Settled model recap · 2026-09-22

MLB Game & Prop Picks Recap for September 22, 2026: 3-4, -1.52 units

Review 1 game Pick and 6 pitcher-prop Picks from September 22, 2026 with the exact saved sportsbook prices, model probabilities, lineup or player values, final results, and plain-English postgame review.

Published after the saved Picks settled. The original prices and inputs stay attached.

Record3-4-0

Net units-1.52u

Game Picks1

Player Props6

One-minute recap

The card missed: 1 game Pick and 6 pitcher-prop Picks finished 3-4 for -1.52 units. The useful question is which assumptions failed, not how to explain the result after the fact.

Decision coverage note

Decision coverage was incomplete in the pitcher strikeout lane (14 opportunities expired despite a prior decision opportunity) and the additional pitcher props lane (41 opportunities expired without a decision). This recap is complete for the 7 settled Picks that were actually recorded, but the slate does not meet full-lifecycle acceptance. No missing decision was reconstructed, and these incidents do not count as Picks or change the record above.

Saved Picks and reviews

Open any Pick for the why

MLB Game Picks

1 game Pick, reviewed from the saved lineup, player, team, and moneyline snapshot.

Game Pick

TB moneyline

TB at NYY · TB 0, NYY 2

Lowvig +126lossSee the review ↓

Postgame review

Lost: TB +126

TB did not get there in a TB 0, NYY 2 final. The saved Lowvig +126 price cost 1.00 unit on the game-pick record.

The pregame case started with Lowvig TB +126. We made the side 45.2%; the price needed 44.2%. That left 0.9 percentage points of room over the price's break-even mark. The strongest public push toward TB was Long-run team strength, worth 0.34 probability points in the saved calculation. NYY carried the higher starter rating: Nick Martinez rated 50.5, compared with Carlos Rodon at 52.5. The two confirmed lineups graded about even: TB 51.1, NYY 50.8. The top displayed hitters were Victor Mesa Jr. (56.2) and Ben Rice (56.4). TB had the higher bullpen rating: TB 50.9, NYY 48.5.

The price case was there; the result was not. We will compare this starter, lineup, and bullpen profile with future frozen reads before deciding whether the miss was noise or a repeatable blind spot.

Saved pregame inputs

Model probability45.2%

Price break-even44.2%

Calculated difference+0.9 pts

Selected side

TB

Team form56.9

Lineup form51.1

Starting pitcher50.5

Bullpen available50.9

Opponent

NYY

Team form55.1

Lineup form50.8

Starting pitcher52.5

Bullpen available48.5

Values use the saved 0–100 rolling scale. They are inputs to the selection screen, not direct probability-point weights.

MLB Player Props

6 pitcher-prop Picks, covering strikeouts and reviewed from the saved workload, opponent, and price snapshots.

Pitcher prop

Nick Martinez Over 3.5 strikeouts

TB at NYY · 7 strikeouts · over 3.5

BetMGM -110winSee the review ↓

Postgame review

Won: Nick Martinez Over 3.5 Ks

Nick Martinez finished with 7 strikeouts, cashing the saved over 3.5 and adding +0.91 units to the prop record.

At BetMGM -110, the model gave the over 55.8% against a 52.4% break-even mark, a +3.4-point gap. The saved projection was 4.6 strikeouts across 23.7 batters faced. The pitcher graded 50.5/100 and the opponent strikeout matchup graded 34.6; the confirmed lineup carried a 23.2% strikeout rate.

The actual result finished 2.4 strikeouts above the point projection. The ticket and the point projection told the same story. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability55.8%

Price break-even52.4%

Calculated difference+3.4 pts

Frozen pitcher-prop inputs

Nick Martinez vs. NYY

Projected Ks: 4.6
Actual Ks: 7

Pitcher value50.5

Opponent K matchup34.6

Lineup K rate23.2%

Lineup coverage100%

Projected batters faced23.7

Evidence quality91%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.29 Ks. Robustness: 80%.

Pitcher prop

Dustin May Over 3.5 strikeouts

MIL at PHI · 2 strikeouts · over 3.5

Betonlineag +110lossSee the review ↓

Postgame review

Lost: Dustin May Over 3.5 Ks

Dustin May finished with 2 strikeouts, leaving the saved over 3.5 on the wrong side of the line for a 1.00-unit loss.

At Betonlineag +110, the model gave the over 51.9% against a 47.6% break-even mark, a +4.3-point gap. The saved projection was 4.5 strikeouts across 21.1 batters faced. The pitcher graded 53.0/100 and the opponent strikeout matchup graded 44.0; the confirmed lineup carried a 20.1% strikeout rate.

