Bet Data Labs is an independent data and analytics service. We are not a sportsbook and we do not accept wagers. No outcome is guaranteed. Past performance does not predict future results.
4 games this season, 6 games last season, 73% on this seasonRole shift
PASS
◇⊘
OVER
20.5
UNDER
20.5
Our win %
52%
Market %
50%
Edge
+2.1pp
Market
Rushing Yards 20.5
DK
1.92
-109
Projection5.2vs20.5(15.3 under)
L5
100%
5/5
L10
100%
10/10
L20
100%
11/11
SZN
100%
11/11
H2H
-
ACC
54%
n=990
ACC is Bet Data Labs' graded record on NFL rushing yards under plays: 54% of 990 settled cards. It measures us, not this player.
Edge
+2.1pp
EV
-0.4%
Alpha Score
0
Recommended
0.00u
PASSWATCHLEANSTRONGPRESS
AB 2-2 on Johnson
◆Best lineBetter number: Under 22.5 at 1.88 (-114) (+27.5% EV)
Why this card
›Rushing Yards projection 5.2 vs line 20.5
›Model read P(under) 66% sits 16pp off the market. Rushing Yards has no graded record yet, so it keeps the default 5.0pp lean, so the read was pulled back to that cap before pricing.
›Edge test fails: a 20.5 line on Rushing Yards sits below where the model has a read. Opportunity bet, not a priced edge.
›Prop market fair P(under) 50% across 2 books
›Calibration: NFL Rushing Yards plays predicted ~57% have hit 54% over 990 graded, of which 496 still describe the current model. Win prob cut 2.9pp.
›Baby line: 20.5 is below the 40 floor for Rushing Yards. Opportunity bet, capped at LEAN.
›Injury kill: CHI has 6 OUT
KILL SIGNALS
⚠ baby line
⚠ CHI has 6 OUT
PLAYS LIKE THIS
Bet Data Labs has graded 990 NFL rushing yards cards. The model predicted 57% on average and they hit 54%. Weighted for model changes since, that history is worth 496 cards about the model running now. This card's win prob was cut 2.9pp for it.
Sharp anchor
SourceBook consensus
Implied P(under 20.5)49.8%
Fair price1.99 · -101
Books de-vigged2
Pulled10/17, 3:19 AM ET
Fair value is the mean of 2 independent two-sided de-vigs. Reference books (BetMGM, Caesars, BetRivers, reduced-juice) count toward this number but can never set the price above, which comes from a Connecticut book only.
Our projection
Projection vs line5.20 vs 20.5
4 games this season, 6 games last season, 73% on this seasonRole shift
Sourcegamelog
Current games4
Prior games6
Prior games fetched7
Current mean5.50
Prior mean2.40
›All model inputs (15)
Market lines
Book
Line
Over
Under
DK★ Best
20.5
1.87 (-115)1.87
1.92 (-109)1.92
FD
22.5
1.88 (-114)1.88
1.88 (-114)1.88
DK implied, with vig (this side): 52.2%
Confidence readOn the fence
52% to hitrange 50-54%price needs 52%
The price's breakeven sits inside our range, so this one turns on exactly where the projection lands.
The band is our projection's standard error run through the same model that sets this card's win probability, so it shows how much the number could move if our read on the player's mean is off. The tick is what -109 has to beat to profit. We grade against that instead of a threshold we invented.
◆ ALPHA SCORE BREAKDOWN
Why 0?
TOTAL
0/100
Win prob+20.9/ 55
Calibrated win probability 52%
Market+-0.5/ 15
Market consensus leans against us (50% our side)
Price+-1/ 15
Negative EV at best price - juice eats the edge
Edge test+0/ 15
Edge test failed
Kill penalty-8/ 0
1 kill signal firing (graduated)
Components sum to 11; published as 0 after a tier rule (baby-line LEAN cap, or no PRESS without the market ≥55% on our side).
Best value is the 22.5 at 1.88 (-114) (27.5% EV) - a better number than the 20.5 this card is priced at.
Line
Win
Push
Loss
Price
EV
20
65%
1%
34%
1.52 (-191) fair
-
20.5now
66%
-
34%
1.92 (-109)
+26.5%
21
66%
Line history
Loading…
Typical pre-close movement
Across 15 replayed days of NFL, this market moves 0.9pp of implied probability on average between six hours out and the close, and the line itself moves 62% of the time (mean 0.88 units of line across every pair measured, movers and non-movers together).
Absolute magnitude over 208 players quoting player_rush_yds, so signed moves cannot cancel. A volatile market is not an inefficient one: this says how much the close knows that six hours out did not, not which way.
