for its recorded method.
PublishedAs of 2026-09-17
Kernel Transfer
This table's only cross-sport-comparable number is the Murphy reliability component, for 2 of 4 sports where a market-side probability decomposition exists (mlb moneyline, soccer_intl moneyline). CRPS (mlb totals/margin), Brier-without-d...
PublishedAs of 2026-09-17
Market Convergence
Published chart; see the linked source module for its recorded method.
PublishedAs of 2026-09-16
Market Disagreement Profile
mlb: at largest disagreement (>=.10, n=8530), market usually right (model_closer_rate=0.199, model_brier=0.1778 vs market_brier=0.1298). soccer_intl: at largest disagreement (>=.10, n=2291), market usually right (model_closer_rate=0.105,...
PublishedAs of 2026-09-16
Market Overreaction
Published chart; see the linked source module for its recorded method.
PublishedAs-of not stamped
Mechanism Ledger Export
Every named mechanism tested, with its recorded verdict and evidence pointer. NULL, REJECT, and NOT_TESTABLE are market-efficiency evidence. This is a descriptive measurement artifact.
PublishedAs-of not stamped
Mechanism Survival
Survival rate = share of TESTABLE hypotheses whose latest recorded verdict is CONFIRMED_LOCAL / CONFIRMED_LOCAL_incl_2026_OOS / REPLICATED. A NULL or REJECT is honest market-efficiency evidence.
PublishedAs of 2026-07-25
Micro Absorption
pregame per-series consecutive-snapshot |devigged_prob| movement, bucketed by minutes-to-commence, from own scraped line_history
PublishedAs-of not stamped
Micro Closing Decay
wnba: Brier sharpens toward the close -- T-6h=0.2454 -> close=0.2280 (delta=+0.0174, n_games=30). soccer_intl: Brier does NOT sharpen toward the close -- T-24h=0.1355 -> close=0.1374 (delta=-0.0019, n_games=13) (UNDERPOWERED).
PublishedAs-of not stamped
Mlb Velo Bands
Published chart; see the linked source module for its recorded method.
PublishedAs of 2026-07-25
Novel Line Half Life
soccer_intl completes half its pre-game line motion by 2.7h before tip; nba not until 4.8h. tennis, wnba move most of their line >6h before tip.
PublishedAs of 2026-09-17
Novel Live Clock Fraction
mlb stays contested later (LCF 0.737) than soccer_intl (LCF 0.599). (nba comeback-rate cross-check 0.0146)
PublishedAs-of not stamped
Novel Load Bearing Index
Most one-star-fragile: DEN (Nikola Jokić, delta_winprob 0.5822). Two estimators name the same #1 player for 1/30 teams (directional cross-check, different seasons).
PublishedAs of 2026-09-17
Novel Market Foresight Premium
mlb: MFP rises from 0.05 to 0.41 over the game (mean 0.15); soccer_intl: MFP rises from 0.68 to 1.49 over the game (mean 0.92)
PublishedAs-of not stamped
Novel Overreaction Harvest Gap
soccer_intl: overshoot 0.286 and model is closer only 22% of the time at maximum disagreement, yielding OHG 0.081 as a measured failure to harvest; mlb: overshoot 0.072 and model is closer 38% of the time at maximum disagreement.
PublishedAs-of not stamped
Novel Schedule Fatigue Tax
Most-taxed schedule: DEN 2025-26 (-0.38 pts/100 ORtg season-averaged); least: MEM 2025-26 (-0.27). League schedule-inequality range 0.11 pts/100 across 90 team-seasons.
PublishedAs-of not stamped
On Off Showcase
Published chart; see the linked source module for its recorded method.
PublishedAs-of not stamped
Player Metric Landscape
Support = raw input-column coverage only, never a claim that a branded metric is reproduced.
PublishedAs of 2026-07-22
Qa Coverage Stats
answer engine: 87/87 regression-bank checks pass (fail-closed statuses graded as PASS); honest coverage 36.6% of answerable questions -- refusals (no_data/not_supported/ambiguous/refused) are the fail-closed design, not a defect
PublishedAs-of not stamped
Reject Graveyard
A REJECT or DEFER is honest market-efficiency evidence.
Partial documentationAs of 2026-09-16
Residual Anatomy
Published chart; see the linked source module for its recorded method.
Partial documentationAs of 2026-09-16
Residual Autocorrelation
mlb: median within-game lag-1 residual autocorr = model 0.971 (n=173 games), market 0.965 (n=173); share>=0.9 model 0.90 vs market 0.86. soccer_intl: median within-game lag-1 residual autocorr = model 0.966 (n=26 games), market 0.952 (n=...
PublishedAs of 2026-09-16
Soccer Calibration Pack
soccer_intl corpus: n=4265 rows -- far smaller than mlb/nba, wider CIs, single-fold reads not durable. Murphy: model Brier 0.3325 vs market 0.1952 (gap=+0.1374); market leads. Minute-bucket ECE (weighted): 0.3748.
PublishedAs of 2026-07-23
Soccer Form Stability
n=153 shared national teams. Spearman rho=0.6419, Pearson r=0.6369 between trailing-10 form win-rate and net-xG strength -- a parallel-forms rank-stability read, NOT a temporal split-half (that series is absent from the store) and NOT a ...
PublishedAs-of not stamped
Statcast Showcase
Published chart; see the linked source module for its recorded method.
PublishedAs-of not stamped
Tennis Showcase
tennis has a served live model but no ingame_grade_joined corpus locally -- this showcase surfaces the two REAL preregistered gate receipts that ARE present (pregame cross-corpus + ingame surface-context detail layer), not a recomputed i...
PublishedAs of 2026-07-24
Tick Microstructure
market-microstructure MEASUREMENT (latency/cadence) -- NOT a trading signal
Partial documentationAs of 2026-09-17
Why Attribution
Largest calibrated win-prob drop: mlb mid(inn4-6):.8-1 -> late(inn7+):0-.2 = -0.8985 (min support n=1462). Any in-game move decomposes into the adjacent-time bucket transition it crossed.
PublishedAs of 2026-07-18
Xsport Structure
Same methodology, per-sport n honest. nba: market ECE=0.006 (n_games=1593), fav_gap=+0.005/dog_gap=-0.001 (favorite-longshot-consistent), comeback~0.015. Not buildable: mlb, soccer, tennis.