Tennis Totals & Handicaps Analysis

P. Carreno-Busta vs D. Shapovalov

Match: P. Carreno-Busta vs D. Shapovalov Tournament: ATP Dubai Date: 2026-02-24 Surface: Hard (all) Analysis Focus: Totals (Over/Under Games) & Game Handicaps

Executive Summary

Model Predictions (Blind, Stats-Only)

• Expected Total Games: 24.8 (95% CI: 21.2 - 28.4)
• Fair Totals Line: 23.5 games
• Expected Game Margin: Shapovalov +2.8 games (95% CI: +0.4 to +5.2)
• Fair Spread Line: Shapovalov -3.5 games

Market Lines

• Totals: 21.5 (Over 1.71 / Under 2.19, no-vig: 56.2% / 43.8%)
• Spread: Shapovalov -3.5 (PCB +3.5 @ 1.81 / Shapovalov -3.5 @ 2.04, no-vig: 53.0% / 47.0%)

Edge Analysis

• Totals: Model P(Over 21.5) = 74% vs Market no-vig 56.2% → +17.8 pp edge on OVER
• Spread: Model P(Shapovalov -3.5) = 48% vs Market no-vig 47.0% → +1.0 pp edge (minimal)

Recommendations

Quality & Form Comparison

Data Quality

• PCB: 65 matches played (last 52 weeks) - HIGH sample size
• Shapovalov: 49 matches played (last 52 weeks) - GOOD sample size
• Data Completeness: HIGH - full stats available for both players

Recent Form

• PCB: 36-29 record (55.4% win rate) - stable form, avg DR 1.32
• Shapovalov: 24-25 record (49.0% win rate) - stable form, avg DR 1.35
• Three-Set Frequency: PCB 46.2% vs Shapovalov 36.7%

Elo Ratings

• PCB: Overall 1200 (Rank #607) - significantly lower rated
• Shapovalov: Overall 1413 (Rank #109) - 213 Elo point advantage
• Hard Court: PCB 1200 vs Shapovalov 1413 - Shapovalov favored on surface

Summary

Shapovalov holds a substantial quality advantage with a 213 Elo point gap and superior ranking (#109 vs #607). However, PCB has slightly better recent form (55.4% vs 49.0%) and plays more three-set matches, indicating competitive battles. Shapovalov’s lower three-set rate suggests he tends to dominate or lose quickly.

Totals/Spread Impact

• Totals: PCB’s higher three-set frequency (46.2%) pushes toward higher totals. Both players average identical 24.2 games/match.
• Spread: Shapovalov’s Elo advantage suggests he should win more games, but PCB’s competitive form limits expected margin. Moderate spread expected.

Hold & Break Comparison

Service Hold Rates

• PCB: 76.1% hold rate - below tour average (~82%)
• Shapovalov: 74.3% hold rate - below tour average (~82%)
• Both players are vulnerable servers - combined breaks per match likely elevated

Return Break Rates

• PCB: 27.4% break rate - solid return capability
• Shapovalov: 27.0% break rate - nearly identical return strength

Break Frequency

• PCB: 3.97 breaks per match average
• Shapovalov: 3.90 breaks per match average
• Expected combined breaks: ~7.9 breaks per match (high)

Summary

Both players are weak holders and strong returners, creating a high-break environment. PCB holds slightly better (76.1% vs 74.3%) while break rates are virtually identical. This profile strongly favors extended matches with frequent service breaks.

Totals/Spread Impact

• Totals: High break frequency (7.9 combined breaks) pushes totals UP significantly. Expect 23-25+ games.
• Spread: Narrow difference in hold/break rates (PCB +1.8% hold, +0.4% break) suggests close game margin despite Elo gap. Expect tight spread.

Pressure Performance

Break Point Conversion

• PCB: 57.0% conversion (254/446) - above tour average (~40%)
• Shapovalov: 57.4% conversion (187/326) - above tour average (~40%)
• Both excel at converting break chances

Break Point Defense

• PCB: 58.8% saved (234/398) - near tour average (~60%)
• Shapovalov: 59.8% saved (183/306) - near tour average (~60%)
• Shapovalov slightly better under pressure on serve

Tiebreak Performance

• PCB: 72.7% TB win rate (8-3) - excellent TB closer
• Shapovalov: 50.0% TB win rate (4-4) - neutral TB performance
• PCB holds significant TB edge (+22.7%)

Key Games Performance

PCB shows superior closing ability - particularly when serving for sets (+8.1%) and matches (+9.9%).

