The whole screened universe, not just the legs that made a card.
Player Board: every prop we panel-read — our data platform's median vs. the line, hit-rate
windows, role share, opponent rank, and a Prop Score where the read was strong enough to score.
Game Board: every game's line and total, where the public's tickets and the sharp money
actually sit, three power-rating edges, the historical systems that apply, and our verdict.
Updated Sat Sep 26 · 2:31 PM ET
Proj = our data platform's median for the market (2025 season, L10 where the panel
showed it); Δ = projection minus the line. Windows are L5 / L10 / 2025 hit rates against the current
number. "Models" counts how many independent projections (platform median, a Las Vegas network's AI model
when posted, a fantasy publication's per-game baseline) sit on the read's side. Rows marked
card are legs published on a card at the price shown there — those prices never
change here. Everything else is a read, not a pick.
Show
Public $ = share of money vs. share of tickets at a major U.S. book; Sharp $ = the
same at a Las Vegas market-making book (small early-week samples — directional until Thursday).
Ratings edge = three published power-rating systems' projected line vs. the market. Systems = historical
Week-1 angles from a Las Vegas betting network's season guide, with their records. Verdicts with a
confidence % are published legs on that day's card; their prices are locked there.
How to read this board
Never seen odds before? Start here — this walks through one real pick, left to right.
Worked example — row 1, Browns
1 · The Odds
-108
The sportsbook's price. A minus number means you risk more than
you win — here, $108 to win $100. A plus number is the reverse.
2 · Implied
51.9%
That price as a percentage: the chance the market gives it.
This is the number to beat.
108 ÷ (108 + 100) = 51.9%
3–4 · Blended / Edge Gap
—
Blank on purpose. These columns need current-season hit-rate
history, and every pick on today's board is from a slate that has no 2026 games played
yet (or where our stats source is down). On those days the read comes from reported
workloads, verified news and line movement instead — and the card says so.
no data > fake data
5 · Matchup
Confirms
The last check: does tonight's opponent support the pick or work
against it? This can veto a leg however big the gap.
The one thing to understand before anything else. Edge Gap ranks picks
against each other — it does not predict how often they win. A +27.7 gap is not a
27.7% profit or an 82% chance tonight. Those historical rates came from a different set of
lines than the one on the board right now. Use it to see which pick is strongest today, not
to forecast a result. Our actual graded hit rate is 52%
— the honest number, on the track record page.
Every column, plainly
Player
Who the bet is on, game underneath. "TEX @ ATH" means Texas away, Athletics home.
Pick
Over = the stat lands above the number, Under = below. "H+R+RBI" is hits plus runs plus runs-batted-in combined.
Odds
The price when we published. These move — check your own book before acting.
Implied
Odds converted to a percentage. Computed from the odds, so it can never be mistyped.
Blended
Weighted hit rate across four windows, so a hot last-five can't outweigh a full season.
Gap ↓
Blended minus Implied. Sorted highest first. Above +40 we treat the line as broken and exclude it.
Matchup
Our read on tonight's opponent — positional defence, pitcher handedness, injuries both sides.
Confidence
How strongly our signals agree on this read — not how often it wins. Derived from Edge Gap and the matchup verdict, capped at 90%.
Prop Score
A 0–10 scorecard for player props built from six panel inputs (below). Stars are the shorthand: ★★★★★ 8.5+, four 7.0–8.4, three 5.5–6.9, two 4.0–5.4. Hover the score for the component breakdown.
Checkbox
Click any row to add it to the parlay calculator. Click again to remove.
The Prop Score, component by component
Component
Max
What earns it
Median cushion
2.5
how far the player's median sits past the line, scaled by stat
Hit-rate windows
2
L10 and 2025 agreement (both ≥80% = full, one ≥70% = half)
Role security
2
share of touches/targets, and whether the sample is THIS role
Opponent rank
1.5
rank on the exact stat pointing the pick's way
Script fit
1
team pass/run rate and expected game state support the volume
Line & market
1
line moved against a strong panel = full; already ran toward us = 0
Injury cap
6.0
a Questionable tag on the player (or his quarterback) caps the total at 6.0, reason printed
Every point traces to a field on the player's matchup panel — median vs the line, hit-rate windows, touch or target share, opponent rank on that exact stat, the team's pass/run rate, and where the line has moved. Because the components are stored, they get graded individually: after enough legs we can see which parts predict hits and re-weight the rubric in public. Nothing under 4.0 is published. Team picks carry confidence instead; they have no panel to score.
What “confidence” actually means
It is confidence in the read, not the odds of winning. A pick at 79% does
not win 79% of the time — our real graded hit rate is
52%, and that is the honest number. The percentage measures how
strongly the signals agree: it starts at 50, rises with Edge Gap (saturating at +35, because
past that a gap is more likely a broken line than a better bet), adds 12 when the matchup
Confirms, subtracts 20 when it Conflicts, and is capped at 90. The cap is
deliberate. Nothing we publish earns a 95.
Team / game-line picks work differently. They have no hit-rate history and
therefore no Edge Gap, so their confidence is assigned by the analysis — and every one
of them carries a written note saying what it rests on.
What the coloured labels mean
+27.7
Green — High. Edge Gap of +25 or more. The strongest tier we publish.
+19.6
Amber — Moderate. A real gap, but smaller. Still on the card, ranked below the greens.
Confirms
The matchup supports the pick. An Under on rebounds against the best rebounding defence, for instance. Our most trusted legs.
Neutral
Neither helps nor hurts. The pick stands on its Edge Gap alone.
Conflicts
The matchup argues against it — so we drop it, however large the gap. You won't see these here; they're listed in the post with the reason.
The parlay calculator, explained
Parlay
Several picks as one bet. Every leg must win or the whole thing loses. Bigger payout, much lower chance.
Book pays
The combined price. +421 means $100 returns $421 profit if every leg lands.
Market's chance
How likely the sportsbook's price says the whole slip is.
Our estimate
Each leg's blended rate multiplied together. Optimistic by nature — see below.
Model vs market
How far apart those two sit. Near 1× is believable; 3× or more means our model is wrong, not that you found free money.
Hits about
Roughly how often it lands by our own estimate. "1 in 3" loses twice for every win.
Correlated legs
Two picks from the same game. Outcomes affect each other, so multiplying overstates your real chance — and most books block these.
Why we show it
Others show a big payout number. We show when the maths stops being credible. Legs multiply our errors faster than your edge.