MLB Betting Model · 57.9% Win Rate · Free Signals

F1Miami GP · May 3 EPLGW34 · This Weekend UCLQF · Live MLBDaily Slate · Live HORSESKentucky Derby · May 2 FA CUPSemi-Finals NASCARTalladega · Apr 27 F1Miami GP · May 3 EPLGW34 · This Weekend UCLQF · Live MLBDaily Slate · Live HORSESKentucky Derby · May 2 FA CUPSemi-Finals NASCARTalladega · Apr 27

2026 Season · Live Analytics

WHERE DATA
ANALYTICS MEETS SPORT.

Quantitative sports analytics for Formula 1, Premier League, La Liga, Champions League, and MLB. We build the models, run the numbers, and publish where the market is wrong — and by exactly how much.

Latest from The Chalk Line

Loading...

NEWSLETTER

Analytics model value.
Delivered anywhere.

Edge %, fair probability, and why the market is wrong — before every race, gameweek, and daily slate.

YOUR EMAIL

No spam. No picks factories. Just the number and the reasoning.

ANALYTICS

Full model access.
One password.

BET/FADE signals, win probabilities, and edge % — for every sport, every event. We show our work.

F1 Race model · Win probabilities

EPL xG model · GW signals

MLB Daily slate · Moneyline + props

UCL Knockout · Edge signals

HORSES Derby · Value picks

THE PHILOSOPHY

One bet per event.
NO BET is a real signal.
We show our work.
The market is wrong
— occasionally.

Round 2 · 2026 Season · Pre-Race

JAPAN RACE

SUZUKA CIRCUIT · 53 LAPS · 5.807 KM  ·  Loading live model data...

2026 WINS

            RUS: 1ANT: 1
          

R3

JAPAN

Mar 29

R4 · SPRINT

MIAMI

May 3

R5 · SPRINT

CANADA

May 24

R6

BARCELONA

Jun 7

RACE WIN

—

Loading...

TOP PICK

—

Loading...

FAST LAP

RUSSELL

+3.2% edge

FADE

VERSTAPPEN

−3.9% vs field

2/2

Correct Calls · 2026

67%

SC Probability

2026

Reg Reset Year

Win Probability

Pit Strategy

Circuit Analysis

Value Bets

Value Map

Race Win Probability · Model vs. Market

↓ Export CSV

Driver

Model %

Book %

Edge

Signal

K. ANTONELLI

Mercedes

28.4%

20.0%

+8.4%

BET

G. RUSSELL

Mercedes

22.1%

18.5%

+3.6%

BET

C. LECLERC

Ferrari

16.8%

20.0%

−3.2%

FADE

L. HAMILTON

Ferrari

14.2%

12.5%

+1.7%

PASS

O. PIASTRI

McLaren

8.4%

10.0%

−1.6%

FADE

L. NORRIS

McLaren

6.2%

8.3%

−2.1%

FADE

M. VERSTAPPEN

Red Bull

2.8%

6.7%

−3.9%

FADE

F. ALONSO

Aston Martin

1.1%

0.8%

+0.3%

PASS

Race Analytics

↗ Full View

Win Probability

Loading...

Safety Car Probability

—%

likely

Loading circuit data...

Optimal Pit Windows

LAP 1 ──────────────── 56

ANT

L18–25

LEC

L13–20

PIA

L16–23

VER

L20–29

Tire Strategy

1-STOP · 54% prob

─pit─

2-STOP · 31% prob

──

──

SC-triggered 15% — soft restart favours Ferrari launch

Japan · Race Start

Sun Mar 29, 2026 · 6:00 AM ET

HRS

MIN

SEC

RND

Model Accuracy · 2026

50%

TOP-3

—

Win

—

Podium

Races

Top Value Markets

All →

ANTONELLI WIN

FanDuel · Race Winner

+420

+EV · +8.4%

LECLERC PODIUM

DraftKings · Top 3

-115

+EV · +12.1%

SAFETY CAR YES

BetMGM · Props

-130

+EV · model −195

VER FADE vs PIA

Bet365 · H2H

+105

+EV · +6.3%

NORRIS TOP-6

FanDuel

-200

−EV avoid

Data Stack

INGEST

OpenF1 · FastF1 · Ergast · Betfair · Kinesis

PROCESS

Databricks Delta Lake · dbt · Silver tables

MODEL

LightGBM · MLflow · Feature Store · Gold

SERVE

AWS Lambda · API Gateway · This site

Cup Series · Loading...

NASCAR
Loading...

Loading race data...

MODEL

Win Probability

Value Bets

Track Notes

Value Map

—

Total Laps

—

HP Package

—

BET Signals

—

FADE Signals

Race Win Probability · Model vs Market

CAR

DRIVER

MODEL%

BOOK ODDS

EDGE

SIGNAL

Loading predictions...

This Week's Signals

Track Intel

Race Info

      
      MLB 2026 · Loading...
    

MAJOR LEAGUE
BASEBALL

Loading game data...

GAMES TODAY

—

VALUE SIGNALS

—

FADES

—

SEASON ROI

—

Today's Games

Moneyline

Totals

SIGNALS: STRONG VALUE6%+ edge VALUE4-6% LEAN2-4% · not a bet FADEoverpriced 6%+ PASSno edge NO LINEawaiting odds CONTEXTdisplay only

TODAY'S GAMES · MONEYLINE · TOTAL · NRFI · STARTER PREVIEW

Loading games...

