Sparrow Data

Graded prediction archive

Tonight's board — 2026-08-20

Board generated 4:30 PM ET on 2026-08-20 for the 2026-08-20 slate, from the mixed — confirmed, morning pass. 22 of 29 picks are on confirmed lineups. These are the same picks posted to Discord before first pitch; they are graded tomorrow morning and roll into the record below. No pick on this page is selected differently from the ones in the archive — it is the same board, published earlier.

ELEVATED

Top 15 by delta (tonight's probability minus the bat's own baseline). The headline board.

Batter	Team	Lineup	Tier	Prob	Base	Lift	Why
Bryce Eldridge	SF	confirmed	T2 · Power Bat	0.188	0.149	1.26x	+career barrel rate, +career hard-hit rate
Alec Burleson	STL	confirmed	T2 · Power Bat	0.183	0.147	1.24x	+park factor, +recent hard-hit rate
Elly De La Cruz	CIN	confirmed	T2 · Power Bat	0.183	0.150	1.22x	+park factor, +SP strikeout rate
Austin Riley	ATL	confirmed	T3 · Solid	0.160	0.107	1.49x	+career barrel rate, +recent hard-hit rate
Andrés Chaparro	WSH	projected	T3 · Solid	0.160	0.124	1.29x	+recent hard-hit rate, −SP strikeout rate
Jonathan Aranda	TB	confirmed	T3 · Solid	0.160	0.129	1.24x	+career hard-hit rate, +SP strikeout rate
JJ Wetherholt	STL	confirmed	T3 · Solid	0.160	0.118	1.35x	+park factor, +season fly-ball rate
Jimmy Crooks	STL	confirmed	T3 · Solid	0.153	0.090	1.69x	+park factor, +season hard-hit rate
William Contreras	MIL	projected	T3 · Solid	0.143	0.108	1.33x	+park factor, +career hard-hit rate
José Tena	WSH	projected	T3 · Solid	0.140	0.094	1.49x	+season hard-hit rate, −SP strikeout rate
Dylan Beavers	BAL	confirmed	T3 · Solid	0.140	0.100	1.40x	+park factor, +SP fly balls allowed

SLEEPERS

Top 5 by lift among bats with a LOW baseline — the ones a power list would not surface.

Batter	Team	Lineup	Tier	Prob	Base	Lift	Why
Michael Toglia	CIN	confirmed	T4 · The Field	0.117	0.086	1.36x	+park factor, +recent hard-hit rate
Blaze Jordan	STL	confirmed	T4 · The Field	0.105	0.068	1.54x	+park factor, −recent fly-ball rate
Andrew Vaughn	MIL	projected	T5 · Longshot	0.092	0.071	1.30x	+park factor, −pull tendency
Sean Murphy	ATL	confirmed	T5 · Longshot	0.086	0.065	1.32x	+career barrel rate, −recent hard-hit rate

AROUND SLATE

One bat per game, the best by delta. Broad coverage rather than concentration.

Batter	Team	Lineup	Tier	Prob	Base	Lift	Why
Bryce Eldridge	SF	confirmed	T2 · Power Bat	0.188	0.149	1.26x	+career barrel rate, +career hard-hit rate
Austin Riley	ATL	confirmed	T3 · Solid	0.160	0.107	1.49x	+career barrel rate, +recent hard-hit rate
Jonathan Aranda	TB	confirmed	T3 · Solid	0.160	0.129	1.24x	+career hard-hit rate, +SP strikeout rate
Jimmy Crooks	STL	confirmed	T3 · Solid	0.153	0.090	1.69x	+park factor, +season hard-hit rate
William Contreras	MIL	projected	T3 · Solid	0.143	0.108	1.33x	+park factor, +career hard-hit rate
Cam Smith	HOU	projected	T3 · Solid	0.143	0.124	1.15x	+recent hard-hit rate, +career hard-hit rate
Lawrence Butler	ATH	confirmed	T3 · Solid	0.140	0.127	1.10x	−park factor, +season hard-hit rate
José Tena	WSH	projected	T3 · Solid	0.140	0.094	1.49x	+season hard-hit rate, −SP strikeout rate
Dylan Beavers	BAL	confirmed	T3 · Solid	0.140	0.100	1.40x	+park factor, +SP fly balls allowed

MOUND SUPPRESS

The three starting pitchers whose opposing lineups project LOWEST. Graded inverted: below 1.0 is the win.

