kalshiMarch 13, 202613d ago9 or more upsets in 2026 March Madness Round of 64
Will there be 9 or more upsets in the Round of 64 at 2026 March Madness?
Signal
BUY
Probability
52%
Confidence
MEDIUM
58%
Summary.
The market is pricing 9+ Round of 64 upsets at 46% implied probability, treating 8 upsets as the most likely outcome. However, historical data (2010-2023) shows an average of 9.15 upsets per tournament under this broad definition (any lower seed defeating a higher seed), suggesting the true probability should be approximately 52%. The market appears to be overweighting 2025's extreme anomaly (only 3 upsets) while undervaluing the robust long-term average. Seed-by-seed analysis yields an expected value of 8.3 upsets, just below the threshold but well within normal variance. The broad upset definition critically includes 9-vs-8 matchups (four coin-flip games producing ~2 expected upsets), which creates a structural advantage for YES. While NIL and Transfer Portal talent concentration may be reducing upset rates, regular season data shows stable upset frequencies despite wider point spreads, suggesting tournament variance and single-elimination dynamics still dominate. Major uncertainty exists because Selection Sunday is March 15—just two days away—meaning specific bracket matchups, auto-bid quality, and injury situations remain unknown. The estimated 52% probability represents modest value against the market's 46%, but confidence is tempered (58%) by bracket unknowns and genuine uncertainty about whether 2025 signals a structural shift or statistical outlier.
Reasoning.
Step 1: Establish Base Rate Historical data (2010-2023) shows an average of 9.15 upsets per Round of 64 under this broad definition (any lower seed beats higher seed). This is critically important - we're essentially betting on whether the tournament will be above or below the historical median. Given normal distribution around 9.15 with standard deviation likely 2-3 upsets, the probability of hitting exactly 9+ should be approximately 50-55% in a vacuum.
Step 2: Seed-by-Seed Expected Value Breaking down the 32 Round of 64 games by seed matchups:
- 9 vs 8 seeds (4 games): 51.9% upset rate = 2.08 expected upsets
- 10 vs 7 seeds (4 games): 38.8% = 1.55 upsets
- 11 vs 6 seeds (4 games): 38.8% = 1.55 upsets
- 12 vs 5 seeds (4 games): 35.6% = 1.42 upsets
- 13 vs 4 seeds (4 games): 20.6% = 0.82 upsets
- 14 vs 3 seeds (4 games): 14.4% = 0.58 upsets
- 15 vs 2 seeds (4 games): 6.9% = 0.28 upsets
- 1 vs 16 seeds (4 games): ~0.5% = 0.02 upsets
Total Expected: 8.3 upsets, which is slightly below the 9 threshold.
Step 3: Adjust for Recent Structural Changes The 2025 tournament's 3-upset extreme is a major outlier but signals potential structural shift from NIL/Transfer Portal era. However, critical counterpoint: regular season upset rates remain stable despite wider spreads. This suggests talent concentration may not fully translate to tournament performance where:
- Single elimination creates variance
- Mid-majors peak for tournament preparation
- Matchup-specific factors (defensive style, pace, 3-point variance) still matter enormously
I weight 2025 as a 2-standard-deviation outlier (15% weight) rather than new normal (would need 2-3 more years of data).
Step 4: Market Comparison Kalshi pricing 9+ upsets at 46% implies market expects ~8 upsets as median. This appears to be:
- Overweighting 2025's anomaly
- Slightly undervaluing the historical 9.15 average
- Perhaps reflecting belief in permanent talent concentration
Step 5: Key Uncertainties
- Bracket unknowns: Cannot evaluate auto-bid strength, specific matchups, or injured stars until March 15
- 9 vs 8 games are essentially coin flips and provide ~2 upsets - high variance
- One extra 12-over-5 or 11-over-6 game pushes us over threshold
- 3-point shooting variance can swing multiple games in single-elimination format
Final Estimate: 52% I estimate slightly above 50% because:
- Historical average of 9.15 favors YES
- Market at 46% appears to overweight 2025 anomaly
- Broad upset definition includes near-coinflip 9-seed games
- Tournament variance and 3-point shooting create upset potential even in talent-concentrated era
However, confidence is modest (58%) due to bracket unknowns and genuine uncertainty about structural changes.
Key Factors.