The actual result finished 2.5 strikeouts below the point projection. The projection supported the ticket, but the actual result crossed to the other side of the sportsbook line. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability51.9%

Price break-even47.6%

Calculated difference+4.3 pts

Frozen pitcher-prop inputs

Dustin May vs. PHI

Projected Ks: 4.5
Actual Ks: 2

Pitcher value53.0

Opponent K matchup44.0

Lineup K rate20.1%

Lineup coverage100%

Projected batters faced21.1

Evidence quality92%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.26 Ks. Robustness: 78%.

Pitcher prop

Jared Jones Under 5.5 strikeouts

STL at PIT · 9 strikeouts · under 5.5

Bovada +105lossSee the review ↓

Postgame review

Lost: Jared Jones Under 5.5 Ks

Jared Jones finished with 9 strikeouts, leaving the saved under 5.5 on the wrong side of the line for a 1.00-unit loss.

At Bovada +105, the model gave the under 53.1% against a 48.8% break-even mark, a +4.3-point gap. The saved projection was 4.8 strikeouts across 20.4 batters faced. The pitcher graded 57.2/100 and the opponent strikeout matchup graded 55.1; the confirmed lineup carried a 20.6% strikeout rate.

The actual result finished 4.2 strikeouts above the point projection. The projection supported the ticket, but the actual result crossed to the other side of the sportsbook line. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability53.1%

Price break-even48.8%

Calculated difference+4.3 pts

Frozen pitcher-prop inputs

Jared Jones vs. STL

Projected Ks: 4.8
Actual Ks: 9

Pitcher value57.2

Opponent K matchup55.1

Lineup K rate20.6%

Lineup coverage100%

Projected batters faced20.4

Evidence quality89%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.34 Ks. Robustness: 75%.

Pitcher prop

Kyle Freeland Over 2.5 strikeouts

ARI at COL · 2 strikeouts · over 2.5

FanDuel -125lossSee the review ↓

Postgame review

Lost: Kyle Freeland Over 2.5 Ks

Kyle Freeland finished with 2 strikeouts, leaving the saved over 2.5 on the wrong side of the line for a 1.00-unit loss.

At FanDuel -125, the model gave the over 61.7% against a 55.6% break-even mark, a +6.2-point gap. The saved projection was 4.4 strikeouts across 23.6 batters faced. The pitcher graded 43.4/100 and the opponent strikeout matchup graded 31.7; the confirmed lineup carried a 18.5% strikeout rate.

The actual result finished 2.4 strikeouts below the point projection. The projection supported the ticket, but the actual result crossed to the other side of the sportsbook line. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability61.7%

Price break-even55.6%

Calculated difference+6.2 pts

Frozen pitcher-prop inputs

Kyle Freeland vs. ARI

Projected Ks: 4.4
Actual Ks: 2

Pitcher value43.4

Opponent K matchup31.7

Lineup K rate18.5%

Lineup coverage100%

Projected batters faced23.6

Evidence quality91%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.25 Ks. Robustness: 77%.

Pitcher prop

Logan Gilbert Under 6.5 strikeouts

HOU at SEA · 3 strikeouts · under 6.5

Betonlineag -135winSee the review ↓

Postgame review

Won: Logan Gilbert Under 6.5 Ks

Logan Gilbert finished with 3 strikeouts, cashing the saved under 6.5 and adding +0.74 units to the prop record.

At Betonlineag -135, the model gave the under 59.3% against a 57.4% break-even mark, a +1.8-point gap. The saved projection was 5.7 strikeouts across 23.8 batters faced. The pitcher graded 50.1/100 and the opponent strikeout matchup graded 56.6; the confirmed lineup carried a 21.8% strikeout rate.

The actual result finished 2.7 strikeouts below the point projection. The ticket and the point projection told the same story. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability59.3%

Price break-even57.4%

Calculated difference+1.8 pts

Frozen pitcher-prop inputs

Logan Gilbert vs. HOU

Projected Ks: 5.7
Actual Ks: 3

Pitcher value50.1

Opponent K matchup56.6

Lineup K rate21.8%

Lineup coverage100%

Projected batters faced23.8

Evidence quality92%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.58 Ks. Robustness: 80%.

Pitcher prop

Taj Bradley Under 6.5 strikeouts

MIN at SF · 6 strikeouts · under 6.5

Betonlineag -120winSee the review ↓

Postgame review

Won: Taj Bradley Under 6.5 Ks

Taj Bradley finished with 6 strikeouts, cashing the saved under 6.5 and adding +0.83 units to the prop record.

At Betonlineag -120, the model gave the under 55.8% against a 54.5% break-even mark, a +1.3-point gap. The saved projection was 6.1 strikeouts across 24.0 batters faced. The pitcher graded 53.2/100 and the opponent strikeout matchup graded 65.6; the confirmed lineup carried a 24.9% strikeout rate.

The actual result finished 0.1 strikeouts below the point projection. The ticket and the point projection told the same story. We will compare the same workload, opponent, line, and price profile before changing production weights.

