The simulator expects 0.481 bases allowed per batter faced against him — better than 15th percentile of qualified pitchers. The soft spot is Strikeouts (16th percentile).
503 batters faced · data through 2026-08-22.
▸How to read this chart
Each spoke is a skill estimate adjusted for sample size (not raw season stats), so farther out = better. Spokes are the metric itself; ↓ marks a rate where a LOWER number is better. Percentiles vs. 464 qualified pitchers — 100+ batters faced, starters and relievers pooled.
Plays like…
Same archetype, nearest by rate — pick one to see him side by side with Mitch Keller.
Where he sits among qualified pitchers on the two skills that separate styles.
▸How to read this chart
Two skills separate pitching styles: finishing hitters without free passes (strikeout rate minus walk rate) and keeping the contact they do allow soft (hard-hit rate against). Every dot is a qualified pitcher this season; dot size tracks batters faced. Click any dot to open that pitcher's page.
Mitch Keller
League median
Percentile
Bases per ball in play(lower = better)
.551
.528
15th
Walk rate(lower = better)
9%
9%
60th
Strikeout rate(higher = better)
19%
22%
16th
100th percentile = best in MLB.
Arsenal
What he throws, and what it does
Mitch Keller mixes his four-seam fastball (32%) and sinker (17%). His best pitch has been the four-seam fastball, saving 1.2 bases per 100 thrown vs average.
Pitch mix & swing-and-miss
▲ +37 bases better than a replacement arm this season
allowed 243 bases vs a replacement arm's 280 over the same batters faced
Replacement level is per-batter and can't be split by pitch type, so the bars below use league average.
Bases vs average
All pitches
22.2 bases cost vs league average, across 1,887 pitches — 1.2 per 100.
Lg avg
Four-Seam Fastball
32%
93.3 mph · RHP median 95.1
17% whiffs · Lg median 21%
saved 1.2 per 1007.3 total
Sinker
17%
92.3 mph · RHP median 94.7
15% whiffs · Lg median 13%
cost 0.3 per 1001.1 total
Sweeper
16%
82.0 mph · RHP median 83.5
24% whiffs · Lg median 30%
cost 4.0 per 10012.0 total
Slider
10%
87.2 mph · RHP median 86.9
22% whiffs · Lg median 33%
cost 1.5 per 1002.8 total
Curveball
9%
77.8 mph · RHP median 81.1
37% whiffs · Lg median 29%
cost 3.8 per 1006.4 total
Changeup
9%
88.7 mph · RHP median 87.1
26% whiffs · Lg median 30%
cost 2.6 per 1004.5 total
Cutter
7%
89.8 mph · RHP median 90.1
20% whiffs · Lg median 22%
cost 2.0 per 1002.7 total
▸How to read this chart
What they throw, how hard, and what misses bats — from pitch tracking, regular season only.
Mix by batter side
Four-Seam Fastball
35% vs LHH · RHP median 41%
28% vs RHH · RHP median 34%
Sinker
12% vs LHH · RHP median 17%
26% vs RHH · RHP median 34%
Sweeper
9% vs LHH · RHP median 18%
28% vs RHH · RHP median 29%
Slider
9% vs LHH · RHP median 21%
12% vs RHH · RHP median 31%
▸How to read this chart
How the mix shifts against lefty and righty bats. Medians are among right-handed pitchers who throw that pitch at least 3% of the time — between 120 and 351 of them, depending on the pitch. Overall results by batter side are under Platoon splits.
Mitch Keller's pitch movement, 2026
3 of his 7 pitches have moved 2+ inches from their early-season shapes.
Compared with his own earlier outings this season, not with other pitchers — and after taking out the drift every pitch in the league shows through midsummer.
