Where does our defense leak outs, and what fixes it?
Reading the numbers. Savant OAA as played is the team as it took the field in 2026: every fielder’s outs above average, credited to the clubs he played for; it answers who. Outs above expected is Keystone’s team model; it counts positioning and maps where balls fell in. Across the 30 clubs the two agree at r = .64–.72 (2023–26); leave out infield liners and pop-ups and it’s .60–.71. That part alone agrees at .04–.40 and repeats from one season to the next at −.30–.20 (grounders and the rest: .38–.59). Fielding runs (Savant) add arms and double plays; 2024–26 is runs per full season (1,458 innings), weighted 5/4/3 and regressed. 2027 ratings are the depth chart’s, per full season against an average fielder at the spot; about 10 runs is a win.
The call
Second and third base: +3.8 runs a season, like for like, a toss-up.
Second base: Brandon Lowe (+2.1 projected there) leaves as a free agent; Nick Gonzales projects −0.4, −2.5 (a toss-up). Third base: Nick Gonzales (−7.7 projected there) moves to 2B; Isaac Paredes projects −1.4, +6.3. Brandon Lowe’s +12.4 in 2026 ran far above his +1.9 a season over 2024–26: price his glove on the three seasons, as the rating does. Keeping him at second with Paredes at third fields +2.5 runs more across first, second and third (+2.7 vs. +0.3), about 0.2 wins, a toss-up on the glove; most of what Lowe adds is the bat (QO brief, the Lowe chain). Against the alignment the trade’s wins are measured from, the in-house plan with Jared Triolo at third (+3.6), Isaac Paredes’s glove is −5.0 runs a season: the trade buys a bat, not a glove (third-base brief). The +6.3 above is like for like against Nick Gonzales, 2026’s regular at third.
Catching is the biggest 2027 hole: −5.1 runs by playing time, 6.3 runs below the average club’s catching.
Henry Davis, the 2026 regular (51% of balls in play) and better than any 2027 catcher (+3.7 a full season), is non-tendered (non-tender brief). 2027: Rafael Flores Jr. −5.1 in 266 PA, Endy Rodríguez −7.6 in 268 PA, Derek Berg 0.0 in 101 PA, Christian Bethancourt −2.2 in 36 PA, replacement level, counted as average, 7 PA (over the 671 PA of named catchers alone, as Catching weighs it, −5.2). Like for like the swap is −8.8 a full season; with Henry Davis in Endy Rodríguez’s 268 PA at catcher, the 2027 catching would be about −0.7 (+4.5 runs). See Catching.
Left field projects −3.4 runs in 2027, 3.7 runs below the average club’s left field.
Bryan Reynolds rates −3.9 a full season there, the same regular as 2026. In-house, Jake Mangum rates +2.4 (3,228 batters there in 2024–26, mostly his CF rating carried over: 36% from his own record there): as a late-inning replacement, about two innings a game, that’s +1.4 runs a season.
Shortstop projects −2.1 runs in 2027, 3.2 runs below the average club’s shortstop.
Konnor Griffin rates −3.2 a full season there, the same regular as 2026. He is 21 next summer, and the ratings have no aging term: in the forward test, fielders 25 and under fielded +1.9 runs a season above their rating the next year (±0.7, 137 player-seasons), so his likely understates him and the gap may be smaller than it reads. In-house, Jared Triolo rates −0.7 (4,195 batters there in 2024–26): as a late-inning replacement, about two innings a game, that’s +0.6 runs a season, and the gap between the two gloves (+2.6) is a toss-up.
The left side leaked in 2026, on both measures.
Third and short together were −16.8 runs against the average club on Savant (as played) and −17.1 on the zone model. By spot, Savant vs. the zone: 3B −5.9 vs. −21.1; SS −10.9 vs. +4.0. At third the gap is wider than usual: across the 30 clubs in 2026 the zone figure tracked Savant’s OAA at r = .54 but spread 1.9 times as wide (SD 13 vs. 7 outs), and Pittsburgh’s −28 outs there sit 21 below what its Savant −8 predicts. Savant credits each play from where the fielder stood; the zone counts every ball hit to the area, so read the side as one hole. At 3B, Isaac Paredes (−1.4) replaces Nick Gonzales (−7.7), +6.3 like for like. At SS, Konnor Griffin returns at −3.2 a full season: still below average.
