Browse › Novel Load Bearing Index
Analytics module · as of
Novel Load Bearing Index
Most one-star-fragile: DEN (Nikola Jokić, delta_winprob 0.5822). Two estimators name the same #1 player for 1/30 teams (directional cross-check, different seasons).
confirmednull (a finding)not testabledescriptivepending
Novel Load Bearing Index
Load-Bearing IndexLBI
Most one-star-fragile: DEN (Nikola Jokić, delta_winprob 0.5822). Two estimators name the same #1 player for 1/30 teams (directional cross-check, different seasons).
What it measures
Per team, the win-prob dependence on its single most-indispensable player, under two independent estimators, with a cross-method agreement flag.
The formula
LBI_a = max_player (p_win_with - p_win_without) [Elo on/off]; LBI_b = max_player (win_rate_active - win_rate_missed) [raw]; agreement = same #1 player.
The results
| team | estimator a elo onoff / player name | estimator a elo onoff / delta winprob | estimator a elo onoff / on off net rating delta | estimator b raw withwithout / player name | estimator b raw withwithout / delta win rate | estimator b raw withwithout / n active | estimator b raw withwithout / n missed | agreement same player |
|---|---|---|---|---|---|---|---|---|
| DEN | Nikola Jokić | 0.5822 | 23.713 | Aaron Gordon | 0.1818 | 36 | 44 | no |
| OKC | Shai Gilgeous-Alexander | 0.5171 | 15.763 | Ajay Mitchell | 0.2685 | 57 | 23 | no |
| LAL | Dorian Finney-Smith | 0.4656 | 14.944 | Rui Hachimura | 0.2626 | 68 | 14 | no |
| MIL | Giannis Antetokounmpo | 0.4441 | 14.231 | Kyle Kuzma | 0.3659 | 69 | 12 | no |
| LAC | Ivica Zubac | 0.4051 | 11.965 | Brook Lopez | 0.4038 | 75 | 7 | no |
| IND | Pascal Siakam | 0.3872 | 11.457 | Pascal Siakam | 0.2075 | 62 | 15 | yes |
| ORL | Franz Wagner | 0.3816 | 13.851 | Moritz Wagner | 0.1833 | 36 | 5 | no |
| CHI | Lonzo Ball | 0.3758 | 13.374 | Jalen Smith | 0.3327 | 53 | 19 | no |
| ATL | Mouhamed Gueye | 0.3668 | 13.562 | Dyson Daniels | 0.1789 | 76 | 5 | no |
| CLE | Isaac Okoro | 0.303 | 8.728 | Sam Merrill | 0.2115 | 52 | 28 | no |
| DAL | Dereck Lively II | 0.2878 | 9.34 | Naji Marshall | 0.3378 | 74 | 5 | no |
| NOP | Zion Williamson | 0.2823 | 15.772 | Karlo Matković | 0.3038 | 62 | 12 | no |
| NYK | Karl-Anthony Towns | 0.2821 | 7.982 | Jalen Brunson | 0.2471 | 74 | 7 | no |
| BOS | Luke Kornet | 0.2785 | 8.218 | Neemias Queta | 0.0842 | 76 | 5 | no |
| SAS | Victor Wembanyama | 0.2781 | 11.135 | Carter Bryant | 0.2433 | 71 | 11 | no |
| MEM | Ja Morant | 0.2731 | 8.072 | Santi Aldama | 0.2752 | 43 | 6 | no |
| BKN | Day'Ron Sharpe | 0.2669 | 14.902 | Ochai Agbaji | 0.2 | 20 | 6 | no |
| PHX | Monte Morris | 0.2593 | 8.334 | Ryan Dunn | 0.1548 | 70 | 12 | no |
| MIA | Davion Mitchell | 0.2517 | 8.354 | Kasparas Jakučionis | 0.3061 | 53 | 9 | no |
| GSW | Gui Santos | 0.2378 | 6.653 | Moses Moody | 0.3 | 60 | 10 | no |
| POR | Jabari Walker | 0.2364 | 7.676 | Rayan Rupert | 0.3988 | 48 | 7 | no |
| HOU | Steven Adams | 0.2327 | 6.503 | JD Davison | 0.2637 | 28 | 52 | no |
| SAC | Domantas Sabonis | 0.2249 | 6.611 | Maxime Raynaud | 0.1588 | 74 | 8 | no |
| UTA | John Collins | 0.2156 | 15.371 | Lauri Markkanen | 0.3423 | 42 | 16 | no |
| TOR | Jakob Poeltl | 0.2093 | 7.751 | Immanuel Quickley | 0.169 | 70 | 12 | no |
| MIN | Naz Reid | 0.2092 | 5.85 | Jaden McDaniels | 0.1027 | 73 | 8 | no |
| WAS | Richaun Holmes | 0.2032 | 14.126 | Khris Middleton | 0.2863 | 34 | 15 | no |
| CHA | Moussa Diabaté | 0.1505 | 7.99 | Brandon Miller | 0.4543 | 65 | 17 | no |
| DET | Tobias Harris | 0.1447 | 4.139 | Duncan Robinson | 0.3532 | 77 | 5 | no |
| PHI | Guerschon Yabusele | 0.1371 | 5.279 | Jared McCain | 0.3089 | 37 | 7 | no |
Prior art
INCREMENTALWe searched for prior work before claiming anything. Here is what exists:
On/Off Win Probability tables (inpredictable, stats.inpredictable.com/nba/onoff.php), 'record-without-star' splits, and player-impact-on-win-chances (arXiv 1604.03186) all exist. Single-method fragility is NOT new. The novel twist is a standardized league-wide fragility ranking CROSS-VALIDATED by two independent estimators with an agreement flag.
Declared confounds
- both estimators carry the ROSTER CONFOUND both source files label explicitly: why a player missed is entangled with injury clustering, rest, blowout pulls, and replacement lineup -- none controlled. NOT a causal player-impact estimate.
- the feeds cover DIFFERENT seasons (Elo on/off = 2024-25; raw with/without = 2025-26), so agreement is DIRECTIONAL corroboration, not a paired estimate.
- estimator (a) is win prob vs a league-average opponent on a neutral floor; estimator (b) is raw team win-rate active-vs-missed. The two axes are related but not the same scale.
scripts/platformkit/analytics_showcase/out/cf_star_removal.json
scripts/platformkit/analytics_showcase/out/ctx_lineup_proxy.json
as_of estimator a: 2024-25 (Elo end-of-season) x 2024_25 on/off slice
as_of estimator b: 2025-26 regular season (through 2026-04-12)
edge_claimed: false · descriptive_only
Ask Scout about this