Blog

Ranking benchmark

Model Zero vs the fantasy basketball field

Model Zero won more matchups than it lost against verified preseason boards from ESPN, Yahoo analysts, FantasyPros, Hashtag Basketball, RotoBaller, and SportsEthos. Every result is published.

September 19, 2026· Ball Ranks
ESPNYahooHashtag BasketballFantasyProsRotoBallerSportsEthosBall Ranks
55
boards
14
seasons
32–23
per-game record
32 wins · 23 losses · every source linked

This page covers every verified site and analyst board in the archive.

See the Yahoo analyst record →

The comparison

Pick a year.

Lower miss score wins.

Per-game value
Every board is graded against Model Zero's per-game ranks and the actual per-game finish.
Model Zero went 22 in 2025-26
The whole field
Model Zero's edge against every board
LostWon
Source / analyst
16.7%Even+16.7%
Result
Yahoo · 177 players
-4.3%
RotoBaller · 178 players
-2.6%
SportsEthos · 207 players
+2.1%
ESPN
ESPN · 131 players
+16.7%

Each bar is Model Zero's percentage edge inside that source's own matched-player pool. Use it to compare direction and relative edge—not raw miss scores across differently sized boards.

The table

2025-26, line by line

Source / analystPlayersModel ZeroTheir scoreResult
SportsEthos
20732.68733.377Won by 0.69
RotoBaller
17826.59625.931Lost by 0.67
17728.63127.459Lost by 1.17
ESPN
ESPN
13120.07924.107Won by 4.03

Across every season

Best win, toughest loss, closest call.

Biggest Model Zero win
ESPN
2024-25
Won by 6.28
Biggest loss
Lost by 3.49
Closest match
FantasyPros consensus · FantasyPros
2012-13
Won by 0.01

Sleeper receipts

One for us. One for them.

Model Zero's standout sleeper
Kris Dunn
2022-23 · vs FantasyPros consensus
The pick
#149
Other side
#336
Actual
#65
FantasyPros consensus's standout sleeper
C.J. McCollum
2015-16 · vs FantasyPros consensus
The pick
#93
Other side
#244
Actual
#40

Experimental all-time leaderboard

Everybody, stacked up.

Lower is better. Board size is normalized, multiple boards are averaged within a season, and every season counts equally. No small-sample adjustment is applied.

#13 seasons
RotoBaller
Normalized miss13.98
#26 seasons
Yahoo · includes official boards
Normalized miss14.32
#311 seasons
FantasyPros consensus
FantasyPros
Normalized miss14.42
#49 seasons
Normalized miss14.45
#514 seasons
Model Zero
Ball Ranks
Normalized miss15.13
#62 seasons
RotoBaller consensus
RotoBaller · small sample
Normalized miss15.32
#73 seasons
SportsEthos
Normalized miss15.33
#88 seasons
Hashtag Basketball
Hashtag Basketball
Normalized miss15.64
#92 seasons
RotoBaller · small sample
Normalized miss16.24
#105 seasons
ESPN
ESPN
Normalized miss16.45
#114 seasons
Normalized miss16.79

Normalized miss is each board’s weighted average rank error as a percentage of that board’s full matched-player rank range. Only sources with at least two scored seasons appear; two-season entries are labeled as small samples.

Sources

Check every board.

Every matchup links to the preseason board that was scored. Each board and Model Zero are graded against completed-season per-game results, and the record includes every qualifying matchup—not only Model Zero wins.

View all years
One scoring target
Every board, Model Zero, and the actual finish are ranked by per-game nine-category value.
Important players count more
Weight declines smoothly from 2× for the best actual finisher in each shared pool to 1× for the last.
Same player pool
Each matchup is scored only on shared players who appeared in at least 20 games that season.

* First-year players are excluded from both sides of every matchup.