Pitcher	Lineup	Tier	Prob	Base	Lift	Why
Kyle Bradish	confirmed	T1 · Slugger	1.213	1.392	0.87x	strong suppression spot
Ian Seymour	confirmed	T1 · Slugger	0.897	1.043	0.86x	strong suppression spot
Randy Dobnak	confirmed	T1 · Slugger	0.848	1.012	0.84x	strong suppression spot
Gage Jump	confirmed	T1 · Slugger	0.799	0.965	0.83x	strong suppression spot

MOUND VULN

The single starting pitcher whose opposing lineup projects highest.

Pitcher	Team	Lineup	Tier	Prob	Base	Lift	Why
Brady Singer		confirmed	T1 · Slugger	1.259	1.076	1.17x	vulnerable spot — hitters' night

Results appear on this night's page once the games are final and graded.

The record

Every pick this model has published since 2026-07-09, and how each one graded. The boards are posted before first pitch and graded the next morning against what actually happened. Misses are shown exactly as prominently as hits — that is the point of publishing this.

How to read it. BASE is how often this bat homers in a normal game. PROB is tonight, given the pitcher, the park and the weather. The gap between them is the entire product — a high PROB on a slugger is not a finding, a high PROB on a bat whose BASE is low is. Ratios below compare home runs actually hit against what those same bats' baselines expected: 1.00x means the picks did exactly what ordinary bats would have.

Rolling 30 days, by section

ELEVATED

1.31x 95% CI [1.03, 1.57] · n=381 picks

Top 15 by delta (tonight's probability minus the bat's own baseline). The headline board.

SLEEPERS

1.12x 95% CI [0.46, 1.86] · n=139 picks

Top 5 by lift among bats with a LOW baseline — the ones a power list would not surface.

AROUND SLATE

1.20x 95% CI [0.92, 1.48] · n=340 picks

One bat per game, the best by delta. Broad coverage rather than concentration.

FAVORITES

0.83x 95% CI [0.63, 1.02] · n=284 picks

Top 10 by raw probability. Published for transparency; NO edge is claimed here — expect roughly 1.0x baseline.

MOUND SUPPRESS

0.64x 95% CI [0.48, 0.80] · n=97 picks (inverted — below 1.00x is the win)

The three starting pitchers whose opposing lineups project LOWEST. Graded inverted: below 1.0 is the win.

MOUND VULN

1.36x 95% CI [1.06, 1.67] · n=44 picks

The single starting pitcher whose opposing lineup projects highest.

Since 2026-07-09, cumulative

Section	Picks	Home runs	Baseline expected	Realized vs baseline
ELEVATED	503	90	63.6	1.42x 95% CI [1.15, 1.68] · n=503 picks
SLEEPERS	185	19	12.1	1.57x 95% CI [0.94, 2.27] · n=185 picks
AROUND SLATE	462	73	51.8	1.41x 95% CI [1.12, 1.70] · n=462 picks
FAVORITES	388	74	87.1	0.85x 95% CI [0.68, 1.02] · n=388 picks
MOUND SUPPRESS	118	84	128.0	0.66x 95% CI [0.53, 0.79] · n=118 picks
MOUND VULN	51	63	48.3	1.30x 95% CI [1.03, 1.58] · n=51 picks

The calibration curve

Of the bats we said had a ~14% chance, did about 14% homer? This is the claim nobody else publishes, and it is the one that cannot be cherry-picked. Every graded pick on a batter board is here (1152 picks, deduplicated across sections). Buckets with fewer than 50 picks are greyed — they are too small to read.

Predicted range	Picks	Avg predicted	Actually homered
6%–10%	76	8.5%	9.2%
10%–14%	171	11.7%	9.4%
14%–18%	349	15.3%	16.6%
18%–25%	505	21.2%	19.0%
25%–101%	51	26.8%	15.7%

Upper bar = predicted, lower = actual. A well-calibrated model has them close in every row that has enough picks to judge.

Most recent night

Full board for the most recent night: 2026-08-20. Tonight's board (2026-08-20) is published but not yet graded; grading runs the following morning. The most recent fully graded night is 2026-08-19. Every night since 2026-07-09 has its own permanent page in the daily archive.

1957 published picks across 40 nights; 1707 graded so far.

Sparrow Data — every pick published before first pitch and graded the next morning. Archive covers 2026-07-09 through 2026-08-19. Nothing here is ever edited or removed after the fact. Methodology · Raw data