About

Squared 2020 is a blog maintained and operated by statistician Justin Jacobs. This blog is primarily used to show how different types of statistical and analytical methods can be used in development and analysis of basketball analytics. Justin is a member of the American Statistical Association, a former NCAA basketball player, a former and current researcher in NBA Front Offices, and is a recipient of the Presidential Early Career Award in Science and Engineering; the highest STEM award for statisticians offered by the U.S. Government.

A PhD graduate in Statistics from the University of Maryland – Baltimore County and MS graduate in Mathematics from the University of Wisconsin – Milwaukee, Justin served for nearly a decade as a research statistician in the Department of Defense. From January 2016 through January 2018, Justin served a Principal Research Statistician at Sandia National Laboratories in Livermore, California. Justin’s research efforts are in spatio-temporal statistics, manifold learning, ranking analytics, streaming analysis, and recommender systems. Justin’s work has led to over 30 publications within the Department of Defense and a patent on pseduo-GPS methods. The views and commentary of this blog are strictly Dr. Jacobs’ own and does not reflects the views or opinions of his employers.

As a former collegiate player and statistician, Justin leverages his knowledge in both areas to create basketball analytics; as well as identify the statistical and mathematical properties of mainstream advanced analytics to identify areas of improvement, uncover misunderstood quantities, and establish rationale for using the analytics in question. Since the 2012-13 NBA season, Justin has worked with several NBA front offices in the areas of analytic development, coaching strategy, player valuation, and player performance through machine learning, artificial intelligence, and spatio-temporal statistics. Also during this time, Justin has worked with two NCAA teams between 2014 and 2016 on developing metrics for understanding player movement and rotation quality in games.

Disclaimer:

From January 2018 through August 2018, Justin joined the Orlando Magic full time to serve as their Senior Basketball Researcher. During that time, Justin took a brief hiatus on Squared2020 from focusing on machine learning, artificial intelligence, spatio-temporal analysis, and other advanced statistical methodologies on various types of NBA data. Given this experience, it is noted that no postings on this site are reflective of the research or analysis being performed in Orlando.

Similarly, posted work during this time does not reflect the current work or data being used for research or analysis being performed with his current team.

Justin encourages you to keep an eye out at some of the important conferences for sports analytics: Note: Reference of these conferences are strictly of the opinion of Justin Jacobs; not any of his present or previous employers.

 

 

Also, feel free to donate to improve the site: make it upgraded with a hosting site to better distribute code and have interactive designs. With the last round of donations, an upgrade was made and now plug-ins can be used; which allows for easier code copying!

Tiered Data Release — Tier 1: Season Summary Dataset

This product provides an authoritative, season-level summary of NBA player and team usage derived from hand-curated, video-based lineup reconstruction. Unlike box-score aggregates, all playing time has been corrected using inferred on-court presence, ensuring internal consistency between minutes, possessions, and scoring. Tier 1 is designed for researchers, writers, and analysts who need accurate season summaries and corrected exposure without access to proprietary possession-level data. Tier 1 includes player–season summaries with corrected minutes played, offensive and defensive possession counts, points scored and allowed while on court, season-aggregated on/off and net rating summaries, and team-level season totals with usage context. By design, Tier 1 does not include 5-on-5 lineup data, substitution timing or stint boundaries, possession sequences or play-by-play equivalents, or player impact model coefficients such as RAPM. Higher tiers introduce inferred rotation structure, lineup behavior, and possession-level inference. Tier 1 datasets are available for the 1984–85 through 1995–96 NBA seasons and are sold on a per-season basis. After completing checkout, purchasers should email the season they wish to receive along with proof of purchase to the address listed on the receipt. As additional games are reconstructed or annotations are refined, quarterly updates to the selected season are provided at no additional cost.

$99.00

Tiered Data Release — Tier 2: Rotation & Usage Profiles

This product provides season-level inference of NBA player rotation structure and in-game usage patterns, derived from hand-curated, video-based lineup reconstruction. Rather than reporting total minutes alone, Tier 2 characterizes when players were used within games and how roles evolved across the season. Tier 2 is designed for analysts, researchers, and advanced media users who need insight into rotation behavior, closing tendencies, and usage structure, without access to possession-level or lineup-level data. Tier 2 includes time-aligned player presence profiles aggregated across the season, quarter-level and game-clock usage summaries, inferred starter, bench, and closing roles, rotation stability and variability measures, and team-level rotation structure indicators. Rotation profiles are inferred from internally consistent minute- and presence-level reconstruction aggregated across observed games. By design, Tier 2 does not include 5-on-5 lineup data, substitution timestamps, stint boundaries, possession sequences, play-by-play equivalents, or player impact model coefficients derived from possession-level estimation. Higher tiers introduce full possession-level stints, lineup matchups, and player impact modeling. Tier 2 datasets are available for the 1984–85 through 1995–96 NBA seasons and are sold on a per-season basis. After completing checkout, purchasers should email the season they wish to receive along with proof of purchase to the address listed on the receipt. All purchased seasons include ongoing quarterly updates reflecting newly reconstructed games and improved rotation inference at no additional cost.

