The Closing Line Report: August 2026
August lost money again: our straight picks went 32-41-1, -13.34 units and a -18.3% return at flat stakes, even though 40 of the 71 we could measure (56.3%) beat the closing line. One league did most of the damage. MLB went 4-11 in August and 28-51-2 since May, and on September 22 we took it off our card; the deep dive below shows why, and what we found wrong in our own data while checking.
Correction note, September 29, 2026: three August picks, two tennis and one MLS, had stored closing lines captured at or after their game’s start, and we replaced each with the last snapshot of that exact game before its start (Correction 008 on our methodology page). August’s beat-the-close rate stays 56.3% (40/71), and average closing line value is +1.34% rather than the +1.32% printed below; in the league and bet-type tables, tennis moves from +2.00% to +2.05% and moneylines from +0.78% to +0.81%. The same correction reached ten July tennis picks, two of which moved from missing the close to beating it, so season to date 120 of the 238 tracked bets beat the close rather than 118 (still 50%), with average closing line value -0.27% rather than -0.57%. In the deep dive’s table of every graded straight bet since May 11, every other league moves from 54.6% [47.8, 61.3] (112/205) to 55.6% [48.8, 62.2] (114/205) and all straight bets from 51.1% [45.1, 57.1] (134/262) to 51.9% [45.9, 57.9] (136/262); MLB is unchanged. The win-loss record and return did not change, and the rest of this issue is left exactly as published.
The Closing Line Report is our monthly self-audit. Every number below comes from the same public ledger behind our live track record, and the market data comes from the odds archive we capture ourselves. We publish the month whether it flatters us or not.
The Scoreboard: August 2026
In August 2026 our straight single-game picks went 32-41-1 (-13.34u at a flat one-unit stake) and we beat the closing line 56% of the time (40/71 tracked), with average closing line value +1.32%. ROI for the month was -18.3% across 74 graded picks. On samples this size ROI is mostly variance, so closing line value is the number we hold ourselves to.
That beat rate carries a 95% confidence range of 44.8% to 67.3%, so one month proves nothing in either direction.
Season to date through August 2026: 173-174-5 (-33.48u, ROI -9.7%) across 352 straight picks, and we beat the closing line 50% of the time (118/238 tracked), with average closing line value -0.57%.
Results by League
WNBA was the month’s best league: 8-4 (+3.32u), with 9 of 12 tracked picks beating the close. MLB was the drag again at 4-11 (-7.81u), even with 8 of 15 beating the close, and the two NFL losses are preseason games (see Data Notes).
Results by Bet Type
Spreads carried the month at 19-12 (+5.64u), led by tennis game handicaps (9-5) and WNBA point spreads (8-3). Moneylines went 13-29-1 for -18.98 units.
Props and Parlays
Tracked separately from the headline: featured player props went 6-5-0 (-0.20u) and we beat the closing line 46% of the time (5/11 tracked), with average closing line value -1.07%; and parlays went 1-3-0 (-0.46u), settled all-or-nothing at combined odds; they are high-variance by construction and never blended into the headline record.
Best and Worst Calls of the Month
Ranked by closing line value, not by result: the market’s verdict on our number, before the game decided anything. This month we ranked only values we could verify twice, against the stored closing line and against the pre-game close rebuilt from our odds archive, because the largest raw value compared prices across two different handicap numbers (see Data Notes).
Atlanta drifted from our +125 to +111 before first pitch, which came nearly two hours late after a rain delay, so the market moved to our side before a ball was thrown. The Braves took a 2-0 lead into the bottom of the ninth and lost 3-2 in 10 innings: a number that aged well on a ticket that lost anyway.
We took Clara Tauson at +198 against Linda Noskova in Cincinnati, the price lengthened to +220 by the start, and Noskova won 7-6(3), 6-2. For the second issue running, our worst call of the month was a bet against Noskova, who also beat our July pick in the Wimbledon final.