Model Info

📊 LightGBM · ML signals only
📅 5,003 games · 2024–2026
🎯 36 features · AUC 0.605
🔧 Clamped [25%–75%] + 40/60 blend
💰 Vig-adjusted edge (6.5%)
🟢 STRONG VALUE: 6%+ edge
🟡 VALUE: 4–6% edge
🔴 FADE: −6% (overpriced)

Season Record

Schedule

      
      2025-26 SEASON · KNOCKOUT STAGE
    

CHAMPIONS
LEAGUE

UEFA CHAMPIONS LEAGUE · QUARTER-FINALS · DIXON-COLES MODEL · MARKET-EDGE SIGNALS

CHAMPIONS LEAGUE

—

QF / SF fixtures

BET SIGNALS

—

This round

UPDATED

—

Last model run

⭐ Champions League

All Matches

UEFA CHAMPIONS LEAGUE · MARKET EDGE MODEL · FAIR PROB VS BOOK IMPLIED (VIG ADJUSTED)

HOME
AWAY
MATCH
SIGNAL ODDS
SIGNAL

Loading Champions League signals...

Top Signals · This Round

UCL 2025-26

TOURNAMENT INFO

StageQuarter-Finals

QF Leg 1Apr 7-8, 2026

QF Leg 2Apr 14-15, 2026

FinalMay 30 · Budapest

UCL model includes aggregate context — second leg pressure, must-win situations, and defensive approach all factored in.

COUGAR'S CORNER 🐆

EXPERT OVERRIDE · THIS WEEK

PSG to advance

"Liverpool are tired. Mbappé is angry about something. PSG in 90."

MODEL

—

COUGAR

—

            One override per week. Gut feeling vs algorithm.

            Transparency is everything.

Model Notes

· Market-anchored blend (10% model / 90% market).
· BET signal requires ≥4% edge after vig removal.
· FADE signal requires ≥6% market overweight.
· Odds: The Odds API · US bookmakers.
· Refresh: Run Notebook 25 in Databricks.

2026 THOROUGHBRED SEASON · LIVE MODEL

KENTUCKY DERBY
2026

CHURCHILL DOWNS · 1¼ MILES · $5,000,000 PURSE · MAY 2, 2026

NEXT · DERBY PREP

Arkansas Derby

Apr 11 · Oaklawn Park

DERBY PREP

Blue Grass Stakes

Apr 11 · Keeneland

FEATURED RACE

Kentucky Derby

May 2 · Churchill Downs

TRIPLE CROWN

Preakness Stakes

May 16 · Pimlico

TRIPLE CROWN

Belmont Stakes

Jun 6 · Belmont Park

DERBY CONTENDERS · MODEL WIN PROBABILITIES

Loading predictions...

Kentucky Derby

Wood Memorial

Santa Anita Derby

Arkansas Derby

Blue Grass

#
HORSE · TRAINER / JOCKEY
WIN%
BEYER
M/L
EDGE
SIGNAL

Loading Derby predictions...

⚠ POST POSITIONS NOT YET DRAWN — Bias adjustment pending. Draw expected Wednesday April 29.

Derby Value Markets

Loading signals...

Derby Weather · Churchill Downs

Fast

Track Condition

72°F

Forecast Temp

            Forecast becomes reliable April 27 (5 days before Derby). Model adjusts probabilities automatically based on track condition.
          

MUDDERS (benefit if wet)

Coal Battle · Owen Almighty · Render Judgment

FAST TRACK ONLY (avoid if wet)

Caviar Dreams · Journalism · Chunk of Gold

Post Position Bias · Churchill Downs

Historical win rates by post — 1¼ miles dirt

1

2

3

4

5

6

7★

8

9

10

            Post 7 is the best in Derby history (+42% ROI). Posts 1-4 are disadvantaged by traffic on the first turn. Posts 15-20 face an extreme distance penalty at the longer distance.
          

Trainer Angles · Churchill Downs

Brad Cox — 23.3% CD

Horses: Render Judgment, Strong Opinion · Best trainer ROI at Churchill Downs +14%

Bob Baffert — 21.0% CD

Horse: Chunk of Gold · 7 Derby wins, highest win rate at Churchill Downs +8% ROI

Todd Pletcher — 17.9% CD

Horses: Citizen Bull, Locked, Majestic Eagle · 52 Derby starts, 2 wins. High volume trainer.

Steve Asmussen — 20.0% CD

Horse: Coal Battle · Most wins all-time as trainer. 18 Derby starts, 0 wins — watch for breakout.

