Four WNBA games tonight, verified against the schedule before anything else: Los Angeles at Connecticut (7:00 PM ET), Indiana at Toronto (7:00 PM ET), Atlanta at Las Vegas (9:00 PM ET) and New York at Chicago (9:00 PM ET). Yesterday's post flagged three of these four as appearing on a board dated the day before they were played; today they are genuinely today.
The three largest edges on the board are all dropped, and all for the same reason. Not a role-change problem like yesterday — a defensive one. Each of the three biggest numbers asks a player to do something against an opponent that is good at stopping exactly that thing, or asks for an Under against an opponent that is bad at stopping it. That is what the matchup layer is for, and today it fired three times before we got to a publishable leg.
The Card
Prices, hit-rate windows and positional defensive ranks captured from our props screen (our premium data platform) Tue Aug 18 · 8:15–8:35 AM ET. Implied probability and Edge Gap are computed in code from the odds, never typed by hand. Implied probability includes the vig, so Edge Gap is a ranking tool, not a profit projection.
Read this before the card. All three published legs are priced at a single book, and two of the three are first-half markets rather than full-game ones. A one-book line is the most fragile kind of price we publish — there is no second quote to sanity-check it against, and it can move or vanish without warning. Two of the three also carry our data platform's standing "team injuries may impact playing time" flag. Shop your own number and confirm the player is starting.
The reasoning
Jonquel Jones — Over 13.5 First-Half Pts+Reb+Ast (−119). The largest edge that survives the gate. 100% over her last five, 80% over her last ten, 85% over her last twenty, 64% on the season, against a price implying 54%. Chicago ranks 12th in the league at defending this category, which for an Over is direct support — this is the clearest Confirms on the board. The weakness is the head-to-head number: 20%. We treat head-to-head as a tiebreaker only and it is a small sample, but 20% is low enough that we are printing it rather than burying it. Her 2025 rate of 35% is also a long way below her 2026 rate of 64%, which is the shape of a player whose role or health improved — the good version of the pattern that killed both of yesterday's candidates, but the same pattern.
Brittney Griner — Over 9.5 First-Half Pts+Reb+Ast (−119). 80% last five, 80% last ten, 65% last twenty, 72% season. Los Angeles ranks 14th — last — at defending this category. That is the strongest possible matchup support for an Over, and it is why a leg with a middling last-twenty number is on this card. Head-to-head is 33% and again we are not weighting it. The flat-to-declining window shape (72 season, 65 over twenty, 80 over ten, 80 over five) is stable rather than streaky, which is what we want to see after yesterday.
Isabelle Harrison — Under 19.5 Points+Rebounds (−119). 100% last five, 80% last ten, 70% last twenty, 67% season. Indiana ranks 9th at defending this — mid-table, which is neither support nor conflict, so this is a genuine Neutral rather than a Neutral by default. It is the only full-game market on the card. Her 2025 rate of 97% against a 2026 rate of 67% is a thirty-point drop year over year, and we do not have an explanation for it.
The three we dropped
1. Caitlin Clark Under 10.5 Assists (−148) — the biggest number on the board, roughly a +26 gap, dropped twice over. She clears this in 90% of her season, 85% of her last twenty and 80% of each of the last ten and five. Toronto ranks 13th at defending this category, meaning they are among the league's worst at limiting it — the wrong direction for an Under. That alone is a conflict.
The second reason is the one we wrote into our own rules yesterday: when our number and the market's disagree by thirty points on a household name, the likely explanation is that we are misreading something. A 90% season rate priced at 60% is exactly that disagreement. Her head-to-head figure is 0% — she has gone over this line in every recorded meeting with Toronto. We could not reconcile it, so we did not publish it.
The second reason is the one we wrote into our own rules yesterday: when our number and the market's disagree by thirty points on a household name, the likely explanation is that we are misreading something. A 90% season rate priced at 60% is exactly that disagreement. Her head-to-head figure is 0% — she has gone over this line in every recorded meeting with Toronto. We could not reconcile it, so we did not publish it.
2. Caitlin Clark Under 4.5 Rebounds (−148) — roughly a +23 gap, dropped on the same defensive rank. 100% last five, 90% last ten, 85% last twenty, 74% season — a genuinely strong profile, and this exact leg won for us on August 16 against New York. Tonight the opponent is different: Toronto ranks 12th at limiting rebounds, which means more rebounds available, which is the wrong side of an Under. Head-to-head is again 0%.
We are being deliberately consistent here. On August 16 we published this leg because New York ranked 3rd defensively and that supported the Under. If the rank is the reason to take it at 3rd, the rank has to be the reason to pass at 12th. Publishing it both ways would mean the layer is decoration.
We are being deliberately consistent here. On August 16 we published this leg because New York ranked 3rd defensively and that supported the Under. If the rank is the reason to take it at 3rd, the rank has to be the reason to pass at 12th. Publishing it both ways would mean the layer is decoration.