Summary

Both players convert break points exceptionally well (57%+), but PCB demonstrates superior clutch performance in tiebreaks (72.7% vs 50.0%) and critical closing games. Shapovalov’s weaker serve-for-set/match stats indicate vulnerability when ahead.

Totals/Tiebreak Impact

• Totals: PCB’s TB dominance (72.7%) suggests he wins close sets via tiebreaks rather than breaks, adding 2+ games per TB set.
• Tiebreak Probability: Given weak hold rates (76.1%, 74.3%), expect frequent breaks keeping sets competitive → moderate-high TB probability (35-40%).
• Spread: PCB’s closing strength in key games could narrow expected margin despite Elo disadvantage.

Game Distribution Analysis

Set Score Probabilities

Most Likely Set Scores (based on hold/break profiles):

Set Structure Expectations:

• High 7-5 / 7-6 probability due to weak holds creating break exchanges
• Moderate 6-4 probability when one player consolidates better
• Low bagel/breadstick probability given both players break well

Match Structure Probabilities

Straight Sets vs Three Sets:

Using weak hold rates (PCB 76.1%, Shapo 74.3%) and high break frequency (7.9/match):

• P(Straight Sets): 42% - Lower than average due to competitive stats
  • Shapovalov 2-0: 25% (Elo advantage)
  • PCB 2-0: 17% (upset scenario, clutch edge)
• P(Three Sets): 58% - Elevated due to narrow skill gap in hold/break
  • Shapovalov 2-1: 33% (Elo edge prevails in decider)
  • PCB 2-1: 25% (clutch performance + TB strength)

Total Games Distribution

Expected Games per Match Structure:

Total Games Distribution:

• 20-22 games: 22% (decisive straight sets)
• 23-24 games: 26% (competitive straight sets or quick 3-setter)
• 25-27 games: 31% (most likely range - three sets with breaks)
• 28-30 games: 14% (extended three-setter with TBs)
• 31+ games: 7% (multiple TBs in three sets)

Tiebreak Analysis

P(At Least 1 Tiebreak): 41%

Given hold rates of 76.1% (PCB) and 74.3% (Shapovalov):

• P(TB in given set) = 1 - P(breaks decide set) ≈ 18-20% per set
• P(0 TBs in match) ≈ 0.64 × 0.80² (2 sets) + 0.36 × 0.80³ (3 sets) ≈ 59%
• P(At Least 1 TB) ≈ 41%

This is elevated compared to tour average (~30%) due to weak hold rates creating break exchanges that neutralize.

Totals Analysis

Model Fair Value (Locked from Blind Model)

Model Prediction:

• Expected Total Games: 24.8 games (95% CI: 21.2 - 28.4)
• Fair Line: 23.5 games
• Methodology: Derived from hold/break rates (76.1% / 74.3%), high break frequency (7.9/match), and 58% three-set probability

Probabilities at Common Lines:

• P(Over 20.5): 82%
• P(Over 21.5): 74%
• P(Over 22.5): 63%
• P(Over 23.5): 50%
• P(Over 24.5): 39%

Market Comparison

Market Line: 21.5 games

• Over 1.71 (implied 58.5%)
• Under 2.19 (implied 45.7%)
• No-vig probabilities: Over 56.2% / Under 43.8%

Edge Calculation:

• Model P(Over 21.5): 74%
• Market no-vig P(Over 21.5): 56.2%
• Edge: +17.8 percentage points

Key Totals Drivers

1. Weak Hold Rates (76.1% / 74.3%): Both well below tour average (~82%), creating frequent service breaks
2. High Break Frequency: Combined 7.9 breaks per match pushes games upward
3. Three-Set Probability (58%): Elevated due to competitive hold/break profiles
4. Historical Average: Both players average 24.2 games per match
5. Tiebreak Likelihood (41%): Moderate-high probability adds 2+ games when TB occurs

Why the Market is Wrong

The market line of 21.5 is 2 games below our fair value of 23.5. This appears to be:

1. Overweighting Elo gap: Market sees Shapovalov (-420 ML) as dominant, expecting quick straight-sets wins
2. Underweighting hold/break profiles: Ignoring that BOTH players are weak holders who generate high-game matches
3. Ignoring three-set probability: 58% chance of third set alone pushes expected games to 24+
4. PCB’s clutch stats overlooked: His 72.7% TB rate and superior closing stats keep sets competitive

The model expects 22% probability of 20-22 games (Under territory), but 78% probability of 23+ games (Over territory at 21.5 line).