Center field fell short in 2026, on both measures.
Savant as played: −5.4 runs against the average club (23rd of 30); the zone model: −7.4. Oneil Cruz’s arm (+3.1) made up part of it: with arms and double plays, today’s group graded +2.9 there (average team +5.8). 2027: Oneil Cruz returns at +3.2 a full season; with everyone who plays there, center field projects +3.3, about the average club’s (+0.1).
Right field: today’s group graded −11.4, mostly from players who play elsewhere in 2027.
Ryan O'Hearn −3.6 (moves to 1B), Bryan Reynolds −2.2 (moves to LF), Jhostynxon Garcia −1.0 (is traded). −5.3 of the group’s runs came on throws, −3.9 of them Esmerlyn Valdez’s. As the 2026 team played, the range was −3.3 runs against the average club on Savant (20th of 30) and +2.6 on the zone model. Esmerlyn Valdez projects −1.6 a full season: the projection expects his throwing to regress toward average, so arm accuracy and decisions are the development lever; with everyone who plays there, +0.8.
Isaac Paredes at 3B: a bat-first fix, and the two measures disagree on him.
Savant OAA there: 0 (2023), −2 (2024), −4 (2025), −3 (2026). Keystone’s zone model: +8.1 outs on 333 balls (2023), +4.1 outs on 349 balls (2024), −4.2 outs on 261 balls (2025), −20.7 outs on 326 balls (2026). In 2026 the zone figure at 3B tracked Savant loosely across fielders (r = .35 over 28 with 150+ balls, against .66–.70 in 2024–25); his −20.7 sits 20.6 outs below what his Savant −3 predicts. Arm at 3B: 85.0 → 81.7 → 79.5 → 78.8 mph (3B median 83.3, 10th percentile 79.8). The depth chart’s rating, which leans on Savant, is −1.4 a full season. If the zone model is right about him (−2.9 outs per 100 balls over 2024–26, not regressed), he would be about −9.5 runs on a full season of 3B balls, 0.8 wins below the rating: a downside, not the forecast. HOU’s alignment is in the zone figure; pro scouting and the positioning staff can separate it from his range before the deal.
In-house gloves the plan doesn’t start.
At LF, Jake Mangum rates +2.4 (3,228 batters there in 2024–26, mostly his CF rating carried over: 36% from his own record there) vs. Bryan Reynolds’s −3.9: +1.4 runs a season as a late-inning replacement. At 3B, Jared Triolo rates +3.6 (4,532 batters there in 2024–26, mostly his SS rating carried over: 40% from his own record there) vs. Isaac Paredes’s −1.4: +1.1 runs a season as a late-inning replacement. At RF, Billy Cook rates +5.7 (345 batters there in 2024–26, mostly his CF rating carried over: 6% from his own record there) vs. Esmerlyn Valdez’s −1.6: +1.6 runs a season as a late-inning replacement, but he opens in the minors (1 option left, estimated). The largest with at least a quarter of the rating from his own record there is about 1.4 runs a season (0.1 wins).
Keep the running game.
Pittsburgh gained 7.4 runs on the bases (6th, today’s roster). The plan’s moves: in, Isaac Paredes (−1.9); out, Brandon Lowe (−0.2) and Henry Davis (−1.8) (2026 baserunning runs, actual): +0.1 net, counting whoever takes a departing regular’s time as an average runner. The regressed 2027 projection over the plan’s playing time is on Running Game.
By position vs. the average team
Today’s roster: each player’s 2026 fielding runs (with any club), credited to the positions where he earned them; range from where he stood, plus arm and double plays. ● Pittsburgh · | MLB average team.
Shaded band = range across all 30 teams. Gap = Pittsburgh minus the average team; about 10 runs is one win.