Model Methodology

18 features including Beyer Speed Figures, pace ratings, trainer/jockey stats at Churchill Downs, post position bias, class level, and Derby points.
Edge calculation compares model win probability to morning line implied probability adjusted for 17.25% takeout.
Updates: Post draw (Apr 29) · Final scratches (May 1) · Race day morning line

FIFA WORLD CUP 2026 · ML MODEL LIVE

FIFA WORLD CUP
2026

USA · CANADA · MEXICO · 48 TEAMS · 104 MATCHES · JUNE 11 – JULY 19

DAYS TO KICKOFF

84d

TEAMS

SIMULATIONS

10K

MODEL BET SIGNALS

—

GROUP STAGE

72 Matches

Jun 11 – Jul 2

KNOCKOUT

R32 → QF → SF

Jul 5–16

FINAL

MetLife Stadium

July 19, 2026

US MATCHES

11 Venues

NY, LA, Dallas, Miami +

OUTRIGHT WINNER · MODEL WIN PROBABILITIES

Loading predictions...

#
TEAM · GROUP / STYLE
MODEL
MARKET
ODDS
EDGE
SIGNAL

Loading World Cup predictions...

Outright Value Markets

Loading signals...

Model Methodology

ELO Monte Carlo — runs 10,000 full tournament simulations. Each match uses ELO win probability. Results aggregated to win percentage.
6 key features: ELO rating · FIFA ranking · xG history · Head-to-head record · Squad age · Tournament experience
Updates after every match — ELO adjusts in real time as results come in during the tournament.
Edge calculation compares model probability to sportsbook implied probability (vig-adjusted).

Host Nations · Betting Angle

🇺🇸 USA — Home Advantage

Playing in front of home crowds at MetLife, Rose Bowl, AT&T Stadium. ELO boosted +50 pts for home advantage. Model gives 8–12% win probability vs market 7.7%.

🇲🇽 Mexico — Passionate Support

Will play group stage games in Mexico. Passionate fanbase boosts ELO. Watch for value if they draw a favorable group.

🇨🇦 Canada — Rising Team

Only second World Cup ever. Young squad with Alphonso Davies. Long odds at +100/1 but model gives slight value vs market.

Data Sources

BALLDONTLIE API — free World Cup 2026 data including live odds, fixtures, standings
API-Football — free tier, 100 calls/day, fixtures + livescores
OpenFootball (GitHub) — historical results back to 1930 World Cup
Betfair Exchange — live market odds (existing credentials)

2026 NCAA TOURNAMENT · IN PROGRESS

MARCH MADNESS
2026

LUCAS OIL STADIUM · INDIANAPOLIS · APRIL 6, 2026 · CHAMPIONSHIP

TEAMS LEFT

—

BET SIGNALS

18d

TO FINAL

Duke

MODEL FAVORITE

Championship Odds

First Round

By Region

Upset Picks

#
TEAM · SEED / REGION
WIN%
MKT%
ODDS
EDGE
SIGNAL

Loading tournament predictions...

Championship Value Bets

Loading signals...

Tournament Schedule

${[ ["First Round","Mar 19–20","🔴 LIVE"], ["Second Round","Mar 21–22","⏳"], ["Sweet 16","Mar 27–28","⏳"], ["Elite Eight","Mar 29–30","⏳"], ["Final Four","Apr 4","⏳"], ["Championship","Apr 6","⏳"], ].map(([r,d,s])=>`

${r}

${d}

${s}

`).join('')}

Key Injury Watch

DUKE — CALEB FOSTER

Guard injury — status uncertain. Model reduces Duke win prob by 3% if out

MICHIGAN — L.J. CASON

Guard injury looms large — Wolverines depth tested in tournament

GONZAGA — BRADEN HUFF

Out first two rounds — Zags offense significantly diminished without him

Model Methodology

          Features: KenPom efficiency, seed, SOS, recent form, injury adjustments, bracket path difficulty. Edge = model win prob vs sportsbook implied prob. BET threshold: edge > 3% + Kelly > 0.3%. Tournament in progress — model updates after each round.
        

2025-26 SEASON · GAMEWEEK 31

PREMIER
LEAGUE

ENGLAND · MATCHDAY PREDICTIONS · xG MODEL · VALUE BETS

LEADER

Arsenal

70 pts

TITLE GAP

9 pts

vs Man City

BET SIGNALS

—

This GW

GAMES LEFT

Season ends May 24

League Table

GW Predictions

xG Analysis

Value Bets

Top Scorers

Title Race

SIGNALS: BET4%+ edge vs book FADEoverpriced by book PASSno edge found

          Dixon-Coles Poisson model · Attack/defense ratings from rolling xG · Home advantage adjusted · Edge = model probability − vig-adjusted book implied probability · Market-anchored (10% model / 90% market early season) · BET signal requires ≥4% edge after vig removal · Odds from The Odds API (FanDuel, DraftKings, Bet365)
        

GAMEWEEK PREDICTIONS · DIXON-COLES POISSON MODEL · HOME ADVANTAGE ADJUSTED

Loading...

This GW · Value Markets

Loading signals...

xG Regression Watch

TEAMS OVERPERFORMING xG — FADE NEXT 3-5 MATCHES

Loading...

Title Odds

Loading...

Relegation Zone

Loading...

