The Board Was Full At 11 AM. By First Pitch, It Was Empty.
At 11:01 this morning our line check logged 48 real FanDuel markets, 24 pitcher strikeout lines and 24 outs recorded lines, spread across 13 of the day's 15 games. Paul Skenes was on it. Yoshinobu Yamamoto was on it. Dylan Cease carried the highest number on the whole sheet. At 4:18 this afternoon, when the model actually ran for July 25, the same fetch came back empty. Not thin. Not one or two survivors after the filter. Zero pitcher prop markets, on any of the 27 events FanDuel had listed for today and tomorrow combined.
Two Reads, Six Hours Apart
Every morning before the model runs, we log whatever FanDuel has posted for the day's pitcher strikeout and outs recorded markets, mostly so a real-line archive keeps building for future backtests. This morning that log came back full. Thirteen of the day's fifteen games carried live two-way pricing, missing only Royals at Tigers and Athletics at Twins, which simply had not been posted yet. Everything else was there. Ryan Feltner at 4.5 strikeouts against the Brewers. Robert Gasser answering at 5.5 on the other side of the same game. Shota Imanaga at 5.5 opposite Paul Skenes at 6.5 in Pittsburgh. Yoshinobu Yamamoto and Nolan McLean both at 6.5 in the same Dodgers-Mets game, a rare split where neither team's starter carried the lower number.
Then the day moved on, the way it always does. Some of today's games were early ones. The Padres finished up in Miami. The Diamondbacks and Nationals wrapped in Washington. By the time our board actually runs each afternoon, most of the day's early games are usually done and the evening slate is usually still hours from first pitch, still sitting there waiting to be bet.
Not today. At 4:18 the board made its live call to FanDuel for the same 27 events it could see, today's remaining games and a first look at tomorrow's slate. It came back with an empty dictionary. No strikeout markets. No outs recorded markets. Not for Skenes, not for Yamamoto, not for anyone.
What The Morning Sheet Actually Looked Like
It is worth showing the full sheet rather than describing it, because "the market disappeared" is an easy thing to write and a specific line sheet is the only way to prove it was ever real. Here is every pitcher strikeout market that existed on FanDuel at 11:01 AM this morning.
The full morning sheet had some real texture to it, not just a wall of numbers. Dylan Cease carried the highest strikeout line on the board at 7.5, priced close to even money on both sides, -111 over and -115 under, ahead of Sonny Gray in the same game at just 4.5. Hunter Brown and Sean Burke were both lined at 5.5 in the Astros-White Sox game, and FanDuel leaned different directions on each of them at the same time, -132 to back Brown's over and -102 to back Burke's over, meaning the book liked Brown to clear his number more than it liked Burke to clear his, even though the numbers matched. Mitch Bratt and Foster Griffin, the least established arms on the sheet, both sat at pick-em pricing, -113 on both sides for Bratt at 3.5.
Where It Actually Went
Here is the honest limit of what we can tell you. We do not know, from outside FanDuel's system, why a full morning sheet of pitcher strikeout and outs recorded markets across 13 games came back empty nine hours later on the same slate. Sportsbooks pull markets and repost them throughout the day for reasons that rarely make it outside the building, whether that is a trading decision, a technical hold, or something as mundane as a batch of markets waiting to be refreshed. What we can tell you is only what we measured directly: real two-way pricing existed at 11:01 AM, and the identical query against the identical page returned nothing at 4:18 PM.
We also checked the one thing we could verify independently, which is where today's games actually stood. A live read of MLB's schedule feed shows five of today's 15 games, the earliest starts of the day, already final, and the other ten in progress. That tells you the day had moved on. It does not, on its own, explain why markets for games that had not yet reached first pitch were also missing when our board ran. We are not going to dress up a guess as a finding here. The gap is real and it is documented above. The cause is not something we can verify, so it does not get a confident sentence here.
Why The Model Does Not Mind An Empty Board
Our props model does not run on its own opinion. It blends a stats-based projection against the de-vigged market price, weighted 0.50 toward the model for strikeouts and only 0.15 for outs recorded, because outs recorded is the noisier market of the two. A pick only gets flagged when that blended number disagrees with the market by at least 2.5 percentage points for a lean or 4 points for a strong play.
The reason the market gets that much weight is not modesty. It is the model's own track record against itself. Its backtest, a 2024 and 2025 walk-forward test plus a live-line replay of 193 real FanDuel contracts from May 24 through May 31 of this year, showed the de-vigged market beating a stats-only projection on Brier score, 0.246 for the market against 0.261 and 0.266 for the two stats-only versions. Neither approach, market-anchored or stats-only, produced a positive return once real vig was applied. That is not a hidden detail. It is written into the model's own documentation as a reason to stay anchored to the market rather than trust the projection alone.
