About Us

About Sherlock Picks

A sports analytics platform powered by reproducible modeling, transparent reporting, and community support.

Our Mission

Sherlock Picks exists to make sports modeling understandable and usable for everyday members. The goal is not to promise wins, but to provide disciplined probability estimates, risk-aware staking ideas, and clear context around each recommendation.

Every model on this platform uses walk-forward backtesting so each fold is evaluated on forward periods the model has not seen during training. We track calibration, expected value, and historical behavior so members can evaluate performance with full context.

How It Works

01

Data Ingestion

Historical game results, upcoming schedules, injuries, market odds, and league-specific stats are ingested and normalized.

02

Backtest and Calibration

Models are evaluated with rolling walk-forward folds, then calibrated and scored by quality metrics before recommendation policies are applied.

03

Publishing and Access

Predictions, prop outputs, and reports are synced into the site. Feature visibility is role-based with supporter tiers.

Current Coverage

The live system currently supports 18 sports/leagues in the core pipeline, plus NBA player-prop modeling and an NBA first-basket module.

NFL NBA MLB NHL MLS EPL La Liga Bundesliga Serie A Ligue 1 Liga MX Primeira Liga Champions League Europa League College Football Men's College Basketball UFC WNBA

Technology Stack

LightGBM
Core model family
XGBoost
Model comparison option
CatBoost
Categorical support
Optuna
Hyperparameter tuning
Python
Pipeline runtime
Laravel
Site and sync layer
Stripe
Tip payments
MySQL
Application storage
Community Funded

How this project grows

Sherlock Picks is funded by supporters. Tips help cover hosting, data operations, and new feature work. Current priorities include better calibration quality, stronger prop coverage, and clearer in-app performance evidence.