2025-26 SEASON · MATCHDAY 29

LA
LIGA

SPAIN · MATCHDAY PREDICTIONS · xG MODEL · VALUE BETS

LEADER

Barcelona

70 pts

TITLE GAP

4 pts

vs Real Madrid

BET SIGNALS

—

This MD

GAMES LEFT

Season ends May 24

League Table

MD Predictions

xG Analysis

Value Bets

Top Scorers

Title Race

SIGNALS: BET4%+ edge vs book FADEoverpriced by book PASSno edge found

          Dixon-Coles Poisson model · Attack/defense ratings from rolling xG · Home advantage adjusted · Accent-normalized team matching · Edge = model probability − vig-adjusted book implied probability · Market-anchored (10% model / 90% market) · BET signal requires ≥4% edge · League avg goals: La Liga 1.42 per team
        

MATCHDAY PREDICTIONS · DIXON-COLES POISSON MODEL · HOME ADVANTAGE ADJUSTED

Loading...

This MD · Value Markets

Loading signals...

xG Regression Watch

TEAMS OVERPERFORMING xG — FADE NEXT 3-5 MATCHES

Loading...

Title Odds

Loading...

Relegation Zone

Loading...

ADMIN ONLY · MODEL RESULTS TRACKER

RESULTS
TRACKER

LOG RESULTS AFTER EACH RACE / MATCH / RACE DAY · TRACKS MODEL ACCURACY OVER TIME

BET SIGNALS

—

WIN RATE

—

BREAK-EVEN

52.4%

NET UNITS

—

ROI

—

AVG ODDS

—

FADES CORRECT

—

      WIN RATE = BET signals only · NET UNITS = 1 unit per signal at book odds · ROI = net return on total wagered · FADES = faded team lost
    

All Sports

EPL

La Liga

MLB

Horses

World Cup

AUTOMATED RECONCILIATION

          Results auto-populated by Notebook 13 after each gameweek or daily slate.

          Run schedule: EPL → Monday · F1 → Tuesday · MLB → Daily · Horse → Race day

          Each run fetches actual scores, matches against saved predictions, scores every call.

DATE
SPORT
MATCH · MODEL CALL
SCORE
RESULT
O/U
xG ERR
GW

No results logged yet. Add your first result above.

Performance by Sport

Log results to see breakdown

Monthly Win Rate

Log results to see trend

Best Calls

Wins at longest odds appear here

Export & Info

            Data source: S3 · auto-updated by Notebook 13

            Path: api/results/history.json

            Updates: Monday (EPL) · Tuesday (F1) · Daily (MLB)

ANALYSIS · PICKS · OPINIONS

THE
CHALK LINE

DATA-DRIVEN TAKES · NO NARRATIVES · NO HYPE

All

Premier League

Formula 1

Horse Racing

Analysis

Loading...

Latest Picks

Loading...

Model Track Record

Loading...

Topics

Premier League Formula 1 Horse Racing xG Analysis Bankroll Strategy World Cup

SPORT / TIER	SIGNALS	WIN %	NET UNITS	ROI	STATUS
Loading...

⬡ INVESTOR & BUSINESS OVERVIEW · CONFIDENTIAL

ONE IS ENOUGH

A data-driven multi-sport betting analytics platform built on enterprise-grade infrastructure. Live across F1, NASCAR, EPL, La Liga, MLB, and Horse Racing. This document covers the business case, market opportunity, technical architecture, competitive differentiation, and 18-month roadmap.

Background

Market

Data Pipeline

Differentiation

Business Plan

Roadmap

✍️ Blog Drafts

📱 Socials

THE FOUNDING INSIGHT

The Analytical Gap Between Data and Market Is Widest In Niche Sports

The major US sportsbooks — FanDuel, DraftKings, BetMGM — have poured hundreds of millions into NFL, NBA, and college football analytics. Those markets are deeply efficient. The edge is thin and shrinking.

The same is not true for Formula 1, NASCAR, international soccer, MLB totals, and horse racing. These markets are priced by generalist traders using basic form data. A model that ingests telemetry, pitcher-level stats, xG, and track history finds edges that simply don't exist in the major American leagues.

That gap is the business.

WHY NOW

Four Converging Forces

1. F1's US Expansion Is Accelerating

Formula 1 signed Betway as its first Official Betting Operator in 2026. 800M global fans. 3 US races. US betting legal in 35+ states. The sport is actively building betting infrastructure for the first time in its 76-year history.

2. MLB Is The Largest US Daily Betting Market

162 games per team. 30 teams. 2,430 games per season. Daily signal volume that dwarfs any other sport. Our LightGBM v3 model: 5,003 training games, 36 features, 40/60 market blend, tiered signals (STRONG VALUE / VALUE / FADE). 57 tracked signals at 57.9% win rate, +10.7% ROI, 62.9% fade accuracy.

3. Free Enterprise Data Infrastructure Now Exists

OpenF1, FastF1, Jolpica, MLB Stats API, ESPN API — the data stack that previously required enterprise licensing is now accessible to independent developers. We run on Databricks + AWS at under $100/month.

4. Prediction Markets Opening A New Affiliate Channel

Kalshi — CFTC-regulated event contracts — launched sports markets in 2025 with an affiliate program. This adds a second monetization layer alongside FanDuel/DraftKings, targeting a more analytically sophisticated audience who self-selects into our content.

THE BRAND

One Is Enough

The name resonates with racing fans — championships decided by a single point. As a brand it communicates our philosophy: we don't give you 15 picks a week and hope some land. We find markets where the model has genuine edge, explain why, and publish the full track record openly.

