Dashboard

Latest report + tracker overview

Current Bankroll
$60.23
from $48.26 start · 25% growth
Main Bets Record
3-1
75.0% hit rate
Total P&L
+$11.97
on $31.92 staked
ROI
+37.5%
P&L / total staked
📄 Latest Report — Thursday, September 24, 2026

Daily Bet Research Report — Thursday, September 24, 2026

Start: 11:12 ET | End: 11:19 ET

Bankroll: $60.23 | Stake (12.5%, one play): $7.53 Record: 3-1-0 (Main bets, main leg, since 9/14 reset) | Hit rate: 75.0% Slate: 13 games · 9 analysed · 31 sources · Playbooks: mlb.md, nfl.md · Verified at 11:12 ET

Section 2: Game Analysis

Sport Game Time (ET) DK Line Note Play?
MLB STL @ PIT 12:35 PM PIT -163 (1.61) / O-U 7.0 PNC live feed reads 11 mph, In From CF, 61°F — and Skenes' ERA sits 0.64 above his xERA while Leahy's sits 1.24 below his. #1, #2
MLB MIA @ CHC 2:20 PM CHC -205 (1.48) / O-U 7.5 Wrigley wind from ~60° at 8.1 mph = blowing in (7.72-run park version), but Boyd (−0.669) and Phillips (−1.212) have both been outpitching their contact quality. #4
MLB MIL @ PHI 6:05 PM PHI -136 (1.73) / O-U 8.0 Painter's ERA−xERA gap of +1.084 is the biggest buy signal on the board, but only one source confirms today's probables — H3 blocks it. No — H3 unmet
MLB CLE @ BOS 6:45 PM BOS -131 (1.76) / O-U 7.0 Espino absent from the qualified xERA leaderboard, so career starts are unverified and H4 exposure is unresolved. No — H4 unresolved
MLB TB @ NYY 7:05 PM NYY -155 (1.64) / O-U 6.5 Schlittler's 1.94 ERA overstates him by 0.91 of xERA and Seymour's 4.38 understates him by 0.73 — the two cancel, and H3 is unmet anyway. No — H3 unmet
MLB CIN @ ATL 7:15 PM ATL -225 (1.44) / O-U 7.5 Mahle (−0.077) and Singer (−0.061) are both dead-on their xERA, so there is no mispricing to find. No — no signal
MLB ARI @ COL 3:10 PM not priced Coors Field. No — H1 Coors
MLB LAA @ SEA 9:40 PM SEA -207 (1.48) / O-U 7.0 Woo (+0.318) and G. Rodriguez (+0.377) both mildly unlucky, pointing under, but T-Mobile wind was not obtained and H3 is unmet. No — H3 unmet
MLB SD @ LAD 10:10 PM LAD -172 (1.58) / O-U 7.5 Glasnow's +0.518 gap favours LAD, but 1.58 already implies 63.3% and Pivetta is missing from the xERA leaderboard. No — H3 unmet
NFL ATL @ GB 8:15 PM GB -7.5 (1.90) / O-U 46.5 Atlanta is scoring 13.0 PPG on 263 yards/game against Green Bay's 22.0 on 453 — a combined pace of 35 into a 46.5 line. #3
MLB CHW @ KC 2:10 PM no DK price returned Dobnak's −1.548 gap is the strongest fade signal on the slate and I could not price it. No — unpriced
MLB NYM @ TEX 2:35 PM no DK price returned Thornton −1.227, Rocker −0.131; unpriceable. No — unpriced
MLB HOU @ ATH 9:40 PM no DK price returned Barnett's +1.759 gap is headline-sized and I could not price it. No — unpriced

Unanalysed: 4 of 13. Coors (H1, excluded before analysis) plus three games — CHW @ KC, NYM @ TEX, HOU @ ATH — where the ESPN DraftKings odds endpoint returned count: 0 on repeated calls at both default and limit=50. The playbook forbids pricing from memory, so they were dropped rather than guessed.

Section 3: Today's Plays

# Play Sport Market Odds Confidence Calibrated Edge %
1 Pittsburgh Pirates ML (vs STL) MLB Moneyline 1.61 66% 63% +0.9%
2 Under 7.0 — STL @ PIT MLB Over/Under 1.86 57% 54% +0.2%
3 Under 46.5 — ATL @ GB NFL Over/Under 1.90 56% 52% −0.6%
4 Over 7.5 — MIA @ CHC MLB Over/Under 1.90 55% 51% −1.6%

Negative edge flags: #3 (−0.6%) and #4 (−1.6%) both price worse than my calibrated estimate. They are listed because ranking is by Calibrated Confidence and hit rate is the objective, not because the number is good.

Correlation note: #1 and #2 are the same game. They are alternatives, not a pair.

All prices DraftKings, fetched this run. No play is outside 1.50–2.10.