$299.00

Tiered Data Release — Tier 3: Possession-Level Stint Dataset

This product provides full possession-level 5-on-5 stint data reconstructed from hand-curated, video-based lineup annotation. Tier 3 represents the atomic game-level record from which season summaries, rotation profiles, and player impact estimates are derived. Tier 3 is designed for professional analysts, research groups, and modeling teams who require direct access to lineup context, possession attribution, and scoring outcomes, and who intend to perform independent impact estimation, matchup analysis, or simulation. Tier 3 includes game-dated possession-level stints with complete offensive and defensive lineups, corrected playing time within each stint, offensive and defensive possession counts, points scored and allowed, and explicit substitution boundaries defining lineup changes. All stints are internally consistent with reconstructed minute- and possession-level exposure and are suitable for downstream statistical modeling without reliance on box-score minutes or inferred play-by-play sequences. By design, Tier 3 does not include raw video, broadcast footage, annotation tooling, model code, or pre-computed player impact coefficients. Higher tiers introduce standardized player impact models, uncertainty quantification, and derived analytic products built directly from the Tier 3 dataset. Tier 3 datasets are sold on a per-season basis and provide game-dated, possession-level 5-on-5 stint data for the selected NBA season. Tier 3 datasets are available for the 1984–85 through 1995–96 NBA seasons and are sold on a per-season basis. After completing checkout, purchasers should email the season they wish to receive along with proof of purchase to the address listed on the receipt. Quarterly updates incorporating newly reconstructed games and consistency improvements are provided for the selected season at no additional cost.

$2,999.00

Tiered Data Release — Tier 4: Player Impact Models & Derived Analytics

This product provides standardized player impact estimates and derived analytic products computed directly from the Tier 3 possession-level dataset. Tier 4 represents the modeling and inference layer, translating raw lineup and possession structure into interpretable player-level measures with explicit uncertainty. Tier 4 is designed for professional analysts, research organizations, and decision-making groups who require validated impact estimates without independently implementing possession-level modeling pipelines. Tier 4 includes pre-computed player impact coefficients derived from possession-level lineup data, season-level offensive and defensive impact estimates, associated uncertainty bounds, and model-based summaries suitable for comparison, ranking, and downstream analysis. All estimates are generated using internally consistent possession attribution and exposure correction, and are aligned exactly with the underlying Tier 3 dataset for the selected season. By design, Tier 4 does not include raw possession-level stints, lineup tables, substitution boundaries, annotation tooling, or model source code. Tier 4 datasets are sold on a per-season basis and are available only in conjunction with, or derived directly from, the corresponding Tier 3 season dataset.

$4,999.00

Tiered Data Release — Tier 5: Custom Analytics & Advisory

Tier 5 provides bespoke analytic services and advisory support built directly on the underlying reconstructed datasets. This tier is designed for organizations, researchers, and media partners who require tailored analysis, custom modeling, or expert interpretation beyond standardized dataset releases. Tier 5 engagements draw on over a decade of applied experience across multiple NBA organizations and research partners, including work in advanced spatiotemporal analytics, draft and free agency modeling, and the development of novel analytic methodologies. Prior work has supported professional teams and external institutions, including collaboration with industry research groups such as Microsoft’s NBA-AI initiative. This purchase covers an initial scoping and consultation engagement to define objectives, feasibility, and deliverables. The scoping fee is credited toward any subsequent Tier 5 project agreement. All Tier 5 services are scoped individually and delivered under a custom agreement defining objectives, deliverables, and permitted use. The Tier 5 purchase covers an initial consultation and scoping engagement to assess objectives, data requirements, feasibility, and proposed deliverables. If a Tier 5 project agreement is initiated, the consultation fee is credited toward the contracted hours or project cost. If no engagement proceeds, the consultation fee is retained.

$500.00

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.