Launching Soon · 3 Sport Models

See the edge
before the books do.

We don't sell picks. We built the model. Obsidic runs thousands of Monte Carlo simulations per event to map probability distributions, quantify market inefficiencies, and surface edges across MLB, PGA, and Tennis so you can make smarter decisions with better data.

✓ You're in. Watch your inbox.
LIVE PREVIEW
OBSIDIC · TENNIS DASHBOARD
Daily Intelligence
Tuesday, July 8
15 games tracked
Slate Avg
8.7
runs
Total Vol
131
proj runs
Avg SP K
5.8
per start
Top YRFI
62%
TEX@COL
Best Edge
+4.2%
NYY ML
Projected Totals
10 9 8 7 O/U avg 8.8 10.1 9.8 9.0 8.5 7.8 7.2 TEX COL NYY LAD CLE MIA
Strikeout Leaders
1 Skenes
8.4
2 Cole
7.4
3 Ohtani
6.9
4 Wheeler
6.1
Top Confidence Pick
NYY
Cole
vs
BOS
Houck
62.4%
NYY projected winner
Total
9.2
Spread
-1.5
YRFI
54%
Quick Insights
🔥3 games projected over 9 runs
📋87% of lineups confirmed
💨Coors has 1.012 carry · hitter friendly
💰Best edge: NYY ML at +4.2%
Today's Matchups
NYYvsBOS
Cole9.2 O/UHouck
LADvsSF
Ohtani8.5 O/UWebb
TEXvsCOL
deGrom10.1 O/UQuantrill
ATLvsPHI
Fried8.1 O/UWheeler
⚡ Interactive Preview

Your Daily
Command Center

Three sports. Three purpose-built models. Explore today's projections with full probability breakdowns, simulation distributions, and edge quantification. Make informed picks with data.

🎾
LIVE MODEL
Tennis
ATP & WTA match predictions · 2,381 matches backtested
71.8%
Best-of-5 Acc
65.6%
Overall ML
113
ML Features
10K
Sims / Match
Sinner vs Alcaraz63%
Djokovic vs Fritz69%
Swiatek vs Gauff67%
Explore Tennis Dashboard →
⚾
LIVE MODEL
MLB
Game winners, run totals, player props · 2,416 games backtested
80.2%
≥70% Conf Acc
64.5%
Overall Winner
3.4M+
Pitches Tracked
5K
Sims / Game
NYY vs BOS · ML72%
Judge 1.5+ Hits68%
Ohtani 8.5+ K66%
Explore MLB Dashboard →
COMING SOON
⛳
COMING SOON
PGA Tour
Tournament sims, Strokes Gained, H2H matchups
6
SG Components
200+
Players Modeled
H2H
Matchup Markets
5K
Sims / Tourney
Scheffler T571%
McIlroy T1064%
Hovland T2073%
Explore PGA Dashboard →
COMING SOON

The PGA Golf model is currently in development.
Sign up to get notified when it launches.

TENNIS
· DAILY ANALYSIS
Daily Intelligence
Saturday, March 1
8 matches tracked
Best Edge
+23.8%
Brengle ML
Matches
8
ATP + WTA
Tournaments
4
active today
Avg Elo Gap
+62
best plays
Sims / Match
10K
Monte Carlo
Win Probability by Match
Hanfmann
53.2%
Brengle
59.9%
Sasnovich
64.8%
Townsend
49.4%
Darderi
46.8%
Surface Split
H
6
Hard
C
2
Clay
G
0
Grass
Tour Split
ATP
3
Men's
WTA
5
Women's
Match Props Overview
MatchAvg GamesTB%Straights%
Hanfmann vs Darderi 23.4 40.1% 56.6%
Brengle vs Gibson 21.9 21.0% 59.7%
Townsend vs Stearns 23.3 29.2% 54.6%
Day vs Sasnovich 22.1 33.8% 55.2%
Best Play · 3 Stars
Indian Wells · Hard
MB
M. Brengle
Elo 1476
vs
T. Gibson
Elo 1472
TG
59.9%
Brengle projected winner
Edge
+23.8%
EV/$100
+$65.97
ML Odds
+177
Avg Games21.9
Straights%59.7%
TB%21.0%
1st Set61.1%
Quick Insights
🎯Brengle ML +177: +23.8% edge, 3-star play
🔥Hanfmann +164 vs Darderi: +15.3% edge
🎾4 tournaments active: Santiago, IW, Merida, Austin
📊113 ML features per match, 10K sims each
Today's Elo Rankings
1 Sasnovich
1712
2 Stearns
1683
3 Townsend
1658
4 Hanfmann
1640
5 Darderi
1545
Today's Matches
ATP Santiago · Clay
Hanfmann vs Darderi
Elo 1640 Edge +15.3% Elo 1545
WTA Indian Wells · Hard
Brengle vs Gibson
Elo 1476 Edge +23.8% Elo 1472
WTA Austin · Hard
Townsend vs Stearns
Elo 1658 Edge -1.6% Elo 1683
WTA Indian Wells · Hard
Day vs Sasnovich
Elo 1520 Sasnovich fav Elo 1712
📊 Backtested Performance