Launched 2026 under Eagan Digital LLC. Live signals across 6 sports. Public results tracker updated daily. @OIEdge on X.

This is a credibility statement in a space dominated by noise.

THE INFRASTRUCTURE

Production-Grade From Day One

The platform runs on Databricks (used by 60% of the Fortune 500) and AWS. Six live sports pipelines. Bronze → Silver → Gold Delta Lake architecture. LightGBM models trained on 4,900+ historical MLB games. Results reconciled automatically via ESPN API nightly.

This infrastructure is the defensible moat. A content site can copy our picks. They cannot replicate a Databricks feature store with 3 seasons of pitcher data, a model with a public accuracy record, and a pipeline that scores results automatically every night at 11pm ET.

THE ACQUISITION BENCHMARK

Action Network — a sports betting analytics and affiliate site founded in 2017 — was acquired by Better Collective for $240 million in May 2021 at a 16x revenue multiple. It had grown from $15M revenue in 2020 to ~$40M projected in 2021. It covered NFL, NBA, MLB — not motorsport or horse racing. That gap is exactly where we operate.

$165B

US Legal Sports Betting Handle 2025

Over $600B total since PASPA repeal in 2018. Online sports betting market projected to reach $98.5B revenue by 2032 at 8.8% CAGR.

2,430

MLB Regular Season Games Per Year

Daily signal volume. Moneyline, totals, and run line markets active every day April–October. Largest daily US betting market by game count.

16x

Revenue Multiple · Action Network Exit

Sports betting affiliates typically trade at 5–7x revenue. Action Network commanded 16x due to diversified revenue. We target the same model across underserved sports.

35+

Legal US Betting States

10–20% of sportsbook users originate through affiliates. FanDuel/DraftKings pay $25–$100 CPA per new depositor. Kalshi (CFTC-regulated) adds a second affiliate channel.

MARKET 1 · MLB · DAILY VOLUME

Major League Baseball

2,430 Games · Daily Signals April–October

Largest daily US betting market by game count. Every game is a moneyline, totals, and run line market. Daily affiliate traffic opportunity no other sport matches.

Live Model v3 · 57.9% Win Rate · +10.7% ROI

LightGBM trained on 5,003 games. 36 features. Clamped probabilities [25%–75%] with 40/60 market blend. Tiered signals: STRONG VALUE / VALUE / FADE. Results reconciled nightly via ESPN API. 62.9% fade accuracy.

Kalshi + Sportsbook Dual Affiliate

FanDuel/DraftKings for traditional bettors. Kalshi event contracts for analytically sophisticated readers. Two channels, one daily content pipeline.

Affiliate: FanDuel $25–35 · DraftKings $75 · Kalshi est. $15–30

MARKET 2 · MOTORSPORT

Formula 1 + NASCAR

F1 signed Betway as Official Betting Partner 2026

First time in 76-year history. 800M global fans. 3 US races. Actively building betting infrastructure for the first time.

2026 Regulation Reset = Market Inefficiency

Complete rule overhaul. Bookmakers pricing on 2024–25 data. Already proven: Antonelli correctly predicted at +420 for Chinese and Japanese GPs.

No Direct US Competitor

Action Network covers F1 minimally. Racing Post is UK-only. No US-focused motorsport analytics platform with an ML model exists.

Affiliate: FanDuel $25–35 · DraftKings $75 · Bet365 30% RevShare

MARKET 3 · HORSE RACING

Thoroughbred Racing

$471B Global Market — Largest Betting Sport

More global betting handle than any other sport. Kentucky Derby alone generates enormous single-day handle. US pari-mutuel handle ~$12B annually.

Enormous Analytics Gap

Most horse bettors use DRF printouts and gut feel. No platform combines speed figures, trainer/jockey stats, post position bias, and pace scenarios in one ML model.

Kentucky Derby = Annual Marketing Event

3–5M casual bettors open sportsbook accounts for Derby week. Highest single-event affiliate conversion in US horse racing.

Affiliate: TVG $50–75 · TwinSpires $40–60 · FanDuel Racing $50–100

COMPETITIVE LANDSCAPE — WHO EXISTS AND WHAT THEY MISS

COMPETITOR	STATUS	MLB	F1 DEPTH	ML MODEL	US FOCUS	GAP WE FILL
Action Network	$240M exit 2021 · 16x rev	Editorial only	Minimal	No — editorial picks	Yes	ML model + motorsport depth
The Athletic	NYT-owned · subscriber	Deep journalism	Light	No model	Yes	Quantitative edge vs journalism
Racing Post	UK incumbent · no US	None	Medium	Human tipsters	No — UK only	US market + ML vs tipsters
SportsLine / CBS	Active · general sports	Yes · simulation	Light	Simulation model	Yes	Motorsport depth + transparency
ONE IS ENOUGH	Live · April 2026	LightGBM · 55% ML	Deep · LightGBM	Yes · Databricks	Primary focus	All of the above

Source: Action Network acquired by Better Collective May 2021 for $240M at ~16x revenue. MLB model v3 live April 2026 — 57.9% win rate, +10.7% ROI on 57 tracked signals.

SEGMENT 1

Daily MLB Bettor

Already bets MLB daily on FanDuel or DraftKings. Wants model-based signal with edge %. Primary affiliate conversion target April–October.