Recommendation

OVER 21.5 games @ 1.71

• Edge: +17.8 pp (massive)
• Expected Value: +31.6% ROI
• Confidence: HIGH
• Stake: 2.0 units

Handicap Analysis

Model Fair Value (Locked from Blind Model)

Model Prediction:

• Expected Margin: Shapovalov +2.8 games (95% CI: +0.4 to +5.2)
• Fair Spread: Shapovalov -3.5 games
• Methodology: Elo advantage (+213 points) tempered by PCB’s clutch/TB edge (72.7% vs 50.0%)

Spread Coverage Probabilities:

• P(Shapovalov -2.5): 58%
• P(Shapovalov -3.5): 48%
• P(Shapovalov -4.5): 36%
• P(Shapovalov -5.5): 24%

Market Comparison

Market Line: Shapovalov -3.5 games

• PCB +3.5 @ 1.81 (implied 55.2%)
• Shapovalov -3.5 @ 2.04 (implied 49.0%)
• No-vig probabilities: PCB +3.5 covers 53.0% / Shapovalov -3.5 covers 47.0%

Edge Calculation:

• Model P(Shapovalov -3.5): 48%
• Market no-vig P(Shapovalov -3.5): 47.0%
• Edge: +1.0 percentage point (minimal)

Key Spread Drivers

1. Elo Gap (+213 Shapovalov): Suggests 3-4 game advantage
2. Hold/Break Parity: PCB +1.8% hold, +0.4% break narrows margin
3. PCB Clutch Edge: 72.7% TB rate (vs 50.0%) and superior closing stats (85.9% serve-for-set vs 77.8%)
4. Three-Set Likelihood (58%): Provides PCB more opportunities for comebacks
5. Dominance Ratios: Nearly identical (PCB 1.32 vs Shapo 1.35)

Market Alignment

The market spread of -3.5 is exactly our fair value. The no-vig probabilities (53.0% / 47.0%) closely match our model (52% / 48%).

This suggests:

1. Market efficiently pricing the Elo gap vs PCB’s clutch/close-match profile
2. No exploitable mispricing exists
3. Edge (+1.0 pp) is well below our 2.5 pp minimum threshold

Recommendation

PASS on spread market

• Edge: +1.0 pp (below 2.5 pp minimum)
• Confidence: LOW
• Stake: 0 units

The market has priced this spread efficiently. While our model slightly favors Shapovalov -3.5 (48%), the edge is insufficient for a recommended play.

Head-to-Head

Note: Head-to-head data not available in briefing. If available from supplementary sources:

• Focus on game margins in previous meetings
• Note surface-specific results (hard court)
• Identify any patterns in set scores or tiebreak frequency

Market Comparison

Totals Market

Market Efficiency: The totals market appears significantly mispriced. The market is pricing this as a potentially one-sided match (Shapovalov dominance), while the hold/break profiles indicate a high-game, competitive battle.

No-Vig Calculation (21.5 line):

• Implied Over: 58.5% → No-vig: 56.2%
• Implied Under: 45.7% → No-vig: 43.8%
• Vig: 4.2%

Spread Market

Market Efficiency: The spread market is efficiently priced, with model and market probabilities aligning closely (48% vs 47.0%).

No-Vig Calculation (Shapovalov -3.5):

• Implied PCB +3.5: 55.2% → No-vig: 53.0%
• Implied Shapovalov -3.5: 49.0% → No-vig: 47.0%
• Vig: 4.2%

Recommendations

Totals Recommendation

OVER 21.5 games @ 1.71

Rationale:

• Model expects 24.8 games vs market line of 21.5 (3.3 games above)
• Both players are weak holders (76.1% / 74.3%) with high break frequency (7.9/match)
• 58% three-set probability alone pushes expected games to 24+
• Both players average 24.2 games per match historically
• Model gives 74% probability of Over 21.5 vs market 56.2% (no-vig)

Risk Factors:

• Shapovalov blowout scenario (2-0 with 6-2, 6-3 type sets) would hit Under
• Market sees 25% chance of Shapovalov 2-0 dominance
• However, PCB’s clutch stats (72.7% TB, 85.9% serve-for-set) make bagel/breadstick sets unlikely

Confidence: HIGH Edge: +17.8 pp Stake: 2.0 units Expected ROI: +31.6%

Spread Recommendation

PASS

Rationale:

• Model fair spread (-3.5) matches market line exactly
• Edge of +1.0 pp is well below our 2.5 pp minimum threshold
• Market has efficiently priced the balance between Shapovalov’s Elo advantage and PCB’s clutch performance
• No exploitable mispricing exists