2027 defense by position
The Winter Plan: each spot’s 2026 regular and 2027 starter on the same projection (runs a full season vs. an average fielder there), and the spot’s total by playing time
| Pos | 2026 Savant | 2026 zone | 2026 regular | 2027 starter | Like for like | 2027 at the spot | Also plays there |
|---|---|---|---|---|---|---|---|
| C | — | — | Henry Davis +3.7 | Rafael Flores Jr. −5.1 | −8.8 | −5.1 | Endy Rodríguez 268 PA · Derek Berg 101 PA · Christian Bethancourt 36 PA · replacement 7 PA |
| 1B | −2.9 21st | +8.2 | Spencer Horwitz −0.1 | Ryan O'Hearn +2.0 | +2.1 | +1.3 | Endy Rodríguez 21 PA · replacement 183 PA |
| 2B | +8.7 3rd | +10.6 | Brandon Lowe +2.1 | Nick Gonzales −0.4 | −2.5 | −0.3 | Termarr Johnson 101 PA · Jack Brannigan 42 PA |
| 3B | −5.9 26th | −21.1 | Nick Gonzales −7.7 | Isaac Paredes −1.4 | +6.3 | −0.6 | Jared Triolo 106 PA · Nick Yorke 10 PA |
| SS | −10.9 26th | +4.0 | same | Konnor Griffin −3.2 | — | −2.1 | Jared Triolo 309 PA |
| LF | +1.0 12th | −0.5 | same | Bryan Reynolds −3.9 | — | −3.4 | replacement 76 PA |
| CF | −5.4 23rd | −7.4 | same | Oneil Cruz +3.2 | — | +3.3 | Jake Mangum 220 PA |
| RF | −3.3 20th | +2.6 | same | Esmerlyn Valdez −1.6 | — | +0.8 | Jake Mangum 256 PA · Ronny Simon 78 PA |
| All | −18.7 | −3.5 | Starters −10.4 | −2.9 | −6.1 | 30th of 30 clubs · average club +8.9 |
2026 Savant: range runs as the team played, against the average club (no catchers). 2026 zone: Keystone’s runs on balls hit there, positioning in, against the average club. 2027 figures are against an average fielder at each spot; clubs start their better gloves, so the average club sits above zero (+8.9 in all; hover a spot for its average), and red marks a spot below the average club’s. The 2027 ratings regress every record the same way, so the change is like for like; 2026 totals aren’t (big seasons shrink toward zero either way). Starters play 67% of a spot’s time on average here; the rest goes to the bench, call-ups and replacement level (counted as average). A season lands within about ±8 runs of a rating two times in three, so a gap under 5.0 runs between two players is a toss-up (gray), 7.1 for two spots changing at once.
| Alignment | 1B · 2B · 3B | Infield runs | Wins | Payroll |
|---|---|---|---|---|
| 2026’s regulars, projected | Spencer Horwitz · Brandon Lowe · Nick Gonzales | −5.7 | — | — |
| Lowe leaves; third base in-house | Ryan O'Hearn · Nick Gonzales · Jared Triolo | +5.2 | 79.1 | $126.0M |
| The plan: Lowe leaves; Isaac Paredes at third, Gonzales to second | Ryan O'Hearn · Nick Gonzales · Isaac Paredes | +0.3 | 81.7 | $138.6M |
| Lowe accepts the qualifying offer; Isaac Paredes too | Ryan O'Hearn · Brandon Lowe · Isaac Paredes | +2.7 | 83.5 | $160.9M |
| Lowe accepts; no acquisition | Ryan O'Hearn · Nick Gonzales · Brandon Lowe (little or no MLB time there) | −1.3 | 80.8 | $148.4M |
| Lowe re-signed at his market price (3 years × $30.8M); Isaac Paredes too | Ryan O'Hearn · Brandon Lowe · Isaac Paredes | +2.7 | 83.5 | $168.5M |
| Lowe at first base, Ryan O'Hearn to DH, Spencer Horwitz to the bench, Gonzales at second | Brandon Lowe · Nick Gonzales · Isaac Paredes | +2.8 | 82.9 | $160.9M |
| Keep Lowe, trade Gonzales | Ryan O'Hearn · Brandon Lowe · Isaac Paredes | +2.7 | 82.7 | $158.6M |
Wins and payroll from the Winter Plan’s chain (decisions feed; current rules). Infield runs are the three starters’ 2027 ratings at those spots, a full season each. A rating with little or no MLB time there (under 300 batters in 2024–26) is mostly the spectrum prior.
Where balls fall in
The team as it played that season: outs above expected on every ball in play, by where it was hit, whoever made the play, positioning included (Keystone expected-outs model). Warm = outs gained; cool = outs leaked; runs at each zone’s value.