→ Free tier · Affiliate · Kalshi

SEGMENT 2

F1 / NASCAR Fan

Watches races, bets occasionally. Wants BET/FADE guidance without needing to understand the math. Highest-intent traffic around race weekends.

→ Free tier · Race week affiliate

SEGMENT 3

Serious Bettor

Already uses Action Network for NFL/NBA. Wants the same analytical rigour for MLB and motorsport. Values model transparency, ROI tracking, and full probability tables.

→ Patreon $12/mo · Pro subscription

SEGMENT 4

B2B / Media

Regional sportsbooks, F1 media, fantasy operators lacking internal data science. License MLB + motorsport prediction API at $5K–20K/month. Year 3 revenue stream.

→ Data licensing · B2B API

TECHNICAL ARCHITECTURE · DATABRICKS + AWS · 6 SPORTS LIVE

LAYER 1 · DATA INGESTION

All free or key-based REST APIs · no scraping

Soccer · EPL + La Liga

ESPN API (esp.1) · standings, fixtures, scores. Understat · xG and xGA per team and player. The Odds API (us,uk,eu regions) · h2h and totals markets from FanDuel, DraftKings, Bet365, DraftKings. Notebooks 11 (EPL) and 15 (La Liga) run Monday 8am. Dixon-Coles Poisson model applied in notebooks 12 and 16.

F1 · Telemetry + Results

OpenF1 · 18 free endpoints, 3.7Hz telemetry, pit stops, race control messages. Jolpica API (Ergast replacement) · full race results 2020–2025, qualifying times, constructor standings. FastF1 Python library · sector times, tyre compound per stint, pace delta from pole. Notebooks 01–02 ingest, notebook 03 trains LightGBM on 12 features tracked in MLflow.

NASCAR + Horses

The Odds API (NASCAR_cup_series) · win market odds from US books. racing-reference.info · race results fallback for signal scoring. DRIVER_DB · hardcoded track history by track type (intermediate/short_track/superspeedway/road_course). Notebook 18 auto-detects current race from 2026 schedule. Notebook 07 handles Kentucky Derby (May 2). Equibase integration pending for past performance data.

Storage: Databricks Delta Lake · dbc-f2f991dd-48fb · All credentials in secrets scope "f1"

LAYER 2 · MODELS + SIGNALS

Edge = model prob − book implied prob (6.5% vig adjusted)

EPL + La Liga

Dixon-Coles Poisson model. Attack/defense ratings from rolling xG. Dixon-Coles correction for low-scoring games. Poisson probability matrix for all scorelines. Home/draw/away win probabilities. BET signal: edge > 4%. FADE: edge < −4%. Accent normalization for Spanish team names. League avg goals: EPL 1.35, La Liga 1.42.

F1 · LightGBM

12 features: grid position, quali pace delta, rolling 5-race avg, DNF rate, circuit win rate, SC probability, overtake index. Three binary classifiers: win, podium, top-6. MLflow experiment tracking. Win LogLoss 0.2599 · Podium 0.0926 · Top-6 0.4075. BET >5% edge, FADE <−4%.

NASCAR · Book-anchored

Base = book implied probability (market signal preserved). Adjustments: track win history, track specialist score, tire management rating, manufacturer edge (Toyota +1.2%, Ford 0%, Chevy −1.5%). BET threshold 3% (higher variance). Auto-detects race from 2026 schedule. Signals written to S3 pending file, scored by notebook 13 post-race.

Signal reconciliation

Notebook 13 runs after each event. Fetches actual scores from ESPN + Odds API settled results. Fuzzy name matching with accent normalization. Scores WIN/LOSS per BET/FADE signal. Writes to api/results/history.json. Tracks: EPL, La Liga, NASCAR, F1, Horses per sport. Published openly on Results tab.

LAYER 3 · CONTENT PIPELINE

Claude API → draft review → one-click publish

Notebook 14 · Generation

Runs Mon/Thu/Sat. Calls claude-sonnet-4 with live model data as context. Generates 3–4 posts per run: EPL preview, F1 preview, horse racing, evergreen analysis. Each post has title, subtitle, lede, sections array, picks card. Writes to api/blog/drafts.json — never directly to live feed.

Admin Review · Draft UI

Admin tab → Blog Drafts shows all pending posts with full body, sections, and picks card. Actions: Publish (approve + go live), Edit (title/excerpt inline), Reject (delete from queue). Red badge shows pending count. All actions call Lambda directly — no downloads or manual S3 uploads required.

Lambda · oie-blog-publisher

AWS Lambda (Python 3.12) behind API Gateway HTTP API. Actions: approve, reject, edit, unpublish, update, publish_all. Reads/writes S3 directly. Fires CloudFront invalidation on every write. Auth: Bearer token. CORS configured for oneisenough.com. Cost: effectively $0/mo at current volume. Notebook 19 provides Databricks-side publish fallback.

LAYER 4 · DELIVERY + SITE

Single HTML file · no framework · no server · ~$42/mo total infra

S3 + CloudFront

S3 bucket oneisenough.com serves HTML and all JSON APIs. CloudFront distribution E29K3JKSZ5NQX7 provides HTTPS, global edge caching, and cache invalidation on every notebook run. All JSON endpoints under /api/* are no-cache headers for real-time data freshness. Estimated infra cost: ~$42/mo total.