The Deep Dive: Why We Benched MLB
We stopped publishing MLB picks on September 22 because the league lost on results and on closing line value at the same time, for long enough that bad luck stopped being a credible explanation. From May 11 to September 1 our MLB straight bets went 28-51-2 (-26.78 units, -33.9% ROI), and 38.6% of the tracked ones beat the closing line (22/57), against 54.6% (112/205) for every other league we cover.
MLB made up 22% of our straight bets and 71% of the net loss. The clearest way to see the gap is to compare how often we won with how often our own prices needed us to win: every price implies a break-even rate, and across 79 decided MLB bets ours averaged 52.7%. We won 35.4%. In every other league the two numbers sat 1.3 points apart, 53.3% won against 54.6% needed.
The losses were concentrated. Moneyline underdogs were the worst of it, 2-11 at prices that needed us to win 44.8% of the time, and totals went 4-11-2. The closing line reads those segments differently: 5 of 12 underdog prices still beat the close, so part of that damage was the results, while totals (1 of 6) and the run line (3 of 9) were where our numbers fared worst.
Two of those segments had already been cut before the bench. Here is the order we pulled MLB back. The last two columns show our automatic check, which reads each league’s last 25 graded straight bets and flags any league where fewer than 40% of the tracked ones beat the close: first as it read that day, then as it reads on today’s repaired closing lines (the repair is explained below).
The repair rewrites part of that timeline. On August 24 the check read 37.5% and we chose to keep one MLB pick a day and watch; on today’s closes it reads 40.0% and stays quiet, while on July 29 it would have flagged at 33.3%, nearly four weeks earlier than it did. The bad captures had flattered MLB in July and hurt it in August. The September 22 reading does not hold up on today’s closes. Two of the 25 picks behind it had no closing line until September 28, when we rebuilt both from our archive (see Data Notes); one beat the close and one did not, which moves the reading from 39.1% (9/23) to 40.0% (10/25), exactly on the line, so the check would not have flagged. The season-long gap in the first table is wider than it was.
Before publishing a verdict built on closing lines, we checked the closing lines, and found a fault. Twenty-four of the 60 stored MLB closes had been captured after first pitch. In all 19 we could check against our archive, the same two teams played again within 30 hours, and our capture log shows why: after a game ended, the capture job matched the teams by name and stored the next game’s line. The capture was fixed on September 22; the 19 were rebuilt from our archive’s last pre-game snapshot of the exact game, and 5 June closes that predate the archive were voided.
The verdict survived the repair. MLB’s beat rate moved from 38.3% to 38.2% (21/55), and to 38.6% (22/57) with the two September 28 rebuilds; a separate rebuild from our archive puts it at 36.8% (21/57), so the two measures now sit 1.8 points apart. Eleven MLB verdicts flipped in the repair, 6 from beat to miss and 5 from miss to beat: the bad captures blurred the measurement without tilting it.
Timing does not explain it either, though it was our first suspect after Issue No. 2 found MLB lines moving against us after we published more often than toward us. Our MLB picks went live a median 12.2 hours before first pitch, in line with the rest of the card (12.4 hours), and the ones published 6 to 12 hours out beat the rebuilt close about as often as those published earlier (40.0%, 10/25, against 36.7%, 11/30). After the August 8 changes MLB went 3-5, with 4 of 8 stored closes and 4 of 8 rebuilt closes beaten: too few bets to say whether the changes helped.
MLB player props, which we track separately from the record above, went 17-10 (+5.28 units) with 12 of 25 beating the close. Every prop rides on a main pick, so they leave the card with it.
MLB stays off our card until closing-line evidence gives us a reason to bring it back; that is an owner decision, and we will report it here when we make it. One MLB pick published before the pause, the Rays and Yankees on September 22, is still to be graded and goes in the ledger like every other. Next issue: September, the month football took over our card. Through September 21, 15 of our 24 graded straight picks this month were NFL or college football games, and those 15 stand at 5-9-1.
Data Notes and Method
This issue covers every pick graded for August 2026: 74 straight picks, 11 props, and 4 parlays. Closing lines were captured for 71 of the 74 straight picks; a pick without a verified pre-game closing line is counted in the record but excluded from closing line value.
