Tools · not tickets

Research

Hit rate sheet, public money traps, RLM, injuries, slate notes. Contest picks stay on the Board.

Freshness · contest MON SEP 14 — GUARDIANS ML · BRONCOS +2.5 · DODGERS ML · ROMA ML

Contest MON SEP 14 — GUARDIANS ML · BRONCOS +2.5 · DODGERS ML · ROMA ML (2026-09-14) · Injuries STALE · last Tue · Sep 1, 2026 · 8:27 AM ET · Wire Live pull · Mon Aug 17, 2026 · ~10:30 AM ET

Injuries · staleX wire · historical

Research source menu (desk + Sims)

All ten every day is too much and it conflicts. Four projection systems agreeing is one prior, not four edges. One primary stat site per sport per ticket. Weekly bulk dumps train Sims. Daily is thin.

daily

Odds API / book tape

Open, lock, close, steam, CLV. This is the PRICE half.

daily

ESPN scoreboard + MLB Stats API

Starters, roles, injuries, scores. Free. Already on the ticker.

daily

SportMonks Football (Growth + Premium Odds Lite + Historical)

Soccer DNB, Asian 0, 1X2, totals, H2H, standings, model predictions. Overseas depth Odds API US books do not carry. Live as of Day 15.

daily

SoccerPunt (Bryan daily)

Human soccer card. O/U 2.5, 1H, BTTS/GG, HT/FT, correct score, double chance. Daily tables. PRIOR only — never a ticket by itself.

daily

WinDrawWin (Bryan daily)

Correct-score first, then 1X2 / O/U / BTTS fall out of it. Small/Med/Large confidence. Form strings. Since 2003. PRIOR only.

daily

PredictZ (Bryan daily)

Side-by-side vs WinDrawWin. Form + predicted score. Where two humans DISAGREE is the interesting row. PRIOR only.

MLB

FanGraphs

ZiPS / Steamer / Depth Charts + plate discipline. Best free K-rate prior. Export the leaderboard — don’t browse 40 tabs.

Tennis

Tennis Abstract

Surface Elo, H2H, charting. Use Elo to price, not ATP/WTA rank.

WNBA

Basketball-Reference WNBA

Per-minute, pace, advanced box. Feeds minutes-first props.

Other

Sports-Reference family

Broadest schema when we have no specialist. One sport page, not the whole site.

file

Retrosheet

Every play since 1901. Outing length, hook, park, NRFI backtest. Zip, not a daily visit.

file

Baseball Prospectus

DRA / DRC+ — sharpest public pitcher/hitter metrics. Core boards open, rest paywalled. Don’t stack on top of FanGraphs as a second vote.

file

pybaseball (library)

Fetch layer for Savant / FanGraphs / BRef. Automation, not a site to read.

file

Jeff Sackmann tennis_atp / tennis_wta

Every tour match since 1968 — dates, minutes, surface. This is the fatigue file. Clone weekly, don’t scrape TA daily.

file

Cleaning the Glass

Garbage-time filtered on/off. Solves blowout minutes. Paid / limited. Not a daily public dump.

file

nflverse

PBP back to 1999. The serious free NFL file once preseason/regular starts.

Pick ONE prior family per ticket (e.g. FanGraphs Depth Charts — not ZiPS + Steamer + DRA averaged and called extra edge). Elo beats ranking for tennis price. Cleaning the Glass is a filtered box, not a second vote vs Basketball-Reference. Number minus market is the edge. Better data at the same price is just more confident wrong.

Desk book reports

Bench · Day 15

Sat Aug 15, 2026 · 2:54 AM ET · SportMonks history · pond was moneylines, ocean is base rates

Market: First-half Under 0.5

Model 1H draw 42–46% minus ~9% for 1-1/2-2 → 1H 0-0 ≈ 33–37%. Pin U0.5: Stoke +186 (35%) · Charlton +178 (36%) · Bristol +204 (33%) · SHU +197 (34%). Fair vs their own numbers. No bet today. Day 16 is per-team 1H 0-0 vs Pin, fire only at the 6-pt plus-money bar. Leans stay even to ±130.

Keep

Per-team first-half score distributions — which teams open 0-0, not the league −270
Corner style-persistence vs today’s 10.5 — habits, not luck
Soccer H2H match-facts before any total ships (tennis rule, ported)
Promoted-side last-division rates (Wrexham, Lincoln, Charlton) until October

Ignore

All 3,000 correct-score rows
Scorer props
Model 1X2 as a selector — tape only until it earns a vote
BTTS until it is not just totals in another coat
Cards until referee baselines exist

Chad · Day 15

Sat Aug 15, 2026 · 2:56 AM ET · History only when it is venue-matched to this manager and this line

Market: Corners Under 10.5

Rejected 1H U0.5 (draw ≠ 0–0) and 1H U1.5 (too expensive). Two tests: Charlton–Derby U10.5 Pin −152 needs conservative p ≥ 64% and no-vig ≤ 61%. SHU–Birmingham U10.5 Pin −123, same test. If neither clears, Pass stands.

Keep

Venue-specific, line-matched distributions tied to current manager/squad
First-half score states, corners, BTTS, Asian lines — not just moneylines

Ignore

Correct-score volume
Raw W/L streaks and generic H2H
Duplicate book rows and one-sided prices
1H draw used as a substitute for 0–0 history

Bench · Day 15

Sat Aug 15, 2026 · 2:57 AM ET · Need both sides of 1H 0.5 to run the 8-field

Market: First-half Under 0.5

Model 42–46% 1H draw vs Pin 33–36% on U0.5. Desk treats them as nearly the same animal in the Championship. Blocker: only the Under side was pasted, so no-vig cannot run. Full report with Day 16. Day 15 Official is Pass; three leans already in −130/+130.

Keep

SportMonks 1H-draw model vs Pin 1H U0.5 implied — only historical gap that could flip a ticket
p_sport for corners / 1H / BTTS that OpticOdds cannot invent

Ignore