5 sport pages

EPL (6 tabs: table, predictions, xG, value bets, scorers, title race). La Liga (identical structure). NASCAR (win probability, value bets, track notes, lineup — fully dynamic from S3). F1 (LightGBM signals, live odds, SC probability). Horses (Derby prep, morning line odds). CBB March Madness (seasonal). All fetch live from S3 JSON on page load.

Match card edge UI

probBox() helper renders model probability vs book implied probability with explicit edge % for each outcome. BET/FADE labels with green glow border on positive signals. Edge calculation: model% − (book odds vig-adjusted at 6.5%). Shared between EPL and La Liga match cards. Three site themes: dark (default), light, track (high-contrast sunlight mode). Persisted via localStorage.

Results + track record

Results tab shows full BET signal history across all sports sourced from api/results/history.json. Win/loss/ROI tracked per sport. FADE signals recorded as avoidance (not as wins unless verifiable counterfactual). Directional accuracy tracked internally. Published openly — transparency is a core differentiator vs opaque tout services.

DAILY + WEEKLY PIPELINE SCHEDULE

MONDAY 8AM

→ 12 (EPL bronze + model)
→ 16 (La Liga bronze + model)
(reconcile prior weekend results)
(blog draft generation)

THURSDAY 8AM

(NASCAR signals for weekend race)
→ 02 → 03 → 05 (F1 race week)
(F1 preview blog draft)
(score NASCAR after race Mon)

SATURDAY 8AM

→ 07 (Horse racing bronze + model)
(horse racing blog draft)
(social content generation)
Databricks job scheduling: pending setup

WHAT MAKES US DIFFERENT

Enterprise ML Infrastructure, Consumer Price

Databricks + Delta Lake + LightGBM — the same stack used by FanDuel internally. Six live data pipelines. Automated nightly reconciliation. No competitor in this space runs production-grade infrastructure. Everyone else is running spreadsheets or basic Python scripts.

Full Transparency + Live Track Record

Every prediction logged. Every result recorded publicly. ROI calculated from actual stored odds. MLB v3: 57.9% win rate on 57 signals, +10.7% ROI, 62.9% fade accuracy. Tiered signals (STRONG VALUE / VALUE / FADE) with public track record. No tipster site publishes like this. Transparency is the strategy — it compounds with every correct call.

Daily MLB Volume = Daily Affiliate Traffic

2,430 MLB games per season. Daily signals April–October. Every signal page is a potential affiliate click to FanDuel, DraftKings, or Kalshi. NFL/NBA/college sports competitors fight for the same readers on the same three days per week. We have daily content and daily traffic.

Dual Affiliate Channel — Books + Prediction Markets

FanDuel/DraftKings via Impact.com for traditional sportsbook audience. Kalshi (CFTC-regulated event contracts) for analytically sophisticated readers who self-select into our model-driven content. Two monetization channels, one content pipeline.

US Market Primary Focus

Every motorsport analytics competitor is UK/European focused. US sports betting is the fastest-growing betting market in the world. We cover F1, NASCAR, EPL, La Liga, and MLB — the full spectrum of what US bettors want, served from a US-first perspective.

THE MOAT — WHY THIS IS HARD TO COPY

A content site can copy our picks tomorrow. They cannot copy the infrastructure, the data history, or the model track record.

⬡

Delta Lake Feature Store

4,926 MLB game rows · 3 seasons · 34 features. 6 sport pipelines. Grows automatically every day.

⬡

Public Accuracy Track Record

43 results tracked. ROI calculated from actual stored odds. The history is itself a competitive asset — it compounds every night.

⬡

Automated Pipeline

MLB signals generated daily. Results reconciled nightly. Blog drafts generated on schedule. Zero manual data entry required.

⬡

Brand Authority

7 static pages indexed by Google. @OIEdge on X. Eagan Digital LLC forming. First credible US multi-sport analytics brand with a live track record.

REVENUE MODEL · FOUR STREAMS

STREAM 1 · FASTEST TO REVENUE

Affiliate Marketing

Refer users to licensed sportsbooks (FanDuel, DraftKings, BetMGM, Bet365). Earn per new depositing customer. No licensing required. Same model as Action Network and every major sports betting media company.

FanDuel CPA$25–35 per user

DraftKings CPA$75 per user

Bet365 RevShare30% lifetime

Year 1 target$18,000–36,000

STREAM 2 · CORE PRODUCT

Subscriptions

Free tier drives SEO and affiliate conversions. Pro tier ($14.99/mo) unlocks full model outputs, live race dashboard, qualifying alerts, pit window live updates, and historical accuracy tracker. Elite tier ($39.99/mo) adds API access and Discord.

Free tier$0 · SEO funnel

Pro tier$14.99/mo

Elite tier$39.99/mo

Year 2 target (200 Pro)$35,976/yr

STREAM 3 · SCALE REVENUE

Data Licensing

Once model has 12+ months track record, license prediction outputs to smaller sportsbooks, F1 media properties, and fantasy operators who lack internal data science capability. B2B API feed. White-label analytics for regional books entering the F1 market.