Pitch
Risein · vs Lg median
Breakin · vs Lg median
Vs. early season
Four-Seam FastballFF
Rise 13.9 in
−1.8 vs Lg median
Break 8.4 in
+0.8 vs Lg median
Vs. early season steady
SinkerSI
Rise 8.9 in
+0.6 vs Lg median
Break 13.7 in
−1.6 vs Lg median
Vs. early season changed
SweeperST
Rise −0.1 in
−1.0 vs Lg median
Break −16.7 in
−2.8 vs Lg median
Vs. early season steady
SliderSL
Rise 1.1 in
−0.7 vs Lg median
Break −4.3 in
−0.6 vs Lg median
Vs. early season steady
CurveballCU
Rise −12.0 in
−1.9 vs Lg median
Break −12.5 in
−3.8 vs Lg median
Vs. early season changed
ChangeupCH
Rise 2.7 in
−1.3 vs Lg median
Break 11.6 in
−2.5 vs Lg median
Vs. early season steady
CutterFC
Rise 7.9 in
+0.4 vs Lg median
Break −2.4 in
−0.3 vs Lg median
Vs. early season changed
Early season → nowLg medianCurrent-dot uncertainty
▸How to read this chart
Each dot is a pitch's current shape — horizontal break (arm side vs. glove side) by induced vertical break (rise vs. drop). Both axes are in inches of break. The trail runs from this season's earliest tracked outings — the faint, thin end — up to today's solid dot; a trail with only a few segments just means fewer tracked outings. The soft halo around each dot is how far that dot could sit off on its own, just from ordinary outing-to-outing wobble. Don't read a long trail as proof on its own that a pitch changed. Every pitch in baseball loses a little break through midsummer and gets it back by September — the same shift two full seasons back agree on — and the trail shows that league-wide seasonal swing alongside whatever this pitcher did himself. The sentence at the top of this card has the league's share taken out, so it can read “has not moved” above a trail that looks long. That is the two marks measuring different things, not a mistake. A pitch counts as changed only once it sits at least two inches from its own early-season shape — a verdict about this pitcher against himself, never a comparison with anyone else. League median: the small hollow ring marks it, on the same two measurements. Where it sits far from a pitch's dot, a line joins the two — and on a wider screen the ring carries that pitch's code as well — so a distant ring is never orphaned; where it sits close, neither is drawn, because the pairing is already obvious. League median is missing entirely for a pitch marked *: too few pitchers throw it for one to be worth setting, so it is plotted without a reference ring. For the exact dates a pitch changed, see the by-appearance chart below.
By appearance
How the stuff has held up outing to outing
One point per outing — how his velocity, movement, and pitch mix have moved across the season, with league context for scale. Each pitch keeps the same color in every chart.
Platoon splits
How he fares against lefties and righties
Platoon splits (EB/PA allowed)
Better vs RHB than vs LHB by 0.042 EB/PA allowed (modeled).
50% probable range: +0.033 to +0.049
vs LHB321 PA0.493
vs LHB: shrunk estimate 0.493 EB/PA, 50% probable range 0.477 to 0.511, based on 321 plate appearances. Raw (unshrunk) rate: 0.539 EB/PA. League median vs LHB: 0.454 EB/PA.
Lg median
Raw vs LHB rate (0.539) is off this scale — small samples like this are exactly why the shrunk estimate (the dot) is the trustworthy number, not the raw one.
vs RHB182 PA0.452
vs RHB: shrunk estimate 0.452 EB/PA, 50% probable range 0.435 to 0.470, based on 182 plate appearances. Raw (unshrunk) rate: 0.422 EB/PA. League median vs RHB: 0.440 EB/PA.
Lg median
Raw vs RHB rate (0.422) is off this scale — small samples like this are exactly why the shrunk estimate (the dot) is the trustworthy number, not the raw one.
0.4250.4490.4730.4970.520
← betterworse →
Below league medianAbove50% probable rangeHis raw rateMiddle half of the league
About the typical split for a righty arm (median 0.040). He allows more than the league median against both sides.
▸How to read this chart
Tap or hover any mark for its number, or tab to a row for all of them.
His raw rate is the unshrunk number against that side — small samples make it noisy, which is why the model pulls his estimate toward a more trustworthy value.
“Middle half of the league” is the 25th–75th percentile of all pitchers against that side.
These are bases allowed, so lower is better. Weighted by how often he faces each side, the two splits average back to his modeled overall EB/PA allowed — not the xEB/PA headline at the top of the page, which folds in strikeouts and walks differently.
Shrunk splits beat raw splits on held-out seasons, most decisively at low batters-faced counts (relievers especially), and converge with the raw numbers as exposure builds.
Contact allowed
What hitters do when they connect
Batters go up the middle off him most often, but the hardest contact goes to the pull side.
Pull and oppo are relative to each batter's stance, so this pools all the hitters he faced.
Batted balls allowed
Where every ball hit off Mitch Keller landed this season.
Dot color is estimated bases (EB) — what the contact was worth, regardless of what the defense did with it. So a dark hollow ring in the outfield is a crushed ball his defense ran down. Tap or hover a dot or a row to link the two, and use the ▶ links to watch a play on Baseball Savant.
Earned vs. actual
Batters have been credited about 8 fewer bases than their contact off Mitch Keller earned — his results have run fortunate so far.
Earned off his contact allowed200
Actually credited to batters192
over 363 balls in play · walks and strikeouts excluded on both sides
▸How to read this chart
Every ball in play is worth the bases its exit velocity and direction usually earn. Comparing those earned bases with the bases batters were actually credited shows whether his results have been lucky, unlucky, or fair.
ERA 5.29 · WHIP 1.37 · BABIP against 0.293 (Lg median 0.286)
Times through order
What happens the third time they see him
Best games
His biggest outings this year
Best games
His best outings this season by bases saved vs a replacement arm.