−3.5 runs
−2 outs vs. expected in 2026, 16th of 30: −6 on grounders and flies (17th), +4 on infield liners and pop-ups (7th). The 40 balls with no zone (−0.2 runs) count in these outs totals but not on the map or in its runs total; with them the defense is −3.7 runs.
Savant, as played: −21 OAA (24th).
Biggest leak: 3B, −21.1 runs (30th of 30).
Biggest strength: 2B, +10.6 runs (5th).
Labels: runs, then rank among 30 defenses. Expected outs use exit velocity, launch angle, direction, distance and batter speed, scored out of sample; parks are measured on visiting teams’ balls only. See the model card.
Who the third-base leak cost
Pittsburgh pitchers, 2026: runs lost on balls hit to the third-base zone behind them, ±1 standard error
| Pitcher | GB% | Balls to 3B zone | Runs lost at 3B | All zones | 2027 at 3B |
|---|---|---|---|---|---|
| Paul Skenes | 43% | 49 | −5.1±1.9 | −6.1 | +0.6±1.8 |
| Bubba Chandler | 40% | 49 | −4.2±2.0 | −6.6 | +0.6±1.8 |
| Jared Jones | 38% | 34 | −4.0±1.7 | +0.5 | +0.6±1.8 |
| Carmen Mlodzinski | 43% | 36 | −1.5±1.7 | +1.7 | +0.3±1.4 |
| Yohan Ramírez | 50% | 27 | −0.8±1.6 | +3.6 | +0.3±1.3 |
A few dozen balls each, so only the biggest are clearly more than noise. 2027 at 3B: his planned innings × his 2026 balls to the zone per inning, each worth the like-for-like change at third (Isaac Paredes for Nick Gonzales, +6.3 runs over a full season of 427 balls there); ± is the noise on his own balls. For every arm listed it is under a run a season. Ground-ball arms gain most from a better left side; whether the Winter Plan’s third baseman delivers it is weighed above and in the third-base brief.
Positioning
Infield alignment on ground balls, 2026, by how often the hitter pulls his grounders
| Hitter type | PIT shade | MLB shade | Shade vs. standard |
|---|---|---|---|
| Heavy pull (65%+) | 27% | 32% | +0.5 ±0.3 outs / 100 GB |
| Pull (55–65%) | 6% | 15% | +0.6 ±0.5 outs / 100 GB |
| Balanced / opposite | 1% | 5% | too few shaded (442) |
League 2024–26. Every gain is within two standard errors of zero, and shading is chosen (it goes to the hitters it suits), so these are associations, not effects. Pittsburgh’s infield shades pull hitters less than the league does; matching the league’s rate would be worth about 0.3 runs a season: positioning isn’t the lever. “Strategic” alignments convert far fewer (mostly the infield drawn in with a runner on third, a deliberate trade), so they’re left out.
Pittsburgh fielders
Sorted worst to best on 2026. Range, arm and double-play runs make up fielding runs; 2024–26 is per full season, regressed.
| Player | Pos | Fielding runs 2026 | Range | Arm | 2024–26 | + position | OAA | Prior OAA | Sprint | Arm mph | Baserunning |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Nick Gonzales | 3B | −9.7 | −8.0 | +0.1 | −3.1 | −0.3 | −11 | +1 2B ’23+2 2B ’24+4 2B ’25 | 28.8 | 82.7 | +1.1 |
| Bryan Reynolds | LF | −4.3 | −3.0 | −1.2 | −5.4 | −12.9 | −3 | −4 LF ’23−13 LF ’24−2 RF ’25 | 28.2 | 88.9 | +1.9 |