Injuries
From ESPN · refreshed every 30 min
CHI
Out · 6
Anthony Johnson Jr.OUT
out
Keyshaun ElliottOUT
Elliott (quadriceps) has been ruled out for Sunday's game against the Falcons.
Dayo OdeyingboOUT
Odeyingbo (quadriceps) has been ruled out for Sunday's game against the Falcons.
Kyler GordonOUT
Gordon (calf) is listed as questionable for Sunday's game against the Falcons.
T.J. EdwardsOUT
Edwards (hamstring) has been ruled out for Sunday's game against the Falcons.
Kyle MonangaiOUT
Monangai (thumb/toe) was ruled out for Sunday's game in Atlanta.
Alpha Score breakdown
EDGE TESTFAILS
TierPASS
Kelly fraction (full)0.00u
Quarter Kelly (recommended)0.00u
EV %-0.4%
Edge %+2.1pp
Card ID: c-401872999-RoschonJohnson-RUSHING_YARDSPublished 10/17/2026, 7:19:41 AM
Variance25.30
Modelnfl-v0.1-gamelog-two-season
Pricingpx-shared-v2+logistic-normal-v1
sourcegamelog
current_games4
prior_games6
prior_games_fetched7
current_mean5.50
prior_mean2.40
current_weight0.730
role_shift1
opp_adj1
games11
player_espn_id4426386
season2026
1%
33%
1.50 (-200) fair
-
22.5best EVtighten
68%
-
32%
1.88 (-114)
+27.5%
Win / push / loss come from our projected distribution. A whole-number line can push - an exact landing returns your stake - so buying to the whole number nearest our projection converts a slice of losses into stake-back. EV is shown only where a book actually quotes that line; other rungs show the model's fair price.
Best EV is the highest-value rung inside the trustworthy band. Tighten is the safest rung that stays +EV, for when a leg needs to survive rather than pay big. EV is our model's edge at the price a book actually quotes. Which rungs we trustis now de-vigged: we measure the book's margin on the two-sided main line (its vig-free read is 49% our side there) and strip that same margin off each one-sided rung. A losing-side rung is shown only when the book's own de-vigged price backs the model up, so a longshot can no longer post a phantom edge built on the model's fat tail. EV in (parentheses)is a rung where the model claims far more than the book's de-vigged number: indicative only, never a recommendation.
Questionable / DTD · 2
Tyrique Stevenson Sr.Q
Stevenson (hip) is questionable for Sunday's game against the Falcons.
Caleb WilliamsQ
Williams (hamstring) is listed as questionable for Sunday's game against Atlanta.
Other · 17
D'Andre Swift?
Swift will once again absorb a heavier backfield workload after Kyle Monangai (thumb/toe) was ruled out for Sunday's game against the Falcons, Larry Mayer of the Bears' official site reports.
Case Keenum?
Caleb Williams (hamstring) is listed as questionable for Sunday's game against Atlanta, but if he's not given the green light to return, then it's unclear whether Keenum or Tyson Bagent starts under center for Chicago, Larry Mayer of the Bears' official site reports.
Tyson Bagent?
Whether Bagent starts or serves in a backup role against the Falcons on Sunday depends largely on whether Caleb William (hamstring) is cleared to return from a three-game absence, Larry Mayer of the Bears' official site reports.
Shemar Turner?
The Bears activated Turner (knee) from their reserve/PUP list Tuesday.
Kalif Raymond?
Raymond caught three of four passes for six yards and logged one carry for minus-2 yards during the Bears' 31-7 loss to the Packers on Sunday. He also returned one punt for nine yards.
Montez Sweat?
The NFL announced Tuesday that Sweat has been suspended two games without pay for violating the league's performance-enhancing substances policy, Brad Biggs of the Chicago Tribune reports.
Ozzy Trapilo?
Jayden Loving?
Jamree Kromah?
Cairo Santos?
Santos missed his lone field-goal attempt and converted on his only PAT during the Bears' 31-7 loss to the Packers on Sunday.
Colston Loveland?
Loveland brought in two of five targets for 19 yards in the Bears' 31-7 loss to the Packers on Sunday.
Rome Odunze?
Odunze brought in three of nine targets for 40 yards in the Bears' 31-7 loss to the Packers on Sunday. He also lost a fumble.
Luther Burden III?
Burden secured four of six targets for 47 yards and rushed four times for two yards in the Bears' 31-7 loss to the Packers on Sunday.
Noah Sewell?
Sewell (Achilles) is active for Sunday's game against the Packers.
Darnell Wright?
Joe Thuney?