Saved pregame inputs

Model probability55.8%

Price break-even54.5%

Calculated difference+1.3 pts

Frozen pitcher-prop inputs

Taj Bradley vs. SF

Projected Ks: 6.1
Actual Ks: 6

Pitcher value53.2

Opponent K matchup65.6

Lineup K rate24.9%

Lineup coverage100%

Projected batters faced24.0

Evidence quality88%

Pitcher and opponent values use the frozen 0–100 rolling scale. Projection standard deviation: 2.70 Ks. Robustness: 78%.

Nick Martinez: What had to hold

Strikeout count: the saved expectation was 2.31351 to 6.90348 strikeouts. The pregame model centered on 4.6 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Nick Martinez: What actually happened

The saved decision settled as a win, with +0.91 units on 1 units of reference risk.

Strikeout count finished at 7 strikeouts, outside the saved 2.31351 to 6.90348 range.

Nick Martinez: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

Dustin May: What had to hold

Strikeout count: the saved expectation was 2.23584 to 6.74836 strikeouts. The pregame model centered on 4.5 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Dustin May: What actually happened

The saved decision settled as a loss, with -1.00 units on 1 units of reference risk.

Strikeout count finished at 2 strikeouts, outside the saved 2.23584 to 6.74836 range.

Dustin May: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

Jared Jones: What had to hold

Strikeout count: the saved expectation was 2.45976 to 7.14468 strikeouts. The pregame model centered on 4.8 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Jared Jones: What actually happened

The saved decision settled as a loss, with -1.00 units on 1 units of reference risk.

Strikeout count finished at 9 strikeouts, outside the saved 2.45976 to 7.14468 range.

Jared Jones: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

Kyle Freeland: What had to hold

Strikeout count: the saved expectation was 2.19437 to 6.69051 strikeouts. The pregame model centered on 4.4 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Kyle Freeland: What actually happened

The saved decision settled as a loss, with -1.00 units on 1 units of reference risk.

Strikeout count finished at 2 strikeouts, outside the saved 2.19437 to 6.69051 range.

Kyle Freeland: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

Logan Gilbert: What had to hold

Strikeout count: the saved expectation was 3.09096 to 8.24756 strikeouts. The pregame model centered on 5.7 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Logan Gilbert: What actually happened

The saved decision settled as a win, with +0.74 units on 1 units of reference risk.

Strikeout count finished at 3 strikeouts, outside the saved 3.09096 to 8.24756 range.

Logan Gilbert: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

Taj Bradley: What had to hold

Strikeout count: the saved expectation was 3.43818 to 8.84056 strikeouts. The pregame model centered on 6.1 strikeouts, with uncertainty around that estimate. The case becomes weaker if: the actual count falls outside the saved descriptive range; this alone does not diagnose the cause.

This range was frozen before the game as a descriptive check, not a calibrated confidence interval.

Taj Bradley: What actually happened

The saved decision settled as a win, with +0.83 units on 1 units of reference risk.

Strikeout count finished at 6 strikeouts, inside the saved 3.43818 to 8.84056 range.

Taj Bradley: What enters model review

These are checks of the original assumptions, not proof of what caused the result. A win does not validate every assumption, and a loss does not invalidate every forecast.

The measured observations can enter a source-verified review dataset. This recap changes no model weights; any proposed change requires separate chronological and prospective evaluation.

What comes next

What we take forward

Each result can raise a useful question, but no single game rewrites the model. Any adjustment has to repeat across comparable frozen reads and survive a prospective test first.

Technical receipt

Fact pack: 64b57d1d2982b294f00f7bbf01519af90ed97f22455952187a2ce4e271467a7d

Decision receipts: mlb-current-decision:2f4b5cb3e0db3ec1b94668c0e01e595a, mlb-k-prop-8141101233ea9b3966430111, mlb-k-prop-d38d054d217d7a5ff2a48461, mlb-k-prop-de6a9420cc661674010b2f8b, mlb-k-prop-48a10ec5a01ba269e479bf68, mlb-k-prop-4a96328fd5ebe2ecad1f4218, mlb-k-prop-8735d33e2e418a531cd7c2f3

Settlement receipts: mlb_live_settlement_b4462c05f06a89a4e355026e, 0659bf803114e6cad2c993e349df6911a4c9f4f199cb7414cea2f48f7b3d3e29, 764780787e28cc759ecd76d8a078d5f733907e37a0b2a7a07252f69e8ad60979, 9a82c32b39b0969fc9e0bc1fa83382f1bda8797b5128bf49ff503a68efdd3c41, 864f5451806a683666450c97be7522b8ac43ce3d86fc1b38e773d1d0be229385, 9bb69e762b90357972139af943bc4b700e985e7508daea49d2bf6d498495a9e1, 53781958fee15de1c19f3dc01d70db74fc8a82306dac02a609f1199db1be550d