| Konnor Griffin | SS | −4.2 | −5.1 | −0.2 | −2.9 | +4.6 | −7 | — | 29.9 | 88.1 | +5.2 |
| Spencer Horwitz | 1B | −3.5 | −3.5 | — | +0.1 | −10.9 | −5 | +1 1B ’23+2 2B ’24+1 1B ’25 | 24.7 | 73.6 | −1.9 |
| Esmerlyn Valdez | RF | −2.5 | +1.8 | −4.3 | −1.9 | −9.4 | +2 | — | 28.1 | 86.9 | −0.5 |
| Ryan O'Hearn | 1B | −2.5 | −2.8 | −1.2 | 0.0 | −10.7 | −3 | +4 1B ’23−2 1B ’24+5 1B ’25 | 26.7 | 81.3 | −3.5 |
| Tyler Callihan | 3B | −2.4 | −1.6 | −0.7 | −2.4 | −8.1 | −2 | −1 LF ’25 | 28.0 | — | +0.9 |
| Jhostynxon Garcia | CF | −2.1 | −1.2 | −0.9 | −1.9 | −5.2 | −1 | 0 RF ’25 | 27.7 | — | +0.1 |
| Jake Mangum | CF | −1.3 | −1.1 | −0.3 | +1.0 | −2.4 | −1 | +6 LF ’25 | 29.6 | 87.2 | +1.2 |
| Jacob Gonzalez | SS | −1.0 | −1.6 | — | −0.7 | −2.5 | −2 | — | 26.5 | 83.8 | +0.5 |
| Nick Yorke | 3B | −0.6 | +0.6 | −0.1 | −0.8 | −1.6 | +1 | −2 2B ’240 2B ’25 | 27.7 | — | +0.4 |
| Ronny Simon | RF | −0.4 | +0.2 | −0.6 | −4.0 | −9.3 | 0 | −6 2B ’25 | 28.7 | — | 0.0 |
| Jack Brannigan | 3B | +0.1 | +0.1 | — | +0.1 | +2.6 | 0 | — | — | — | — |
| Kyler Fedko | RF | +0.2 | +0.3 | −0.1 | +0.2 | −1.9 | 0 | — | — | — | −0.5 |
| Jared Triolo | SS | +0.4 | −0.5 | −0.1 | +2.5 | +5.9 | 0 | +5 3B ’23+3 3B ’24+6 SS ’25 | 28.3 | 85.5 | +1.7 |
| Billy Cook | CF | +1.2 | +0.4 | +0.9 | +2.2 | −0.4 | 0 | +2 CF ’240 CF ’25 | 29.3 | 97.8 | +0.5 |
| Oneil Cruz | CF | +1.8 | −1.3 | +3.1 | +2.0 | +5.7 | −1 | 0 SS ’23−1 SS ’240 CF ’25 | 28.5 | 97.7 | +2.6 |
| Brandon Lowe | 2B | +12.4 | +11.2 | — | +1.9 | +3.9 | +15 | +4 2B ’230 2B ’24−14 2B ’25 | 27.0 | 77.7 | −0.2 |
Catching and player gloves
Defensive upgrades
Outside gloves by position, ranked on their 2027 rating there (2,000+ batters at the spot over 2024–26), with availability and contract
| Player | Team | 2027 at 3B | vs. ours | All positions | Availability | Contract |
|---|---|---|---|---|---|---|
| Joey Ortiz | MIL | +12.2† | +13.6 | +9.5 | Unlikely | Arbitration · thru 2029, $23.3M (arb. est.) |
| Ezequiel Duran | TEX | +9.3† | +10.7 | +7.4 | Likely | Arbitration · thru 2028, $16.9M (arb. est.) |
| Maikel Garcia | KC | +7.2 | +8.6 | +6.0 | Unlikely | Signed $7.2M · thru 2031, $71.1M |
| Nick Loftin | KC | +6.4 | +7.8 | −0.3 | Possible | Pre-arb · thru 2030, $9.2M (arb. est.) |
| Daniel Schneemann | CLE | +5.7† | +7.1 | +5.6 | Possible | Pre-arb · thru 2030, $8.6M (arb. est.) |
| Matt Chapman | SF | +5.6 | +7.0 | +5.3 | Possible | Signed $25.2M · thru 2030, $100.7M |
| Alex Bregman | CHC | +5.0 | +6.4 | +4.8 | Unlikely | Signed $35.0M · thru 2030, $140.0M |
| Jazz Chisholm Jr. | NYY | +4.6† | +6.0 | +4.4 | Free agent | Free agent |
| Ernie Clement | TOR | +4.5† | +5.9 | +3.2 | Possible | Arbitration · thru 2028, $11.9M (arb. est.) |
Source: Baseball Savant Fielding Run Value (2023–26), OAA, catcher, arm strength, sprint speed & baserunning leaderboards; Keystone expected outs (Statcast balls in play, 2021–26) and the depth chart’s 2027 ratings. Fielding runs credited to each player’s current team; Savant as played splits each fielder’s OAA among his 2026 clubs by balls in play.