CASE STUDY 02 · COMPLETED ANALYSIS

A closer look at championship teams.

How did the 2024 Boston Celtics compare with the four NBA champions before them? I used a small SQLite dataset to ask a focused question about regular-season records.

SQLSQLiteSports analytics

The five-team comparison

Celtics win rate
78.0%
2024 record
64–18
Champions compared
05

01 / THE QUESTION

The question

The Celtics won 64 regular-season games in 2023–24. Raw wins are awkward to compare across these champions because the 2019–20 Lakers played 71 games and the 2020–21 Bucks played 72. I compared win percentage instead: wins divided by games played.

02 / THE EVIDENCE

Regular-season win percentage · 2020–24 NBA champions
  1. 2024 Celtics64–18 · 78.0%
  2. 2020 Lakers52–19 · 73.2%
  3. 2023 Nuggets53–29 · 64.6%
  4. 2022 Warriors53–29 · 64.6%
  5. 2021 Bucks46–26 · 63.9%

Bars start at zero and use a common 0–100% scale. Percentages are calculated from each team’s regular-season wins and losses, then rounded to one decimal place.

03 / THE METHOD

The SQL method

I kept one row per championship team, with its season, wins, and losses. The query calculates the same measure shown above and sorts the results. The public sample ↗ runs this query against a five-row CSV in an in-memory SQLite database.

SELECT team_name, year,
       wins, losses,
       ROUND(
         100.0 * wins /
         (wins + losses), 1
       ) AS win_pct
FROM championship_teams
ORDER BY win_pct DESC,
         year DESC;

Season records: Celtics, Lakers, Nuggets, Warriors, and Bucks (NBA).

04 / THE TAKEAWAY

The takeaway

Boston had the highest regular-season win rate in this five-champion sample: 78.0%, compared with 73.2% for the next team, the 2020 Lakers. That is a descriptive result about these five seasons. The sample contains only champions, so it cannot establish what predicts a title or which team was strongest across eras.

A predictive study would need all team-seasons, including teams that did not win, and a held-out test period. Other measures such as scoring and assists can describe style, but they cannot fix that sampling problem.

← Back to selected work