What does a +27 change actually tell us?
A swing compares two scoring margins. It does not measure a win probability or prove that a blowout causes the next game’s rebound. Use the sliders to explore endpoint arithmetic, then open the explanation for the chart’s assumptions.
The 47-entry histogram, stored median/win rate, and ten profiles are unverified demonstration fixtures. No historical sample, statistical calibration, or predictive model is supplied. The JSON retains these original fixture fields for transparency.
The default moves from −22 to +5: 5 − (−22) = +27. The intermediate −8.5 is interpolation, not an observed inning. The combined margins at these two endpoints sum to −17, which is a different question from their change.
MLB records Seattle’s 22–0 loss on August 18, 2026. MLB: August 18, 2026 game
On August 19, Seattle led 4–0 after one inning and 7–2 after two, then won 7–5. MLB: August 19, 2026 inning lines
The early +5 endpoint matches that box score; the middle −8.5 point does not. The scrubber selects three model positions. Changing either slider creates a counterfactual rather than replaying the actual game.
The histogram’s counts total 47. Its cumulative count reaches 21 at two runs and 26 at three, so the 24th observation gives a median of three. The retained metadata says four; without the underlying records, neither that metadata nor the 31.9% win rate should be treated as verified MLB evidence.
The profile comparison counts ten authored endpoint swings at or below your selection. Their swings are 21, 22, 23, or 25, so +27 displays 100%. This is a share of toy profiles, not a historical percentile or chance of winning.
Thirteen histogram bars cycle through ten illustrative profiles. Overlay positions are normalized to the chart width; they do not identify a game in that bin or align observed innings. Download JSON to inspect all retained points and PNG for the displayed chart.
Seattle lost 22–0.
Seattle led 4–0 after one inning and 7–2 after two; final 7–5.