Skip to main content

College basketball: BartTorvik ratings and an ESPN box score

What you'll build: the top of BartTorvik's T-Rank table, a game's top scorers from ESPN's player box score, and the team totals pivoted into a side-by-side table.

Sources used​

SourceHostCall
BartTorvik / T-Rankbarttorvik.com/2024_team_results.csvsdv.torvik.torvikRatings({ year: 2024 })
ESPN Site API v2site.api.espn.com/apis/site/v2/sports/basketball/mens-college-basketball/summary?event=401600379sdv.mbb.espnMbbSummary({ section: 'boxscore_player' | 'boxscore_team' })

Offline fixtures: test/fixtures/torvik/torvik_ratings.csv (3 rows of the 2023-24 results CSV, captured 2026-06-18) and test/fixtures/espn/basketball_pbp/mbb_summary_401600379.json.gz (Maryland at Ohio State, 2024-02-10).

A provider that serves CSV​

Not every source is JSON. BartTorvik publishes CSVs (some with headers, some positional) and a couple of headerless JSON arrays, and rejects default programmatic user agents. The torvik family's runtime sends a browser-like UA and hands the raw body to a parser that knows each endpoint's format — torvikRatings reads the headed CSV, torvikPlayerStats the 67-column positional one (column names ported from hoopR). From the outside it is the same { parsed: true } call as everything else, returning rows keyed rank, team, conf, record, adjoe, adjde, barthag, adjt, wab, and ~40 more.

Values come back as strings (CSV has no types); the script casts the handful it prints. The "Duplicate headers found and renamed" line in the output is papaparse noting that Torvik's CSV repeats a rank header — the second becomes rank_1.

Two box-score sub-frames​

summary gives the box score two ways:

  • boxscore_player — one row per player with athlete_display_name, athlete_position, starter, did_not_play, and the stat columns named after ESPN's labels (minutes, points, rebounds, assists, field_goals_made_field_goals_attempted as the 4-6 display string, …).
  • boxscore_team — long form: one row per team × stat with stat_label, stat_display_value and home_away. Long form is the honest shape (ESPN's stat list differs by sport), and pivoting it to one row per stat with a column per team is six lines of JavaScript, which the script shows.

Box-score numbers are also strings here ("26"), so sort after Number().

The script​

examples/06_college_basketball_torvik_and_espn.mjs
// 06 — Men's college basketball: BartTorvik ratings + an ESPN box score.
//
// Shows: a non-ESPN provider on the same `{ parsed: true }` contract. BartTorvik
// serves CSV (with a header for team results) behind a browser-UA check — the
// `torvik` runtime handles both, so the wrapper returns rows like any other. Then
// the ESPN `summary` dispatcher's `boxscore_player` / `boxscore_team` sub-frames
// for one game (team stats are long-form: one row per team × stat).
//
// Sources: BartTorvik — barttorvik.com/2024_team_results.csv
//          ESPN Site API v2 — site.api.espn.com/apis/site/v2/sports/basketball/mens-college-basketball/summary?event=401600379
// Offline fixtures: test/fixtures/torvik/torvik_ratings.csv (2026-06-18, 3 rows),
//                   test/fixtures/espn/basketball_pbp/mbb_summary_401600379.json.gz (Maryland at Illinois, 2024-02-10).

import sdv from 'sportsdataverse';
import { setup } from './_offline.mjs';
import { printTable, round } from './_util.mjs';

setup();

const ratings = await sdv.torvik.torvikRatings({ year: 2024, parsed: true });
printTable(
  ratings.map((r) => ({
    rank: Number(r.rank),
    team: r.team,
    conf: r.conf,
    record: r.record,
    adj_oe: round(Number(r.adjoe), 1),
    adj_de: round(Number(r.adjde), 1),
    barthag: round(Number(r.barthag), 3),
    adj_t: round(Number(r.adjt), 1),
  })),
  ['rank', 'team', 'conf', 'record', 'adj_oe', 'adj_de', 'barthag', 'adj_t'],
  5,
  'T-Rank 2023-24 (torvik_ratings)'
);

