There are no picks on this page. One game, two candidates that screened enormously well, and both were removed at the last check. We publish the no-play with the same detail we would give a card, because a reader who only sees us on the days we like something has no way to judge us.
The slate
Exactly one WNBA game today: Dallas at Golden State, 10:00 PM ET at Chase Center. Our props board also listed Los Angeles at Connecticut, Indiana at Toronto and Atlanta at Las Vegas — all three are tomorrow, Tuesday Aug 18, not tonight. The board had already rolled forward to the next slate while today's game was still hours away. We check this every morning because it is the easiest way to publish a pick on a game that does not exist, and this morning it would have handed us three of them.
Caitlin Clark Under 4.5 Rebounds was on that board at a +14 gap. It is an Indiana–Toronto line for tomorrow. Had we not checked the schedule independently, that is exactly the leg that ends up on a card dated today.
The two picks we did not make
Odyssey Sims Under 15.5 Points+Rebounds+Assists (−119) — roughly a +35 Edge Gap, dropped. On paper this was the largest edge we have generated on any board, in any sport, since this site started: a blended rate near 90 against a price implying 54%. That number is built on a season in which Sims was a rotation guard averaging around 17 minutes. Reporting on her recent run describes a materially larger role — starting, and averaging in the region of 12 points, and close to three rebounds and three assists, on roughly 28 minutes a night. Add those up and her recent production sits above the line, not below it. Her own hit-rate windows agree with the reporting rather than with the season figure: 95% over the last twenty, 90% over the last ten, 80% over the last five. That is a descending gradient, and a descending gradient on an Under is the signature of exactly this problem.
A blended rate weights the season at 40%. When a player's role changes mid-season, that 40% is measuring a job she no longer has, and the Edge Gap it produces is an artefact rather than an edge. We could not establish with confidence what her minutes will be tonight. So it is a drop.
A blended rate weights the season at 40%. When a player's role changes mid-season, that 40% is measuring a job she no longer has, and the Edge Gap it produces is an artefact rather than an edge. We could not establish with confidence what her minutes will be tonight. So it is a drop.
Alanna Smith Under 12.5 Points+Assists (−111) — roughly a +34 Edge Gap, dropped. Same shape, and one additional problem we want to be honest about. Smith's season averages are in the region of five points and one assist in about fifteen minutes off the Dallas bench. A player at roughly six combined points and assists should clear an Under of 12.5 far more often than the 53% the market is pricing. When our number and the market's number disagree by that much on a well-known player, the most likely explanation is that we are misreading something, not that the market has made a thirty-point error. Her windows show the same decay as Sims — 90% over twenty, 80% over ten, 60% over five — and she has already turned in a 21-point, 11-rebound game this season, which is a losing outcome for this Under.
We could not reconcile the gap. Publishing a number we do not understand is worse than publishing nothing, so we did not.
We could not reconcile the gap. Publishing a number we do not understand is worse than publishing nothing, so we did not.
What this says about the method
The blended rate has a blind spot and today found it. Season, last-twenty, last-ten and last-five, weighted 40/30/20/10, is a reasonable way to describe a stable player. It is a bad way to describe a player whose role changed six weeks ago, because the heaviest weight sits on the stalest data. Both of tonight's candidates produced Edge Gaps roughly fifty percent larger than anything on our MLB board today, and in both cases the size of the number was itself the warning sign.
We are adding a standing check rather than a one-off note: when the last-five rate is materially below the last-twenty and the season rate, treat the Edge Gap as suspect until the player's current role is confirmed. A monotonic decline across those windows is not noise; it is usually a job description changing.
We are adding a standing check rather than a one-off note: when the last-five rate is materially below the last-twenty and the season rate, treat the Edge Gap as suspect until the player's current role is confirmed. A monotonic decline across those windows is not noise; it is usually a job description changing.
The matchup layer said yes to both. Golden State ranks 1st in the league defending both of these stat categories, which for an Under is the strongest possible support, and both legs came through the gate marked Confirms. It was the third check — asking whether the history still describes the player — that removed them. Worth recording, given we have spent two weeks asking whether the matchup layer earns its keep: today it was not the layer that saved us.
Yesterday, for what it is worth
Our Aug 16 WNBA card went 1-for-2 — Caitlin Clark Under 4.5 Rebounds hit with three, Angel Reese Under 2.5 Assists missed with six. More interesting is what happened to the three legs the matchup layer made us drop that day: Ezi Magbegor Under 13.5 Pts+Reb would have lost (21), Jade Melbourne Under 7.5 Reb+Ast would have lost (11), and Megan DiLeo Under 1.5 Assists would have won (1). Two saves and one bad drop. That is the first clearly positive day the layer has had, and it goes on the record alongside the negative ones.
Availability, since it mattered
Dallas has Azzi Fudd (knee) and Azaiah James (leg) ruled out, with Jessica Shepard (ankle) upgraded to probable. Golden State has ruled out Iliana Rupert for the season. Neither list resolved the role question above, which is why it remains open.
One caution on our own sourcing: an injury report we pulled while checking Sims turned out to be from June, republished under a headline that read as current. We caught it on the dateline. It is a good reminder that a search result is not a timestamp, and we mention it because it nearly went the other way.
One caution on our own sourcing: an injury report we pulled while checking Sims turned out to be from June, republished under a headline that read as current. We caught it on the dateline. It is a good reminder that a search result is not a timestamp, and we mention it because it nearly went the other way.
Fewer picks is always acceptable. Nothing on this page is a recommendation to bet.
Sources: hit-rate windows, opponent defensive ranks and prices from our props screen (our premium data platform), captured Mon Aug 17 8:05 AM ET. Schedule verified against ESPN. Aug 16 results graded from ESPN box scores. Role and availability reporting via our lineup service and Athlon Sports.