Numbers Don't Lie.
Neither Do Ours.

Every chart below is from genuine out-of-sample backtesting. Not cherry-picked. Not curve-fit. Real predictions scored against real outcomes.

Accuracy by Confidence Tier 2,416 games
90% 75% 60% 45% coin flip 54.2% 50-55% n=598 59.2% 55-60% n=551 64.9% 60-65% n=441 68.6% 65-70% n=347 80.2% 70%+ n=479 Model Confidence
Higher confidence = higher accuracy. The model gets sharper as conviction grows.
Monthly Winner Accuracy 2025 Season
75% 65% 55% 45% 63.4% Apr 404 61.8% May 424 68.0% Jun 403 66.1% Jul 383 62.8% Aug 425 65.3% Sep 377 avg 64.5%
Consistent performance across the full season. June peaked at 68.0%.
64.5%
Overall Accuracy
80.2%
≥70% Conf Picks
2,416
Games Backtested
0.22
Brier Score
68.0%
Best Month (Jun)
Accuracy by Surface 2,381 matches
Grass 70.8% 271 Clay 65.6% 422 Hard 64.8% 1,688 matches
Grass courts show strongest signal. Serve dominance makes outcomes more predictable.
Accuracy by Tour & Format 4,762 Matches
80% 70% 60% 50% 64.7% ATP 1,266 66.6% WTA 1,115 71.8% Best-of-5 252 64.9% Best-of-3 2,129
Best-of-5 hits 71.8%. Longer format reduces variance. WTA edges out ATP.
65.6%
ML Accuracy
71.8%
Best-of-5
2,381
Matches Backtested
113
ML Features
70.8%
Best Surface (Grass)
COMING SOON

The PGA Golf model is currently in development.

Strokes Gained Model Accuracy Projected vs Actual
Off the Tee r=0.91 Approach r=0.87 Around Green r=0.72 Putting r=0.58
Tee-to-green metrics are far more predictive than putting. Our model weights accordingly.
Placement Prediction Accuracy 150 tournaments
100% 80% 60% 40% 68% Top 5 72% Top 10 81% Top 20 91% Made Cut
Wider placement windows hit at higher rates. Made Cut predictions at 91% accuracy.
r=0.91
SG: Off the Tee
91%
Made Cut Accuracy
150
Tournaments Tested
6
SG Components
ShotLink
Data Source
🧠 Under the Hood

Built for Edge,
Not Entertainment

Every component is designed to find where the market is wrong, then quantify exactly how wrong.