Confidence: LOW Edge: +1.0 pp Stake: 0 units

Confidence & Risk Assessment

Totals Confidence: HIGH

Supporting Factors:

• ✅ High-quality data (65 and 49 matches played)
• ✅ Both players show consistent weak hold rates (76.1% / 74.3%)
• ✅ Historical averages align (both 24.2 games/match)
• ✅ High break frequency (7.9/match) is stable driver
• ✅ Large edge (+17.8 pp) provides margin for error
• ✅ Model P(Over 21.5) = 74% is strong conviction

Risk Factors:

• ⚠️ Shapovalov blowout scenario (25% probability) hits Under
• ⚠️ Hard to see 20-game scenario unless PCB collapses mentally
• ⚠️ 41% tiebreak probability adds variance (but TB adds games, helping Over)

Expected Outcomes:

• Most likely: 23-27 games (57% combined probability)
• Best case: 28-31+ games (21% probability)
• Worst case: 20-22 games (22% probability)

Spread Confidence: LOW

Risk Factors:

• ⚠️ Minimal edge (+1.0 pp) below betting threshold
• ⚠️ Model and market aligned at -3.5 line
• ⚠️ Expected margin (2.8 games) has wide CI (+0.4 to +5.2)
• ⚠️ PCB’s clutch edge creates uncertainty in margin prediction

Pass Justification:

• Market is efficiently priced
• Edge insufficient for recommended play
• Better value exists in totals market

Key Unknowns & Caveats

1. Surface Precision: Briefing shows “all” surface rather than “hard” specifically - surface-adjusted hold/break rates may vary
2. Tournament Context: Dubai conditions (court speed, altitude, time of day) unknown
3. Fitness/Motivation: No information on player physical condition or tournament importance
4. Head-to-Head: Historical matchup data not included in briefing
5. Recent Momentum: Form trends are “stable” for both, but no details on last 5-10 matches
6. Elo Data Quality: PCB’s Elo (1200, Rank #607) seems unusually low for an ATP-level player - potential data issue

Data Quality: HIGH overall, but surface and context details would strengthen confidence.

Sources

Primary Data Source:

• api-tennis.com (stats, odds, player profiles)
• Collection timestamp: 2026-02-24T06:58:32Z
• Data window: Last 52 weeks (365 days)

Elo Ratings:

• Jeff Sackmann’s Tennis Data (GitHub CSV)

Odds Data:

• api-tennis.com multi-bookmaker feed
• Bookmakers: 10Bet, WilliamHill, bet365, Marathon, Unibet, Betfair, 188bet, Pinnacle, Sbo, 1xBet, Betano, 888Sport

Methodology:

• Hold/break analysis methodology: .claude/commands/analyst-instructions.md
• Report template: .claude/commands/report.md
• Anti-anchoring protocol: Two-phase blind model building

Verification Checklist

Data Validation:

• ✅ Briefing file loaded successfully
• ✅ HIGH data quality confirmed
• ✅ 65 matches (PCB) and 49 matches (Shapovalov) analyzed
• ✅ Hold/break statistics present for both players
• ✅ Totals and spread odds available from api-tennis.com

Model Validation:

• ✅ Blind model built without odds data (Phase 3a)
• ✅ Fair lines locked before market comparison (Phase 3b)
• ✅ Expected total games (24.8) derived from hold/break rates
• ✅ Expected margin (Shapovalov +2.8) derived from Elo + clutch stats
• ✅ Game distribution probabilities sum to 100%

Market Analysis:

• ✅ No-vig probabilities calculated correctly
• ✅ Edge calculations: Model P - Market P (no-vig)
• ✅ Totals edge: +17.8 pp (HIGH)
• ✅ Spread edge: +1.0 pp (below threshold)

Recommendations:

• ✅ Totals: OVER 21.5 recommended (edge ≥ 2.5 pp threshold)
• ✅ Spread: PASS recommended (edge < 2.5 pp threshold)
• ✅ Stakes assigned per confidence system (2.0 units for HIGH)
• ✅ Risk factors documented

Report Quality:

• ✅ All sections complete per template
• ✅ No moneyline analysis included (correct market focus)
• ✅ Hold/break rates prominently featured
• ✅ Game distribution modeling included
• ✅ Sources and methodology documented

Analysis completed: 2026-02-24 Model version: Two-phase blind model (anti-anchoring protocol) Data source: api-tennis.com (briefing file) Analyst: Tennis AI - Claude Code