Jonah Jackson?
ATL
Out · 1
C.J. HendersonOUT
Henderson (groin) has been ruled out for Sunday's game against the Bears, Tori McElhaney of the Falcons' official site reports.
Questionable / DTD · 1
Za'Darius SmithQ
Smith (neck) is listed as questionable for Sunday's game against the Bears, Tori McElhaney of the Falcons' official site reports.
Other · 23
Tua Tagovailoa?
With Michael Penix (ribs) cleared to play Sunday against the Bears, Tagovailoa will handle backup duties for the Falcons.
Michael Penix Jr.?
Penix (ribs) doesn't have a designation for Sunday's game against the Bears, Tori McElhaney of the Falcons' official site reports.
Kendal Daniels?
Cameron Thomas?
Divine Deablo?
Deablo (hamstring) was a full participant at practice Thursday, Tori McElhaney of the Falcons' official site reports.
Jahan Dotson?
Dotson caught two of four passes for 54 yards during the Falcons' 30-23 overtime loss to the Ravens on Sunday.
Brian Robinson Jr.?
Robinson handled seven carries for 16 yards and did not catch his only target during Sunday's 30-23 overtime loss versus the Ravens. He also returned two kickoffs for 61 return yards.
Jack Strand?
Ethan Onianwa?
Malcolm DeWalt IV?
Christian Harris?
Robert Longerbeam?
Nick Folk?
Folk converted all five of his kicks (three field goals, two extra points) during Sunday's 30-23 overtime loss to the Ravens.
Kyle Pitts Sr.?
Pitts caught five of six targets for 42 yards and a touchdown in Sunday's 30-23 overtime loss to the Ravens.
Drake London?
London caught seven of 11 targets for 118 yards in Sunday's 30-23 overtime loss to the Ravens.
Bijan Robinson?
Robinson rushed 17 times for 86 yards while catching four of seven targets for 32 yards in Sunday's 30-23 overtime loss to the Ravens.
Jake Matthews?
Matthews (groin) is active for Sunday night's game against the Ravens, Tori McElhaney of the Falcons' official site reports.
Cooper Rush?
Rush (coach's decision) is inactive for Sunday's game against the Ravens but will serve as the emergency quarterback, Tori McElhaney of the Falcons' official site reports.
Samson Ebukam?
Yasir Abdullah?
The Falcons placed Abdullah (hamstring) on injured reserve Saturday, Tori McElhaney of the team's official site reports.
Michael Jerrell?
Harold Perkins Jr.?
Perkins recorded 10 tackles (seven solo) including 0.5 sacks and a pass defensed during Monday's 45-24 win at New Orleans.
Austin Hooper?
Hooper caught his only target for nine yards during Monday's 45-24 win at New Orleans.
★ RECENT FORM
Roschon Johnson · Rushing Yards vs 20.5
ESPN gamelog · hit rates vs line
LAST 5
100%
4/4 under · avg 5.5
LAST 10
100%
4/4 under · avg 5.5
SEASON
100%
4/4 under · avg 5.5
Cells scored against the Under, the side this card is on. At 1.92 (-109) the price needs 52% to break even: green clears it with room, red falls short. A high hit rate at heavy juice is not an edge, which is why we grade against the price and not a round number.
Team form
ESPN standings · L10 / streak / splits
AWAYCHI
3-2
L103-2
StreakL1
Road0-0
Run diff+23.0
HOMEATL
2-3
L102-3
StreakL1
Home0-0
Run diff-3.0
Recent form is context, not signal. A team on a heater can still be the wrong side of the number. Cross-check with the projection above.
Game log
Roschon Johnson · Rushing Yards vs 20.5
ESPN gamelog · newest first · scored for the under
100%5/5 under
Oct 11@GBL6Hit
Oct 4vsNYJW16Hit
Sep 29vsPHIW0Hit
Sep 20vsMINL0Hit
Nov 16@MINW0Hit
Nov 9vsNYGW0Hit
Oct 19vsNOW11Hit
Oct 14@WSHW6Hit
Sep 28@LVW0Hit
Sep 21vsDALW0Hit
Sep 14@DETL0Hit
The five most recent games with a readable stat line. Dimmed rows are outside the selected window, not excluded from the record. Hit and miss are scored against 20.5, this card's line, so the same game can be a hit here and a miss on a different rung. The price needs 52% to break even, which is the bar these percentages have to clear.
About ACC Our plays in this bucket (NFL rushing yards under) hit 54.3% over 990 graded cards, 988 of them replayed from historical lines. That is our settled record in this market, not this card's win probability and not a claim about the board as a whole.