Section 4: Reasoning

#1 — Pittsburgh Pirates ML, 1.61

This is the cleanest two-sided read the xERA method produces (MLB §2, [MODERATE] — the playbook is explicit that xERA is the best obtainable estimator, not the best one, and that its direction is better established than its strength). Paul Skenes carries ERA 3.91 against xERA 3.27, a gap of +0.641 — his results are worse than his contact quality, so the market is fading him on outcomes. Kyle Leahy runs the other way: ERA 3.65, xERA 4.89, gap −1.238, comfortably past the playbook's 1.00 flag threshold, meaning his respectable ERA is concealing genuinely poor pitching. Skenes unlucky and Leahy lucky in the same game is the exact configuration §2 is built to find. The line has not moved — PIT opened −164 and sits −163, so no sharp money disagrees, and I am not late. Supporting, not carrying: St. Louis is 77-81 and eliminated, and §6 flags September lineups as the thing to distrust in exactly this spot. The risk that beats me: I never saw either lineup card. St. Louis just beat Pittsburgh 5-1 last night off the same rotation, and one run of xERA does not survive Skenes walking four in the third. Source: statsapi probablePitcher hydrate plus ESPN's scoreboard probableStartingPitcher block, and the Baseball Savant expected-statistics leaderboard.

#2 — Under 7.0, STL @ PIT, 1.86

MLB §4 sets the order of operations for a total as park → wind → pitchers → bullpen → umpire, and the first two both point down hard here. PNC Park's live game feed — the playbook's preferred wind source because it is observed rather than forecast — returned "11 mph, In From CF", 61°F, Sunny. Eleven miles an hour straight in from centre in 61-degree air is a meaningful suppressant at a park that already plays large, and it is a 12:35 PM first pitch, so this reading is current rather than a projection. Skenes' xERA of 3.27 is the pitching half. [STRONG] on the wind mechanism (§4 treats wind as the volatile, underpriced input), [MODERATE] on the xERA half. The total has not moved off 7.0 from open, which means the market has not repriced the wind — that is the whole case. The risk that beats me: Leahy's −1.238 gap says Pittsburgh should hit him, and a four-run Pirates inning clears a 7.0 total on its own regardless of what the flags are doing. I also could not obtain either bullpen's usage from last night, which §3 calls the most underpriced factor available — this under is being made without the factor that most reliably decides late-inning run scoring.

#3 — Under 46.5, ATL @ GB, 1.90

NFL §2's fallback metric — points per game, since EPA is confirmed unobtainable — has Atlanta at 13.0 points on 263 yards per game and Green Bay at 22.0 on 453. That is a combined current pace of roughly 35 into a 46.5 total, and the yardage gap is not a rounding artefact: 190 yards a game is close to the largest offensive separation you will see between two teams on the same field. Both sides are on a short week, which §5 tags [MODERATE] as the most defensible situational factor in the sport. H3 is satisfied — MET Norway puts Lambeau at 2.5 m/s (5.6 mph) at kickoff, so wind is a non-factor in either direction and H2 does not apply. The line is entirely static: spread 7.5 open and current, total 46.5 open and current, which under H1(b) is weak evidence that no quarterback news has landed that I am missing. The risk that beats me: I could only obtain points scored, not points allowed, so this is half a total estimate rather than a whole one — Atlanta's 13 PPG could sit opposite a Green Bay defence that has been shredded, and I would not know. Sample is also two or three games. Quarterbacks are presumptive starters — whoever started last week; no current injury feed was available, and inactives drop after this report is written.

#4 — Over 7.5, MIA @ CHC, 1.90

H2 is satisfied and this is what the data says. MET Norway puts Wrigley at 3.6 m/s (8.1 mph) from ~60° at the 2:20 PM first pitch; against the park's 31° centre-field bearing that is wind blowing in, which historically produces the 7.72 runs/game version of Wrigley — [STRONG], the 410-game 2021-25 sample. DraftKings has the total at 7.5, marginally below that wind-in mean, and the wind is under the 10 mph threshold where the gap widens. On top of that, both starters have been outperforming their contact quality and should regress worse: Matthew Boyd ERA 4.06 / xERA 4.73 (−0.669) and Tyler Phillips ERA 3.45 / xERA 4.66 (−1.212, past the flag threshold). Total unmoved from 7.5 at open. The risk that beats me: this is a thin edge stacked on a historical mean, the wind-in figure is a league-wide average rather than a forecast for these nine innings, and the statsapi cross-check the playbook prefers returned "weather":{} — so the out/in call rests on my own degree conversion, not on an observed reading. If 60° is really crosswise rather than in, the whole frame changes.

Sources

Research limitations

The three counts: 9 of 13 games analysed · 31 sources fetched · 9 games where a Tier-1 factor was unavailable.

Every analysed game has confirmed probable pitchers (or presumptive QBs, for the NFL) plus at least one Tier-1 factor carrying a real number. But a Tier-1 factor was missing in all nine, detailed below.

