Team Matchup
Team vs team
Quantile predictions turned into calibrated Monte-Carlo samples for spread, total and win probability — with the full distribution, not one number.
- Spread
- Over / Under
- Total range
- Win sim
NBA & NCAA Men's Basketball
Every prop, one tool.Modeled properly.
Points, rebounds, assists, blocks, steals, threes, turnovers, combos and full team matchups — each one returns a calibrated Monte-Carlo projection, not a guessed number.
POINTS/REBOUNDS/ASSISTS/BLOCKS/STEALS/3-POINT/TURNOVERS/COMBO PROPS/TEAM MATCHUP/NBA + NCAA/
The engine room
Nine-plus models. One simulation core.
Player Props
Points · Reb · Ast · Blk · Stl · 3PT · TO · Combos
Per-player distributions built on dynamic form features and opponent defensive rating, with a neural fallback when game logs are thin.
- Prop line
- Distribution
- Box score
- Confidence
College Hoops
NCAA Men's D-1
A parallel data stack for college basketball — team and player logs, transfer-portal aware roster mapping, the same simulation engine.
- Team logs
- Player form
- Freshman fallback
- Daily seed
The pipeline
From tip-off data to a number you can act on.
Stage 1
Build the game
Model artifacts load, features are built from nightly-refreshed schedules, rosters and logs, and quantiles are predicted.
Stage 2
Simulate it
Calibrated Monte-Carlo samples are drawn from those quantiles — thousands of versions of the same game.
Stage 3
Ask anything
Change a line, a minute total or a prop and the answer recomputes instantly against the samples you already have.
By the numbers
Every answer is counted, not guessed.
Nothing here is a formula's output. Each probability is a frequency over tens of thousands of simulated games — which is why you can ask a different question of the same run and get an answer instantly.
25,000
Monte-Carlo samples per team matchup
Player props draw 20,000 and combined props 30,000. Home and away scores are sampled together with shared game-factor noise, because real games have a shared character.
9
Prediction models
One team model, seven props, one combiner
99
Quantiles per head
The whole predicted shape, in one pass
1,588
Player cards
686 current · 902 all-time
30
NBA teams
Complete league coverage
13
Nightly refresh phases
Incremental, isolated failures
06:00
ET daily refresh
Logs, rosters, ratings, form
College basketball runs on the same engine with a separate data stack — see exactly what is covered, including what is not.
Pricing
Pay for predictions, not a subscription.
Purchased credits never expire and every model costs the same one credit — a full team matchup is priced identically to a single player prop. Buy more at once and the rate drops.
One credit buys
One model run — the simulation itself. Once it exists you can change the sportsbook line, test a range, flip between analytics tabs and re-read the distribution as many times as you like at no extra cost, because those questions query samples you already paid for.
Re-running does cost
Changing the scenario — different minutes, a different defender, a new opponent — rebuilds the simulation and spends another credit. That is the expensive part: 20,000 to 30,000 samples per run.
Billing is not live yet. SportzPRED is in private beta and access is by password, so nothing is for sale on this page today — these are the rates credit packs will launch at. Prices are in USD and exclude any tax that applies where you are.
Stop guessing the line.
Pick a game, set up the scenario, run the sim, and read the distribution instead of the headline.
Launch SportzPRED