Seventy-one sportsbooks quote Jets @ Steelers. The spread ranges from -3.5 to +2.5. That is a 6-point gap . Seventy-one books quote Chiefs @ Bucs. Every single one says -5.5. That is a zero-point gap . The gap is the EV analytics price. The bigger the gap, the more room for edge.
What EV analytics price actually means
EV analytics price answers one question: how much better is your bet than the final market line? You bet Bills -3 on Wednesday. The line closes at Bills -5. You captured 2 points of positive EV. That is the price. It is not a fee. It is the math edge in every wager.
The market shows you the price through dispersion — the gap between the highest and lowest line across all books. Zero dispersion means every book agrees. The market has spoken. Your EV price is near zero unless you have private info. High dispersion means books disagree. That disagreement is where EV lives.
Live proof: 32 games, three sports, one pattern
Our odds feed captured 32 games on 2026-08-21. The pattern is immediate:
- MLB run lines: 11 games, 4-7 books each, zero dispersion on every single one. Braves @ Brewers -1.5 across 7 books . Giants @ Red Sox -1.5 across 6 books . Nationals @ Marlins -1.5 across 4 books . The market is locked. EV price = near zero.
- NFL spreads: 14 games, 71 books each, dispersion ranges from 0 to 6 points. Chiefs @ Bucs locked at -5.5 across all 71 books . Jets @ Steelers spans 6 points (-3.5 to +2.5) across 71 books . That 6-point window is the EV price on display.
- College football: 7 games, 38-71 books, dispersion 0-1 point. Massive favorites (NMSU @ FSU -31.5 across 71 books ) show zero dispersion. Tighter games (Sacramento St @ EMU -9.5 to -8.5 across 71 books ) show 1 point — enough for a half-point of CLV (closing line value) if you pick the right side.
The takeaway: sport choice sets your EV floor. MLB run lines offer almost no dispersion EV — see why run lines are different. NFL and CFB spreads offer real windows — but only on certain games. NFL spread mechanics explain why.
How dispersion maps to capturable CLV
Dispersion is not EV itself. It is the ceiling for EV. If books span 6 points, the most you can capture is roughly half that (3 points) by being on the right side of the midpoint. The Jets @ Steelers game is the textbook case. 71 books. 6-point spread. The midpoint is roughly -0.5. If you bet Steelers +2.5 early and the close lands at -0.5, you captured 3 points of CLV. That is a real EV price on a single wager.
- 0.0-0.5 pts dispersion (most MLB, blowout NFL/CFB): Market is efficient. EV price = 0. You need a model edge, not line shopping.
- 0.5-1.5 pts (many CFB, some NFL): Thin. Maybe a half-point of CLV if you time it right. EV price = small.
- 1.5-3.0 pts (Commanders @ Lions 3 pts , Seahawks @ Titans 1.5 pts ): Real. 1-2 points of CLV capturable. EV price = meaningful.
- 3.0-6.0 pts (Jets @ Steelers 6 pts , Panthers @ Jaguars 4 pts ): Strong. 3+ points of CLV on the table. EV price = high.
Book count weights the signal
Dispersion without book count is noise. Angels @ Rangers shows zero dispersion — but only 2 books quoted it . That is not a tight market. That is a thin market. Conversely, Chiefs @ Bucs shows zero dispersion across 71 books . That is a genuinely efficient price. Our chart weights by book count automatically. The NFL games (71 books) carry far more signal than the 2-book MLB games. When you scan for EV, filter for games with 30+ quoting books. The dispersion on those games is the real EV price. Sharp vs public money shows why book count matters.
One practical step using live dispersion
Open the odds page, sort by dispersion descending, filter for 30+ books. Bet the 3+ point dispersion games. That is the play. Track your CLV to prove the edge. Know the vig you are fighting. Build a model if dispersion alone is not enough.
What would change your mind
If Jets @ Steelers closes within 1 point of the opener next week, dispersion EV is dead for now. Watch the 6-point window . If it collapses, the market has gotten too efficient for this approach.
Bet responsibly — set limits, never chase losses.
Model calibration: predicted vs observed
Predicted win probability bucket vs the empirical win rate inside that bucket on the test set. Points on the y=x reference line are perfectly calibrated; points below mean the model is overconfident in that bucket.
EV per $100 across win rate × odds grid
Expected value of a $100 stake at each combination of true win rate and market odds. Anywhere the cell is positive you have a long-run profitable bet; the magnitude shows how aggressive Kelly will size it.
