How It Works — MVP Capping

// THE_PROCESS

ML models analyze game data

Our machine learning models ingest historical game data, player statistics, team trends, and real-time odds to produce a predicted outcome for each matchup. Models are built specifically for MLB and are continuously evaluated against real results.
Predictions are published before the game

Each model prediction is logged at the time it is released — before the first pitch. The odds available at release time are recorded alongside the prediction. Nothing is added, changed, or removed after the game is played.
Results are tracked automatically

Once the game finishes, the actual result is matched against every prediction. Wins, losses, and pushes are all counted — nothing is filtered out. The model record reflects every call the model made, good or bad.
Performance is reported in units

Results are expressed in units rather than dollar amounts so the record is independent of bet sizing. Standard unit calculations are applied using the recorded release-time odds. ROI is computed across all predictions for each model.
The full record is available to everyone

The public prediction record — every call, every result — is visible to all visitors, no account required. Registered users gain access to deeper analytics, individual model performance pages, and backtesting tools.

// PLATFORM_FEATURES

🤖

Machine Learning Models

Predictive models built on real MLB data. Each model has its own strategy, feature set, and tracked record.

📋

Pre-Game Predictions

All predictions are locked in and timestamped before game time. No after-the-fact entries.

📈

Live Odds Tracking

The line at time of release is what gets tracked — no adjusting after odds move or results come in.

✅

Full Outcome Recording

Wins, losses, and pushes are all included. Nothing is filtered out to inflate the model win rate.

🔒

Immutable Record

Past predictions cannot be deleted or backdated. The model record stands as-is.

📊

ROI Analytics

Return on investment is calculated across all predictions using release-time odds and standard unit sizing.

🔬

Backtesting (Pro)

Pro subscribers can run backtests on historical data to evaluate and compare model performance over time.

🌐

Public Transparency

The entire prediction record is publicly accessible. No sign-up required to view model history.

// KEY_TERMS

Term	Definition
Unit	A standardized measure of bet size used so results are comparable regardless of the dollar amount wagered. One unit equals a fixed percentage of a bankroll.
ROI	Return on investment. Calculated as net profit (in units) divided by total units wagered, expressed as a percentage.
Release-time odds	The odds that were available when the model prediction was published. These are what the unit calculation is based on — not the closing line.
Win / Loss / Push	Win: the prediction was correct. Loss: the prediction was wrong. Push: the game ended in a tie or the bet was refunded. All three outcomes are counted in the record.
Model	A machine learning algorithm trained on historical data to predict the outcome of a game. Each model has its own feature set and tracked performance history.
Backtest	Running a model against historical data to measure how it would have performed in the past. Available to Pro subscribers.

// PLANS

Free

Browse the platform and view the public prediction record. No payment required.

Basic Early access

ML model predictions, performance tracking, and ROI analytics — free during early access.

Pro

Everything in Basic, plus the ability to run backtests on your own or existing models.

Ready to see the model record?

No account needed to browse. Sign up when you want to follow along with full analytics.