🎰
Monte Carlo Engine
5,000–10,000 simulations per event mapping the full probability distribution of every possible outcome.
Up to 10K sims/event
📊
Machine Learning Algorithm
Gradient-boosted trees trained on 5+ years of data with Bayesian hyperparameter optimization.
12,150+ games backtested
🌦️
Environment Adjustments
Wind, temperature, altitude, and park factors for MLB. Surface speed and conditions for tennis. Course layout and weather for golf. Every model accounts for where the game is played.
Context-aware across all sports
🤝
H2H Matchup Context
Tennis H2H by surface and format. MLB pitcher-batter splits and platoon matchups. Golf head-to-head market pricing. Every prediction knows who's facing who.
Matchup-aware across all sports
🔬
113+ Engineered Features
113 features for tennis (Elo, serve stats, fatigue). 80+ for MLB (Statcast, bullpen, park). 40+ for golf (Strokes Gained, course fit). Each model is purpose-built.
Sport-specific pipelines
📄
Published Methodology
Every data source, every feature, every training decision is documented publicly. No black boxes. Read the full methodology.
100% transparent
📱 Any Screen, Full Edge

Your Edge.
Everywhere.

Desktop at home. iPad on the couch. Phone at the bar. Watch the dashboard scroll through MLB, PGA, and Tennis. The same powerful analytics adapt to every screen.

🎾 Tennis
⛳ PGA
⚾ MLB
⚙️ Pipeline

Data In.
Edge Out.

From raw data ingestion to published methodology, every step is automated, backtested, and transparent.

01

Ingest Raw Data

Statcast pitch data for MLB. ATP/WTA match and serve stats for tennis. ShotLink performance data for golf. All processed through automated pipelines.

AUTOMATED
02

Feature Engineering

113+ custom features for tennis, 80+ for MLB, 40+ for golf. Serve dynamics, Elo ratings, park factors, Strokes Gained, matchup context, and more.

113+ FEATURES
03

Model Prediction

Gradient-boosted ML models + up to 10,000 Monte Carlo simulations map entire probability distributions per event.

10K SIMS
04

Edge Detection

Compare model probabilities against market odds across all three sports. Surface mispriced lines and quantify exactly how much value exists.

ALL 3 SPORTS
05

Full Transparency

Every data source, feature, and training decision is published. Read the full methodology for each sport. No black boxes, no hidden logic.

OPEN SOURCE
💬 Early Testers

Don't Take Our Word.
Take Theirs.

What beta testers and early access members are saying about the Obsidic projection engine.

JM
Jake M.
Sports Bettor · 6 years
⭐⭐⭐⭐⭐
"I've used every public model out there. FanGraphs, Statcast tools, you name it. First time I ran Obsidic's projections against my own, it was picking up edges I wasn't even looking for. The F5 innings market alone completely changed how I bet."
DC
Derek C.
DFS & Sports Analytics
⭐⭐⭐⭐⭐
"The player prop projections are absurd. It takes into account multiple factors not just past performance. That's the difference between getting +110 and getting -130."
RT
Ryan T.
Quantitative Bettor · Former Equity Trader
⭐⭐⭐⭐⭐
"I ran the backtest data against closing lines for the full 2025 season. Consistent CLV across moneylines, totals, and F5s. This isn't a tout service selling picks. It's a legitimate statistical edge with the data to back it up."
MW
Marcus W.
Sharp Bettor · 10+ years
⭐⭐⭐⭐⭐
"What sold me is the park factor modeling. Most public tools treat every stadium the same. Obsidic adjusts for fence distances, wind patterns, elevation, down to the batter level. That kind of granularity matters when you're grinding totals and F5s every day."
AP
Alex P.
Analytics Enthusiast
⭐⭐⭐⭐⭐
"I just love the data. The dashboard Obsidic is building is genuinely beautiful to use, the projections update daily, and it's taught me more about how games actually play out than years of watching baseball."
SK
Sam K.
Bettor · MLB Focus · 4 years
⭐⭐⭐⭐⭐
"My biggest problem was always second-guessing myself. With Obsidic I'm not guessing. I can see the probability distribution, the edge over the line, and the model's confidence. I finally have a process I trust enough to stick to."
🚀 Early Access
Stop Crunching.
Start Winning.

Up to 10,000 simulations per event. ML-driven projections. 3.4M+ pitches and 46,400+ tennis matches analyzed. MLB and Tennis models are live. PGA Golf coming soon. Sign up to get access to all three sport models as they launch.

✓ You're in. We'll reach out soon.

🔥 Only 237 founding member spots remaining