Set 2, Game 3: Eala changes rally tempo, saving 2 break points with aggressive inside-out forehands to establish 3-1 lead, swinging momentum heading toward the decider.
Set 2 1st-Serve Win %
76%
Set 2 Break Pts Saved
4 of 5
S2 Winners (Eala)
14
Jovic UEs (Set 2 Shift)
18
Primary Decider Leverage
Aggressive Return
Match Breakdown & Deciding Scenario LedgerAudited Splits
Set / Frame
Status Type
Score / Margin
1st Serve Win %
BP Conv / Saved
Rally Length (>6)
Momentum Driver
Set 1
CONFIRMED FACT
Jovic 6-3
64% (Jovic)
3/4 Converted
42%
Jovic deep kick-serve & heavy top-spin
Set 2
CONFIRMED FACT
Eala 6-3
76% (Eala)
4/5 Saved
56%
Eala inside-out forehand attack & depth
Set 3 (Decider)
MODELED PROJECTION
Eala 6-4
72%
68% Saved
48%
Baseline Aggression (+18%)
Saved decider coefficient arithmetic
Read the explanation
Source begins with fifty-percent edge. Saved serve input seventy-two minus baseline sixty-five gives seven, times coefficient point four five adds three point one five. Bars compare baseline and intermediate fifty-three point one five at eight pixels per percentage point. Assigned model probabilities and saved match facts are not independently validated. Saved aggression eighteen times coefficient point two eight contributes five point zero four. Add to intermediate fifty-three point one five to reach fifty-eight point one nine before other terms. Bars compare intermediates at eight pixels per percentage point. This heuristic does not prove aggression causally improves actual match performance. Saved rally term adds point six, and opponent-pressure term subtracts one point six. Net minus one moves intermediate fifty-eight point one nine to fifty-seven point one nine before output rounding, within source clamps. Bars compare these values at eight pixels per percentage point. Full raw native model JSON/reset and native responsive first-footer checks preserved; match truth and prediction calibration unproved.