API data feed$5K–20K/mo

White-label$2K–8K/mo

Media licensing$1K–5K/mo

Year 3 target$120K–240K/yr

PROJECTED P&L · 3 YEAR

ITEM	YEAR 1	YEAR 2	YEAR 3
Affiliate — MLB (FanDuel/DraftKings)	$8,400	$42,000	$120,000
Affiliate — Motorsport (FanDuel/DraftKings)	$6,000	$24,000	$60,000
Affiliate — Kalshi prediction markets	$2,400	$12,000	$36,000
Affiliate — Horse Racing	$3,600	$18,000	$48,000
Subscriptions / Patreon	$1,800	$28,800	$86,400
Data licensing / B2B API	$0	$12,000	$120,000
TOTAL REVENUE	$22,200	$136,800	$470,400
Databricks + AWS	($960)	($2,400)	($6,000)
Odds API + data feeds	($480)	($960)	($1,800)
Domain + hosting + LLC	($420)	($240)	($480)
Content / SEO	($1,200)	($4,800)	($12,000)
TOTAL COSTS	($3,060)	($8,400)	($20,280)
NET PROFIT	$19,140	$128,400	$450,120

Note: Year 1 assumes affiliate accounts active Q3 2026 (pending LLC approval). MLB affiliate revenue modeled at 0.3% conversion of monthly unique visitors at $50 avg CPA. Kalshi modeled conservatively at 50% of FanDuel volume. Patreon/subscriptions assume 25 patrons by end of Year 1. Infrastructure costs remain low due to serverless architecture.

KEY RISKS + MITIGATIONS

Risk: Regulatory changes to affiliate model

Mitigation: Diversify across all three revenue streams. Subscription revenue is fully independent of affiliate regulation. Data licensing is B2B and regulatory-neutral.

Risk: Model underperforms over extended sample

Mitigation: Transparency is the hedge. We publish accuracy openly. A model that explains its uncertainty builds more trust than one that hides it. Blog content remains valuable regardless of model performance.

Risk: Data API access restricted or priced out

Mitigation: OpenF1 and FastF1 are open-source community projects with broad usage. Jolpica is a community replacement for Ergast. We also have Betfair API as a paid alternative. Analysis of public results data is always legally permissible.

Risk: Larger competitor enters the niche

Mitigation: Brand authority + track record moat. The first credible data-driven motorsport analytics brand in the US market compounds with every race weekend. History and accuracy records cannot be bought overnight.

18-MONTH PRODUCT ROADMAP · Updated April 2026

COMPLETE

Phase 1 · Foundation + Multi-Sport Launch

March – April 2026

✅ DONE

6 Live Sports Pipelines

F1, NASCAR, EPL, La Liga, MLB, Horse Racing on Bronze/Silver/Gold Delta Lake

✅ DONE

MLB LightGBM Model v3

5,003 training games. 36 features. 40/60 market blend. Tiered signals. 57.9% win rate, +10.7% ROI on 57 tracked signals. 62.9% fade accuracy.

✅ DONE

Automated Results Tracker

ESPN API nightly reconciliation. Public track record. ROI per $100 wagered.

✅ DONE

Blog + SEO Infrastructure

7 static pages indexed by Google. Sitemap submitted. @OIEdge on X.

IN PROGRESS

Phase 2 · Monetisation + Track Record

May – July 2026

PENDING LLC

FanDuel + DraftKings Affiliate

Eagan Digital LLC + EIN + Mercury account. Apply via Impact.com. $25–$100 CPA per depositor.

PENDING LLC

Kalshi Affiliate

CFTC-regulated prediction markets. Sports event contracts. Analytically sophisticated audience.

Q2 2026

Patreon + Mailchimp

Patreon live quietly. $5/$12/$25 tiers. Weekly email to subscribers before affiliate is active.

Q2 2026

Databricks Job Scheduling

MLB pipeline auto-runs daily. F1/NASCAR pre-race auto-run. Zero manual steps.

PLANNED

Phase 3 · Product Depth + Pro Tier

August – October 2026

Q3 2026

Pro Tier Launch

Stripe payments. Full model probability tables. Daily MLB email. Early signal access.

Q3 2026

NRFI Props + Totals Lines

Odds API props endpoint. Real NRFI and totals book lines. True edge on all 3 MLB markets.

Q3 2026

Backtesting Module

Simulate 2024–2025 MLB signal performance. Verified historical ROI. Trust accelerator.

Q3 2026

Kentucky Derby + UCL Pipeline

Horse racing model expanded. Champions League final coverage. Miami GP + Kentucky Derby content.

VISION

Phase 4 · Scale + Data Licensing

November 2026 – September 2027

Q4 2026

NFL + NBA Expansion

Largest US affiliate markets. Additive revenue on proven infrastructure. Year-round coverage.

Q4 2026

B2B API Product

First data licensing contract. Regional sportsbooks, F1 media, fantasy operators.

Q1 2027

Discord Community

Pro tier Discord. Daily MLB signal thread. Race watch parties. High LTV engagement.

Q2 2027

Acquisition Positioning

12+ months verified ROI. Diversified revenue. B2B contracts. Positioned at 8–12x revenue multiple.