GSGhost Screen Analytics
Work with us

Custom basketball analytics, built and tested the same way as our public models.

Ghost Screen Analytics takes on analytics work for teams, agencies, media and fans. You bring the question. We write down how the answer will be tested before running anything, then deliver the result, the method and every test, including the ones that failed.

Start a project

What we do

Player evaluation

How good is a player, what is he good at, and how sure can anyone be? Ratings with ranges, not single numbers.

Example: the Three Gates rim-protection grades.

Draft and development

Prospect scores from college numbers and draft slot, breakout and decline projections, and how much each of those can be trusted.

Example: College to NBA and the Breakout Model.

Contracts and value

Who is underpaid or overpaid, and how the answer changes with the way value is counted.

Example: the NBA's most underpaid player.

Health and workload

Injury risk, games-missed projections and which kinds of load actually matter, built on the NBA's official injury reports.

Example: Load & Durability.

Forecasts and simulations

Season, playoff, award and single-game odds from player ratings, run as many times as the question needs.

Example: the Season Forecast, 1,000,000 seasons a day.

Dashboards and explainers

Interactive pages and charts that let your audience or staff explore the numbers themselves, refreshed automatically.

Example: every page on this site.

Who it is for

Teams and staffSecond opinions on players, lineups, draft boards and workload.
Agents and agenciesContract value and the case for a player, with the numbers shown.
Media and creatorsData for a story, a segment or a show, plus charts ready to publish.
Fans and fantasy playersCustom questions answered properly, from one-off reports to tools.

How it works

  1. Tell us the questionEmail what you want to know, the deadline and what the answer is for.
  2. Agree on the testsBefore any results, we write down what the answer has to pass to count.
  3. Build and runModels from public NBA data, or your own data if you have it.
  4. DeliverA written report, the charts or an interactive page, the method and every test result.

See the work first

Seven public models →Each with its rules, code and pass/fail record. The newsletter →How the work reads when it is written up.

Have a basketball question?

Send a short note covering:

  • the question you want answered
  • who it is for and how it will be used
  • your deadline
Email us ajay@gsanalytics.org