Building this issue turned up measurement problems in our own data; each is disclosed here with what it changes. The deep dive covers every graded MLB straight bet from May 11 through September 1, 2026, re-measured against our archive from June 12, when the archive began.
- Closing lines repaired on September 22. Our capture job could store the next game between the same two teams as a pick’s closing line. We fixed the capture on September 22 and repaired the affected history from June through September, as logged in Correction 004 on our methodology page. In August six closes (five MLB, one tennis) were rebuilt from our archive’s last pre-game snapshot, and every number in this issue uses the repaired values. As an independent check, August measured against pre-game closes rebuilt entirely from our archive beats the close 55.6% of the time (40/72).
- Closes rebuilt on September 28. Five August picks had no closing line, because our capture only followed the games our selection process chose and these were picks we added by hand: two WNBA spreads, one MLB moneyline, and the two NFL preseason spreads. We rebuilt each close from our archive’s last pre-game snapshot of that exact game (Correction 006 on the same page), which moved August from 56.1% (37/66) to 56.3% (40/71). The season figure also counts July picks rebuilt the same way (Corrections 006 and 007), and two MLB picks in the deep dive gained a close this way (July 26 and August 25).
- One bet removed from closing line value. Aleksandar Rakic to win by KO, TKO or DQ at +120 had been scored against the fight’s straight moneyline, which is a different bet, so we removed it from closing line value (Correction 005 on the same page); it still counts in the record. Separately, the builder’s automatic best call, Zeynep Sonmez +4.5 games, shows +20.2% only because it compares prices across two different numbers: the line closed at +5.5, so by the number it did not beat the close. We show the largest value the stored and rebuilt closes agree on instead.
- Preseason labels. The two August rows labeled NFL are preseason games from August 27; a third preseason pick carries the NFL Preseason label.
Capture windows, the pre-game capture rule, grading definitions, and the full repair log live on the same methodology page. Book odds are captured from The Odds API; prediction-market prices, when cited, are attributed inline to their source. All market figures are the markets’ own implied numbers, with the bookmaker’s margin removed where stated; a gap between markets is a difference of opinion, nothing more.
Download the derived data behind this issue: closing-line-report-2026-08.csv. Derived aggregates only, never raw feed rows; cite it freely with a link back.
Today’s slate is on our picks page, graded and added to the ledger every morning. Past results never predict future ones.
Play Safe: Gambling should be fun, not stressful. Set limits, stick to your budget, and never chase losses. If you or someone you know has a gambling problem, call 1-800-MY-RESET or visit ncpgambling.org. For more resources, see our Responsible Gambling page.
Frequently Asked Questions
Quick answers to the questions we hear most about this report and how it is built.
What Is The Closing Line Report?
It is our monthly self-audit. We grade every betting pick we publish against the final score and the closing line, then publish the full month here: record, ROI, closing line value, and the best and worst calls, plus one study built from our own odds archive. Wins and losses both stay in the ledger permanently.
What Is Closing Line Value, and Why Do You Grade Yourselves on It?
Closing line value compares the line we published to the final pre-game line. Over a small number of picks, wins and losses are mostly variance, so whether the market moved toward our number is the more reliable long-run signal of whether the analysis holds up.
Where Does the Data in the August 2026 Report Come From?
Everything is first-party: our public graded picks ledger and the odds archive we capture ourselves. Book odds come from The Odds API and prediction-market prices are attributed inline when cited. The derived aggregates behind each issue are downloadable as a CSV.
Why Did GamblingSite.com Stop Posting MLB Picks?
Because MLB lost on results and on closing line value at the same time. Our MLB straight bets went 28-51-2 from May 11 to September 1, 2026, a -33.9% return, and 38.6% of the tracked ones beat the closing line against 54.6% in our other leagues. When our automatic check flagged the league again on September 22, we paused it, and bringing it back has to rest on closing-line evidence.