Historical average of 9.15 upsets (2010-2023) positions 9+ as slightly above 50% probability baseline
2025's 3-upset anomaly appears to be overweighted by market, creating value on YES at 46% implied odds
Broad upset definition (any lower seed wins) includes ~2 expected 9-over-8 victories that are essentially coin flips
Expected value from seed-by-seed probabilities is 8.3 upsets, just below threshold but within variance range
NIL/Transfer Portal talent concentration is real but hasn't reduced regular season upset rates, suggesting tournament variance still dominates
Bracket unknowns until March 15 create uncertainty - cannot evaluate auto-bid quality, matchup specifics, or injury situations
3-point shooting variance in single-elimination format can swing multiple games and create upset clusters
Scenarios.
Chalky Tournament (Bear Case for YES)
28%2025's pattern continues with talent concentration dominating. Only 6-7 upsets occur, heavily weighted toward 9-over-8 coinflips with few 10+ seed victories. Top programs' roster advantages prove decisive.
Trigger: Strong 1-5 seeds dominate Selection Sunday. Few compelling mid-major auto-bids. Key injuries to double-digit seeds. Early games go chalk, creating momentum for favorites.
Historical Average (Base Case)
44%Tournament produces 8-10 upsets, clustering around historical 9.15 average. Mix of 9-seed victories (~2), competitive 10-12 seed wins (3-4), and one surprise 13+ seed victory. YES hits narrowly or misses by 1.
Trigger: Balanced bracket with mix of strong mid-majors and vulnerable high seeds. Standard 3-point shooting variance. One or two major conference teams underseeded due to late-season slump.
Upset-Heavy Tournament (Bull Case for YES)
28%11-13 upsets occur as double-digit seeds exploit favorable matchups and 3-point variance. Multiple 11-12 seeds win, plus a 13 or 14 seed victory. YES clears threshold comfortably.
Trigger: Weak 4-6 seed line with teams that limped into tournament. Strong mid-major auto-bids (think Saint Peter's 2022). Favorable pace/style matchups for underdogs. Hot shooting from multiple lower seeds.
Risks.
2025 tournament may signal permanent structural shift rather than outlier - NIL effects could take years to fully manifest
Selection Sunday could produce unusually strong 1-8 seed line and weak double-digit seeds, skewing matchups toward favorites
Key injuries to promising upset candidates between now and March 19-20 could reduce upset potential
Recency bias may be affecting my analysis - overweighting long-term historical average vs. recent talent concentration trends
Standard deviation around 9.15 average is likely 2-3 upsets, meaning 6-7 or 11-12 upset years are both plausible
Market may have superior information about 2026 bracket strength that I cannot access before Selection Sunday
Tournament draw luck matters - if best double-digit seeds cluster in same regions, they cannibalize each other's advancement
Edge Assessment.
MODEST EDGE ON YES at 46% market odds
The market's 46% pricing appears to overweight the 2025 anomaly (3 upsets) and undervalue the robust 9.15 historical average. Key edge drivers:
- Historical frequency: 9+ upsets has occurred in approximately 50-55% of recent tournaments, making 46% market odds slightly underpriced
- Definition advantage: Broad upset criteria (any lower seed) includes high-probability 9-over-8 games that alone contribute ~2 upsets
- Mean reversion logic: After 2025's extreme chalk, regression toward 9.15 average is more likely than repeat outlier
- Structural uncertainty cuts both ways: While NIL may reduce upsets, regular season data doesn't support this conclusively
Recommended bet sizing: Small to modest (1-2% of bankroll) given 58% confidence level. The edge exists but is narrow (52% true probability vs 46% market = 6 percentage point edge).
Key decision point: Would significantly upgrade to NO if Selection Sunday reveals dominant 1-5 seed line and weak mid-major auto-bids. Conversely, would increase YES confidence if strong mid-majors earn vulnerable matchups.
Uncertainty caveats: Cannot fully assess until bracket is revealed March 15. If this resolves March 20-21, there's only 5-6 days to react to Selection Sunday information, suggesting wait-and-see may be optimal strategy unless current 46% odds are exceptional value.
What Would Change Our Mind.
Selection Sunday (March 15) reveals an unusually strong 1-5 seed line with dominant teams and weak/injured double-digit seeds, suggesting chalk-heavy bracket
Multiple key injuries to promising mid-major upset candidates occur between now and tournament tip-off on March 19
Bracket draw creates unfavorable matchup clustering where best 10-12 seeds face strong defensive teams or pace mismatches that neutralize their advantages
Market odds move significantly toward YES (above 50%), eliminating the current value proposition
Additional data emerges showing 2025's chalk pattern continuing into 2026 regular season tournament games (conference tournaments March 10-15) with historically low upset rates
Analysis of final Selection Sunday bracket reveals unusually weak mid-major auto-bids compared to at-large teams, reducing competitive 11-13 seed pool
Sources.