Playbooks read: mlb.md (v1, 2026-09-16) and nfl.md (v1, 2026-09-20). No stub playbook was used — NHL and NBA have no games today. No Hard Rule was violated; H1, H3 and H4 each blocked plays, which is the rule working rather than a gap.

Step 1 freshness check: verified against statsapi schedule at 11:12 ET — 12 MLB games, and against the ESPN NFL events endpoint — 1 game (ATL @ GB, week 3, Lambeau Field). Every starter named in this report came from that call. ESPN's scoreboard independently confirmed Skenes and Leahy, which is what makes #1 and #2 H3-compliant.

H3 cost six playable games. The ESPN scoreboard payload populated probableStartingPitcher for only the first five games of the day. site.api.espn.com/.../mlb/summary?event=... contained no probables block at all (tested on 401817062 and 401817069 — greps for probableStartingPitcher, Painter, Wilson, Woo and Rodriguez all returned nothing), and .../competitions/{id}/competitors/{id}/probables returned an empty body on both competitors. So MIL @ PHI, CLE @ BOS, TB @ NYY, LAA @ SEA and SD @ LAD were all priced and all had a usable xERA read, and none of them could clear a Hard Rule that requires two independent confirmations. The single strongest signal on the board went unplayed for this reason — Andrew Painter at +1.084.

Three games returned no DraftKings price. count: 0 from the odds endpoint on events 401817065 (CHW @ KC), 401817067 (NYM @ TEX) and 401817070 (HOU @ ATH), at both default page size and limit=50. Note that 401817061 and 401817062 returned count: 0 on the first call and a full board on retry with limit=50 — so this endpoint is intermittently empty rather than reliably so, and a retry is worth making. The three above failed both times. Two of them carried the largest xERA gaps of the day (Dobnak −1.548 and Barnett +1.759) and could not be acted on.

Bullpen state — unobtained for all nine games. This is the big one, and MLB §3 calls it the most underpriced factor available. statsapi's boxscore endpoint does serve the data: https://statsapi.mlb.com/api/v1/game/{gamePk}/boxscore?fields=teams,away,home,team,name,players,person,fullName,stats,pitching,inningsPitched,numberOfPitches,gamesStarted returned ~76k chars for both games tested (823410 and 823327). The single-game innings figures are all present and greppable in isolation (5.0, 3.1, 1.0, 3.0, 2.0 …), but attributing them to pitchers failed: the fields filter does not suppress the parallel seasonStats.pitching block, so the gap between a fullName and its own game inningsPitched exceeds what a bounded Grep pattern can span, and the saved file sits outside the sandbox mounts so bash cannot slice it. This is a solvable problem and worth a playbook entry — the likely fix is requesting gamePk + fields=teams,home,away,pitchers to get the ordered pitcher-ID list, then matching it against the IP sequence.

Umpire assignments: not attempted. MLB §5 values them at 0.3–0.5 runs/game [STRONG], and §9 lists the source as unverified. With two totals on the card this is a real omission, not a trivial one.

Wind obtained for 2 of 7 relevant venues. PNC Park (observed, via the live feed: 11 mph in from CF) and Wrigley (forecast, via MET Norway) — plus Lambeau for the NFL. Not obtained: Fenway, Yankee Stadium, Truist, T-Mobile, Dodger Stadium. Those five games were no-plays for other reasons, so nothing was priced blind, but had H3 cleared I would have needed them. The statsapi cross-check on Wrigley returned "weather":{} as the playbook predicts for an afternoon game queried at midday, so the out/in call there is my own degree conversion rather than an observed reading.

xERA missing for four of today's starters: David Sandlin, Bryse Wilson, Daniel Espino and Nick Pivetta are all absent from the min=q leaderboard. Per §9 I am reporting that rather than substituting FIP. Espino's absence is what left H4 unresolved on CLE @ BOS.

NFL turnover differential is unusable. The §2 fallback's key metric came back as totalGiveaways: 2.0 and totalTakeaways: 1.0 for both Green Bay and Atlanta — identical values across two teams, which cannot be right for a season-to-date figure. I have treated it as bad data and not used it. totalPointsPerGame and yardsPerGame looked sane and were used. Points allowed per game was not located in the response, which is why #3's total estimate is only half-built.

Calibration caveat, restated: the 3- and 4-point confidence haircuts were measured on a mostly-prop sample and have little history on game lines. Treat all four Calibrated figures as a starting assumption.

Cutoff: finished well inside 11:50 — End 11:19. Parallel fetching moved faster than expected; the run was limited by what the endpoints would serve, not by time.

Recent Reports
Sep 23 Wednesday
Read →
Sep 22 Tuesday
Read →
Sep 21 Monday
Read →
Sep 20 Sunday
Read →
Sep 19 Saturday
Read →
View all reports →
Betting Tracker · Generated 2026-09-24 11:21 · Personal use only