3. Jackie Young Over 28.5 Points+Rebounds+Assists (−114) — roughly a +22 gap, dropped on the opposite problem. 80% last five, 90% last ten, 75% last twenty, 67% season, at near-even money. Atlanta ranks 3rd in the league at defending this category — a top-three defence is the strongest conflict an Over can draw. Her Over 22.5 Points+Rebounds (+19.3) and three first-half Young props are dropped for the same reason. Head-to-head is 40%.
We checked the opposite side, as our own rules require. If Atlanta's third-ranked defence conflicts with a Young Over, a Young Under should confirm — that is precisely how we found Lauren Betts's Over 9.5 Pts+Reb+Ast after dropping her Under. Tonight there was no Under priced on Young in any of these categories at a number we could read, and we are not going to invent one. The angle exists; the market for it did not.
We checked the opposite side, as our own rules require. If Atlanta's third-ranked defence conflicts with a Young Over, a Young Under should confirm — that is precisely how we found Lauren Betts's Over 9.5 Pts+Reb+Ast after dropping her Under. Tonight there was no Under priced on Young in any of these categories at a number we could read, and we are not going to invent one. The angle exists; the market for it did not.
The strongest argument against our #1 pick
Jonquel Jones's leg rests on a first-half line at one book, and the head-to-head number is a real warning we are choosing to ignore.
Start with the market. A single sportsbook quoting −119 on a first-half combined-stat prop is a thin, low-limit market. Thin markets are where stale numbers live — and a stale number is exactly what an Edge Gap screen is designed to find, which means we cannot tell the difference between "the book has not updated" and "the book is right and our history is wrong." With a full-game line at four books we would have a consensus to check against. Here we have one quote and our own arithmetic.
Then the number we are downplaying: 20% head-to-head. We say, correctly, that head-to-head is a small sample and a tiebreaker. But 20% against a specific opponent, in a leg whose entire case is that this specific opponent defends the category badly, is the one place head-to-head bears directly on the argument. If Chicago is 12th overall but has some structural answer for Jones in particular, the head-to-head number is where that would show up, and it is showing up. We do not think that is what it means. We are not certain.
Finally: it is a first-half prop, which means an early foul, a slow start or a coach resting her for four minutes of the second quarter breaks it in a way a full-game line would absorb.
Start with the market. A single sportsbook quoting −119 on a first-half combined-stat prop is a thin, low-limit market. Thin markets are where stale numbers live — and a stale number is exactly what an Edge Gap screen is designed to find, which means we cannot tell the difference between "the book has not updated" and "the book is right and our history is wrong." With a full-game line at four books we would have a consensus to check against. Here we have one quote and our own arithmetic.
Then the number we are downplaying: 20% head-to-head. We say, correctly, that head-to-head is a small sample and a tiebreaker. But 20% against a specific opponent, in a leg whose entire case is that this specific opponent defends the category badly, is the one place head-to-head bears directly on the argument. If Chicago is 12th overall but has some structural answer for Jones in particular, the head-to-head number is where that would show up, and it is showing up. We do not think that is what it means. We are not certain.
Finally: it is a first-half prop, which means an early foul, a slow start or a coach resting her for four minutes of the second quarter breaks it in a way a full-game line would absorb.
What we did not run today
No projections, so no three-signal test. We did not pull our lineup service's daily WNBA projections this morning, which means no Projection Gap and no independent model to agree or disagree with the two signals we do have. All three legs are two-thirds checked.
No slate-wide injury sweep, and this matters more than usual tonight. our data platform's own board carries a standing "team injuries may impact playing time" flag on several of the rows we used, and yesterday we learned the hard way that a hit-rate window built on one role does not transfer to another. We did not independently confirm minutes or starting status for any of the three players on this card. That is the largest uncontrolled variable here and it is the same one that removed both of yesterday's candidates.
No component breakdown on the combo legs. Two of the three are combined-stat props and we rated them against a single positional defensive rank rather than breaking the rank down by component. A team that defends points well and rebounds badly is not the same as a team that is uniformly 12th, and we did not check which kind Chicago and Los Angeles are.
No slate-wide injury sweep, and this matters more than usual tonight. our data platform's own board carries a standing "team injuries may impact playing time" flag on several of the rows we used, and yesterday we learned the hard way that a hit-rate window built on one role does not transfer to another. We did not independently confirm minutes or starting status for any of the three players on this card. That is the largest uncontrolled variable here and it is the same one that removed both of yesterday's candidates.
No component breakdown on the combo legs. Two of the three are combined-stat props and we rated them against a single positional defensive rank rather than breaking the rank down by component. A team that defends points well and rebounds badly is not the same as a team that is uniformly 12th, and we did not check which kind Chicago and Los Angeles are.
Fewer picks is always acceptable. Nothing on this page is a recommendation to bet.
Sources: hit-rate windows, opponent positional defensive ranks and prices from our props screen (our premium data platform), captured Tue Aug 18 8:15–8:35 AM ET. Schedule verified against ESPN.