Get This Via API.
Access real-time prediction market analysis programmatically. Every analysis on this page is available through our REST API.
curl -X POST https://api.rekko.ai/v1/analyze \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"category": "sports", "platform": "kalshi"}'Related Analysis.
Canadian team wins the Stanley Cup before the 2031 season
The market implies a 63% probability that a Canadian team wins the Stanley Cup between 2026-2030, but my analysis estimates a more conservative 52% probability—an 11-percentage-point overvaluation. This is essentially a bet on the Edmonton Oilers' championship window during Connor McDavid's prime (ages 29-33), as all other Canadian teams are non-competitive (Toronto/Vancouver rebuilding, Ottawa a longshot at +3300-4000). While McDavid's team-friendly extension through 2027-28 creates a legitimate 3-year window and the Oilers reached back-to-back Finals in 2024-2025, several factors suggest the market is overpricing this outcome: (1) Edmonton LOST both Finals, creating psychological hurdles that losing finalists historically struggle to overcome; (2) Current injuries are concerning—Leon Draisaitl has been out since March 15 with unclear playoff timeline, and McDavid has hip/groin issues; (3) Colorado upgraded to prohibitive favorite (+275-300) by acquiring Quinn Hughes; (4) The 2029-2030 seasons offer minimal value since McDavid's extension ends after 2027-28; (5) The market appears sticky at 63¢ despite recent negative developments, suggesting recency bias and McDavid halo effect rather than properly pricing injury risks and elite competition. My probabilistic model weights 2027-2028 as peak window years (12-15% each) but assigns only 6% to injury-plagued 2026 and 5% to uncertain 2030, yielding 52% cumulative probability.
Will humans colonize Mars before 2050?
The market is pricing a Mars colony by 2050 at 17.5%, but our analysis estimates just 3% probability—nearly a 6:1 mispricing favoring "No." The critical development is SpaceX's February 2026 strategic pivot to lunar colonization, explicitly delaying Mars missions by 5-7 years. This eliminates the only credible Mars settlement actor until the early 2030s, leaving merely 17-19 effective years for an unprecedented achievement requiring 15-20+ years minimum from today. The resolution criteria demands extreme technical sophistication: 10+ people surviving one full Earth year without resupply, requiring operational ISRU, radiation-shielded agriculture, manufacturing, and nuclear power. NASA's roadmap shows only exploratory missions (late 2030s/2040) with Earth resupply—no government agency has permanent Mars settlement planned. The market appears inefficiently high due to retail Musk enthusiasm not fully incorporating the recent pivot's implications, while sharp money is already favoring "No." The 24-year horizon creates false comfort; detailed milestone sequencing reveals timeline compression is nearly impossible given Mars's 26-month launch windows, 6-9 month transits, and self-sufficiency requirements. Only tail-risk scenarios (AI singularity enabling autonomous construction, or geopolitical space race) preserve ~3% probability.
Will Miami (OH) make the 2026 NCAA Tournament?
The market prices Miami (OH)'s NCAA Tournament selection at 69%, but our analysis estimates the true probability at 78% (+9 percentage points). Miami's 31-1 record—the first undefeated regular season since Gonzaga 2020-21—combined with strong résumé metrics (21st SOR, 33rd WAB) creates a compelling case despite today's MAC Tournament quarterfinal loss. The Selection Committee has historically never rejected a one-loss team, and teams ranked 33rd in Wins Above Bubble are selected at near-certainty rates. The exceptionally weak 2026 bubble (with Texas, SMU, and Indiana also losing recently) reduces competition for at-large bids. Expert bracketologists maintain Miami as a projected selection, likely via the First Four. However, legitimate concerns exist: zero Quad 1 wins, 355th-ranked non-conference strength of schedule, and 91st-93rd KenPom ranking represent unprecedented schedule weakness for an at-large candidate. The committee could theoretically prioritize predictive metrics and schedule quality over win totals, setting new precedent. With Selection Sunday just 3 days away (March 15, 2026), limited time remains for new information to emerge. The moderate positive edge reflects strong historical precedent and expert consensus, balanced against the unprecedented nature of this specific résumé profile and the general efficiency of high-volume prediction markets for major sporting events.