The Full Line Sheet, For The Record
| Game | Pitcher | K Line | Odds (O/U) | Status at 4:18 PM |
|---|---|---|---|---|
| Rockies at Brewers | Ryan Feltner / Robert Gasser | 4.5 / 5.5 | +126/-162 · +108/-138 | Market gone |
| Yankees at Phillies | Ryan Weathers | 6.5 | +128/-164 | Market gone |
| Dodgers at Mets | Yamamoto / McLean | 6.5 / 6.5 | +112/-142 · +116/-148 | Market gone |
| Cubs at Pirates | Imanaga / Skenes | 5.5 / 6.5 | -118/-108 · -148/+116 | Market gone |
| Reds at Cardinals | Greene / Pallante | 6.5 / 4.5 | +102/-130 · +118/-150 | Market gone |
| Diamondbacks at Nationals | Bratt / Griffin | 3.5 / 4.5 | -113/-113 · -128/+100 | Market gone |
| Mariners at Rangers | Woo / Eovaldi | 5.5 / 6.5 | -142/+112 · +112/-142 | Market gone |
| Braves at Orioles | Elder / Young | 4.5 / 4.5 | -136/+106 · +118/-150 | Market gone |
| Padres at Marlins | Eury Perez | 5.5 | -162/+126 | Market gone |
| Astros at White Sox | Brown / Burke | 5.5 / 5.5 | -132/+104 · -102/-125 | Market gone |
| Guardians at Rays | Bibee / Martinez | 4.5 / 3.5 | +132/-170 · +114/-146 | Market gone |
| Blue Jays at Red Sox | Cease / Gray | 7.5 / 4.5 | -111/-115 · -148/+116 | Market gone |
| Angels at Giants | Johnson / Ray | 3.5 / 5.5 | -122/-104 · -132/+104 | Market gone |
Everything on this page is analysis only. There is no tracked pick for July 25 and no units are staked, so the season record is untouched. The morning lines above are historical, not live prices, and none of them should be bet off this page.
The Rule This Costs Us
It would be simple to just run the model against the 11:01 AM lines and publish whatever came out. Nobody reading this page would necessarily know the difference. That is exactly the reason not to do it. A line from six hours before publication is not a price anyone can still get, and writing analysis against a number that no longer exists on the board is a smaller version of the same dishonesty as publishing a pick for a game that has already started. Both dress up hindsight as a projection.
So the rule holds. No market, no grade, no play. It cost us a headline today. Paul Skenes at 6.5, Dylan Cease at 7.5, a Yamamoto-McLean strikeout mirror in the same game, any one of those would have made an easy write-up if we were willing to pretend the morning line was still live. We are not.
Final Verdict
No play on July 25. The market that fed this model existed, in full, at 11:01 this morning, 48 real lines across 13 games. By the time the board ran, it did not. We do not know why, and we are not going to guess in print. What we can verify is that the gap happened, that it happened after a morning where the same fetch worked, and that the honest response to a missing market is an empty board rather than a manufactured one.
Back tomorrow with a full slate, and, with any luck, a market that is still there when the board runs.
Frequently Asked Questions
Because there was nothing to evaluate. The 11:01 AM line check logged 48 real FanDuel strikeout and outs recorded markets across 13 of the day's 15 games. When the model actually ran at 4:18 that afternoon, the same fetch across all 27 events FanDuel had listed came back with zero pitcher prop markets posted anywhere. With no market price to blend against, the model has nothing to grade.
A full sheet. Paul Skenes at 6.5 strikeouts, -148, opposite Shota Imanaga at 5.5. Yoshinobu Yamamoto and Nolan McLean both at 6.5 in the same Dodgers-Mets game. Dylan Cease carried the highest number on the board at 7.5. Hunter Brown and Sean Burke were both priced at 5.5 in the same game, with the market favoring each pitcher's over and under in opposite directions. Thirteen games, 24 strikeout lines, 24 outs recorded lines, all gone by the afternoon.
No. The model anchors to the market because its own backtest, a 193-contract live-line replay in May 2026, showed the de-vigged market beating a stats-only projection on Brier score, 0.246 against 0.261 and 0.266. With no market price available today, there was nothing to disagree with, which is the model doing what it is designed to do rather than failing.
Because a line from six hours before the board runs is not a live price. Sportsbooks pull and repost prop markets for reasons that are not visible from outside the book, and the honest response to a market disappearing is to report that it disappeared, not to quietly keep using the version that is convenient.