Will the Denver Broncos beat the Kansas City Chiefs at home on November 1, 2026?

The Kalshi market has DEN at roughly 47-48% to beat KC at home, below the unconditional NFL home-field base rate of ~57%, implying the market judges KC meaningfully stronger. I find no statistically significant home-field edge in DEN's own recent game log (s4, s7 both insignificant with wide CIs and small n), so I cannot justify a large deviation from the market price. The only directional nudge I apply is from recent-form divergence (DEN's last logged margin +14 on 10/25 vs KC's last logged margin -17 on 10/26), but this is drawn from very small, highly autocorrelated samples (s2, s3) and a failed correlation test, so I weight it lightly. I land essentially at the market price with a small upward adjustment, not a confident departure from it.

Own objection: The market price itself (d5≈0.475) could already be stale or thin — Kalshi sports contracts sometimes lag sportsbook lines, especially 4 days out. If I had the actual espn_spread for this game (d4 failed to load) I might find the 'true' market consensus is quite different from the Kalshi print, meaning I'm anchoring to a potentially mispriced reference point rather than an efficient one. Also, my upward nudge from 0.475 to 0.49 is based on noisy single-game form (DEN +14 last game, KC -17 last game) in series with n=9-10 and p=0.815 insignificance — this could just be overfitting recency into a number that has no real predictive signal, and the honest estimate should stay at or even below the raw market print.

The Kalshi market itself (d1) has priced DEN's win probability tightly around 0.472-0.475 for the four days leading into the game, with high day-to-day autocorrelation and a very narrow bootstrap CI, indicating a stable, converged market consensus rather than a noisy or stale quote. No data available to me as of 2026-10-29 provides an out-of-sample edge over this price. My estimate stays close to the market, with a small downward nudge reflecting the historical pattern of KC dominance in this specific matchup even in DEN home games (not something fully testable here with the series provided).

Own objection: The market price (d1) could itself be stale or thin (Kalshi sports markets sometimes have low liquidity), and the other series (d2-d8) that might capture recent team-specific form, injuries, or scoring trends were not successfully linked to a clear causal/predictive test against the outcome — the only significant correlation found (s6, corr=-0.85) has n=9, flagged explicitly as small-sample and not reliable (CI crosses zero at the upper bound), so I have no validated signal that beats the market price. If there is late-breaking injury news (e.g., a key QB or defensive starter ruled out) between now and kickoff that isn't reflected in d1's last timestamp (2026-10-29), my estimate would be stale in the opposite direction of whatever that news implies.

The market has already converged on DEN's win probability at ~0.475, matching the implied spread-based base rate for a near-pick'em home underdog; price stability and high autocorrelation indicate low crowding and little residual tradeable edge beyond the current quote, absent new information not captured in this dataset.

Own objection: The d5 series (n=211, last values jumping to 0.82 on 2026-10-28/29) is a glaring anomaly: if it is in fact tracking the same game's win probability (or a closely related derivative like a live/model-based probability) and not noise, it implies something priced very differently from d1/d2/d4 — a ~35-point implied-probability gap. If I am wrong to dismiss it as mislabeled/unrelated, the true market-implied probability could be far from 0.475, and my crowding/edge conclusion would be invalid. I have no metadata to confirm d5's definition, which is a real blind spot.

None of the statistical 'signals' presented (s1-s3, s6) survive scrutiny as clean causal drivers of Sunday's Broncos-Chiefs outcome: they are small-sample (n=25-27), show parametric/nonparametric disagreement, carry unverified event-selection criteria, or rest on correlations undermined by severe autocorrelation. The closest thing to an unconfounded estimate is the market price itself (d8≈0.475), which I treat as the primary anchor, adjusting only slightly downward from the generic NFL home-field base rate (~57%) because the market has already priced in whatever specific-game information exists, and I have no verified basis to override it with the fragile in-sample signals.

Own objection: If d7/d8 genuinely are Kalshi's own DEN-win contract price trending toward settlement, then the market itself is the best available forecast and any causal-skeptic discount I apply is just noise-adding, not information-adding — in an efficient market, my job is to defer, not to shave the number for unverified confounds. If I'm wrong about which series maps to the actual contract, my whole anchor is misplaced."

debate skipped: panel unanimous within 0.02

debate skipped: panel unanimous within 0.02

debate skipped: panel unanimous within 0.02

debate skipped: panel unanimous within 0.02

All four analysts land within 0.02 of each other (0.47-0.49) and the deterministic pool is 0.47, but this near-unanimity is largely an artifact of shared evidence rather than four independent confirmations: every analyst's primary input is the same Kalshi market-price series (d1/d2/d5/d8), re-sampled and re-tested with different statistics. Collapsing duplicated evidence, there are really only two independent evidence lines here: (1) the market price itself (~0.472-0.485, tight and stable, cited by all four), and (2) DEN's own recent game log tested for a home-field effect, which is statistically null (p=0.815, p=0.787) and too small-n to be trusted either way. No analyst supplied team-strength, injury, or weather data independent of the market quote. Base rate: unconditional NFL home teams win ~57% historically. The market's ~47.5% implies the committee (and bettors) judge KC strong enough to flip a ~10-point home-field edge into a slight road favorite. I did not find grounds to override this -- the only semi-independent signal (DEN's recent-form divergence, +14 vs KC's -17) is a single-game, highly autocorrelated data point that Analyst A himself down-weighted. Calibration quality across this segment is mediocre for all four analysts (Brier 0.26-0.29, accuracy 0.46-0.56, i.e., near coin-flip historically), which argues for a wider CI than the market's own tight bootstrap band and for treating this as sitting on the market price with real residual uncertainty, not high-confidence agreement. Consensus: 0.47, CI [0.42, 0.53] -- wider than the market's own CI to reflect poor historical analyst calibration and untested small-sample caveats. Expected return: 0.0 -- no analyst identified an edge over the current market price; buying or selling DEN at ~47.5 cents has no demonstrated positive expected value from this evidence set. Top disagreement (implicit): base rate (~57% home win) vs. market price (~47.5%) -- resolved by treating the market as the superior, more information-dense estimate, with the gap explained by team-quality asymmetry rather than mispricing. NOT TRADEABLE on this evidence: the market has already converged (autocorr 0.746, 6 consecutive prints at 0.475) and no committee member found a statistically significant edge over that price. Recommend no position; if a trade is forced, it should be sized as near-zero (e.g., a small, cheap hedge only) given crowding=0.85 and novelty=0.1 -- this is a stale, well-priced market, not a mispricing opportunity.

Consensus 48% · interval 42% to 53%