const EVENT_ID = 401600379;
const players = await sdv.mbb.espnMbbSummary({ event_id: EVENT_ID, parsed: true, section: 'boxscore_player' });
const teams = await sdv.mbb.espnMbbSummary({ event_id: EVENT_ID, parsed: true, section: 'boxscore_team' });
console.log(`${players.length} player rows, ${teams.length} team-stat rows`);

printTable(
  players
    .filter((p) => !p.did_not_play)
    .map((p) => ({ team: p.team_abbreviation, player: p.athlete_short_name, pos: p.athlete_position, min: Number(p.minutes), pts: Number(p.points), fg: p.field_goals_made_field_goals_attempted, reb: Number(p.rebounds), ast: Number(p.assists) }))
    .sort((a, b) => b.pts - a.pts),
  ['team', 'player', 'pos', 'min', 'pts', 'fg', 'reb', 'ast'],
  8,
  'Top scorers (boxscore_player)'
);

// Pivot the long team frame to one row per stat with a column per team.
const abbrs = [...new Set(teams.map((t) => t.team_abbreviation))];
const byStat = new Map();
for (const t of teams) {
  if (!byStat.has(t.stat_label)) byStat.set(t.stat_label, { stat: t.stat_label });
  byStat.get(t.stat_label)[t.team_abbreviation] = t.stat_display_value;
}
printTable([...byStat.values()], ['stat', ...abbrs], 8, 'Team totals (boxscore_team, pivoted)');

Opens a Node sandbox in a new tab with this script as index.mjs; runs live (no API key for ESPN).

Output​

Output of node examples/06_college_basketball_torvik_and_espn.mjs (offline, against the committed fixtures):


## T-Rank 2023-24 (torvik_ratings)
| rank | team        | conf | record | adj_oe | adj_de | barthag | adj_t |
| ---- | ----------- | ---- | ------ | ------ | ------ | ------- | ----- |
| 1    | Connecticut | BE   | 37-3   | 128    | 91.3   | 0.98    | 64.6  |
| 2    | Houston     | B12  | 32-5   | 119.6  | 86.5   | 0.977   | 63.4  |
| 3    | Purdue      | B10  | 34-5   | 126    | 94.5   | 0.965   | 67.1  |
(3 rows, all shown)
30 player rows, 46 team-stat rows

## Top scorers (boxscore_player)
| team | player       | pos | min | pts | fg    | reb | ast |
| ---- | ------------ | --- | --- | --- | ----- | --- | --- |
| MD   | J. Young     | G   | 48  | 26  | 8-22  | 7   | 4   |
| OSU  | B. Thornton  | G   | 43  | 24  | 10-20 | 7   | 4   |
| MD   | D. Scott     | F   | 47  | 19  | 5-9   | 7   | 1   |
| OSU  | R. Gayle Jr. | G   | 37  | 17  | 4-12  | 2   | 3   |
| MD   | J. Reese     | F   | 49  | 13  | 4-15  | 13  | 2   |
| MD   | M. Traore    | F   | 20  | 8   | 4-6   | 5   | 0   |
| OSU  | J. Battle    | F   | 44  | 8   | 4-11  | 6   | 0   |
| OSU  | F. Okpara    | F   | 33  | 7   | 2-3   | 6   | 0   |
(19 rows, first 8 shown)

## Team totals (boxscore_team, pivoted)
| stat               | MD    | OSU   |
| ------------------ | ----- | ----- |
| FG                 | 24-62 | 29-66 |
| Field Goal %       | 39    | 44    |
| 3PT                | 7-17  | 4-17  |
| Three Point %      | 41    | 24    |
| FT                 | 20-30 | 17-20 |
| Free Throw %       | 67    | 85    |
| Rebounds           | 42    | 39    |
| Offensive Rebounds | 11    | 6     |
(23 rows, first 8 shown)

Variations​

  • sdv.torvik.torvikTeamFactors, torvikGameStats, torvikGameSchedule, torvikPlayerStats cover the rest of the site; women's T-Rank is sdv.torvik.bartWbb*.
  • sdv.mbb.kenpom* wraps KenPom for subscribers (login via KENPOM_EMAIL / KENPOM_PW).
  • The stats.ncaa.org-derived season releases (sdv.mbb.loadMbbPbp, …) are the bulk path; see release loaders.

Next steps​