Rosternomics

1991 MLB Draft

Every MLB amateur draft, 1985–2025. Real = career WAR so far. Expected = the model's full-career projection for this slot × position × age-at-draft (HS vs college), ± a predictive band. DVOS = Real minus what the same cohort has produced by the same number of years post-draft (apples-to-apples — so we don't punish a 2024 pick for not yet being a 10-yr veteran). Example: Bazzana (2024 #1, college IF) has Real 0.8; the typical #1 college IF has produced 1.6 WAR by year 2 post-draft, so his DVOS = −0.8 (0.8 behind cohort); long-run Expected ≈ 40 WAR. Model: empirical-Bayes hierarchical, fit on 41 years of realized outcomes. SD leaderboard →

Round:
Year Pick Rd Team Player Pos Age B/T Ht Wt Bonus Bonus% Scouting Dir GM / POBO Real DVOS Expected
1991 1 1.1 NYY Brien Taylor P P 20 COLL L/L 6'3" 220 Bill Livesey Gene Michael 0.0 -19.6 19.6±3.8
1991 2 1.2 ATL Mike Kelly OF LF 21 COLL R/R 6'4" 195 John Schuerholz 0.7 -22.8 23.5±4.2
1991 3 1.3 MIN Dave McCarty IF 1B 22 COLL R/L 6'5" 210 Terry Ryan Andy MacPhail -2.5 -24.1 21.5±3.9
1991 4 1.4 STL Dmitri Young IF 1B 18 HS S/R 6'2" 295 Dal Maxvill 11.8 -7.1 18.9±3.9
1991 5 1.5 MIL Ken Henderson P P 18 HS R/R 6'5" 190 Dick Foster Harry Dalton 0.0 -9.9 9.9±3.9
1991 6 1.6 HOU John Burke P P 21 COLL S/R 6'4" 220 Dan O'Brien Jr Bill Wood 0.0 -7.9 7.9±3.8
1991 7 1.7 KCR Joe Vitiello IF 1B 21 COLL R/R 6'3" 220 Art Stewart Herk Robinson -0.8 -15.6 14.8±4.0
1991 8 1.8 SDP Joey Hamilton P P 21 COLL R/R 6'4" 240 Randy Smith Joe McIlvaine 16.3 +8.4 7.9±3.7
1991 9 1.9 BAL Mark Smith OF LF 21 COLL R/R 6'3" 235 Gary Nickels Roland Hemond 0.7 -7.8 8.5±3.9
1991 10 1.10 PHI Tyler Green P P 21 COLL R/R 6'5" 185 Jay Hankins Lee Thomas 2.0 -5.9 7.9±3.7
1991 11 1.11 SEA Shawn Estes P P 18 HS R/L 6'1" 210 Woody Woodward 18.6 +13.1 5.5±3.7
1991 12 1.12 CHC Doug Glanville OF CF 21 COLL R/R 6'2" 175 Jim Frey 10.0 +5.6 4.4±3.8
1991 13 1.13 CLE Manny Ramirez OF LF 19 HS R/R 6'0" 225 Hank Peters 56.0 +48.0 8.0±3.7
1991 14 1.14 WSN Cliff Floyd OF LF 19 HS L/R 6'4" 230 Gary Hughes Dave Dombrowski 23.6 +15.6 8.0±3.7
1991 15 1.15 MIL Tyrone Hill P P 19 HS L/L 6'6" 215 Dick Foster Harry Dalton 0.0 -5.5 5.5±3.7
1991 16 1.16 TOR Shawn Green OF RF 19 HS L/L 6'4" 190 Pat Gillick 29.9 +21.9 8.0±3.7
1991 17 1.17 ANA Eduardo Perez IF 1B 22 COLL R/R 6'4" 240 Bob Fontaine Jr Mike Port 1.6 -9.2 10.8±3.7
1991 18 1.18 NYM Al Shirley OF OF 18 HS R/R 6'2" 210 Roland Johnson Frank Cashen 0.0 -8.0 8.0±3.7
1991 19 1.19 TEX Benji Gil IF SS 19 HS R/R 6'2" 210 Sandy Johnson Tom Grieve 2.1 -0.6 2.8±3.7
1991 20 1.20 CIN Pokey Reese IF 2B 18 HS R/R 5'11" 190 Julian Mock Bob Quinn 6.6 +3.9 2.8±3.7
1991 21 1.21 STL Allen Watson P P 21 COLL L/L 6'1" 212 Dal Maxvill 5.2 -0.3 5.5±3.7
1991 22 1.22 STL Brian Barber P P 18 HS R/R 6'1" 170 Dal Maxvill -0.0 -5.5 5.5±3.7
1991 23 1.23 BOS Aaron Sele P P 21 COLL R/R 6'3" 220 Eddie Kasko Lou Gorman 29.2 +23.8 5.5±3.7
1991 24 1.24 PIT Jonathan Farrell OF OF 20 COLL R/R 6'2" 185 Jack Zduriencik Larry Doughty 0.0 -4.4 4.4±3.7
1991 25 1.25 CHW Scott Ruffcorn P P 22 COLL R/R 6'4" 210 Ron Schueler -1.0 -6.4 5.5±3.7
1991 26 1.26 OAK Brent Gates IF 2B 21 COLL S/R 6'1" 180 Dick Bogard Jr Sandy Alderson 3.4 -0.4 3.8±3.7

Model: empirical-Bayes hierarchical — smooth monotone slot curve × position group (P/IF/OF/C/DH) × age bin (HS <20 vs COLL 20+) × maturity. The slot curve is forced monotone-decreasing within each position+age cell, so pick N can't have lower expected than pick N+1 by construction. Predictive ± band = √(σ²·(1 + 1/n_cell)) — pooled within-cell residual SD plus the cell-mean uncertainty for a new pick at that cell. Fallback chain when a (slot, pos, age) cell is too sparse: → (slot, pos) → (slot). Position × age effects: pitchers debut earlier (high realized at 2 yrs, lower career due to attrition); college catchers ~2× HS catchers career WAR at top slots; HS slot-1 IF/OF carry premium upside (Harper/Correa/A-Rod pattern). Coverage: position 100% (1985–2025), bio (born/ht/wt) ~97%, age 97%, signing bonus only ~11% (sparse pre-2008 — MLB Stats API hasn't backfilled). SD attribution follows SABR (single unambiguous name per team-year), GM uses our hire/POBO records. Team/GM development effects intentionally NOT in the model — that signal belongs on the DVOS-attribution side (per-SD on /sds, per-GM in this table), or it would collapse to zero by construction.