Two days after the Houston Rockets beat the Dallas Mavericks 135-117 in Macao, the teams meet again at the Venetian Arena for the second of the 2026 NBA China Games. Tip is 6:00 a.m. ET. Houston is 1-0 in the preseason and never trailed in the opener. Dallas is 0-1 and is still learning its own playbook under a first-time head coach. Our pick is Dallas Mavericks +2.5.
This is basketball, so there are no pitchers or park factors to weigh. What matters here is roster continuity, rotation intent and the unusual variance of an exhibition game, and that is the frame we use below.
The number
- Pick: Dallas Mavericks +2.5 (spread) - Model probability: n/a - Market-implied probability: n/a - Edge: None reported - Model confidence: None reported
Our model engine returned no data for this game. That means we have no projection, no probability and no measured edge. We cannot claim the model sees value at +2.5, and we will not reconstruct one. This pick goes on our public ledger like every other, but readers should treat it as a context-driven lean, not a quantified edge. Engines are typically thinnest on preseason games, because playing-time inputs are the least reliable at this point in the year.
Why the model leans this way
With no engine output, the case for Dallas +2.5 rests entirely on the researched context.
1. Preseason minutes often decide outcomes. The regular season begins October 20. Coaches use exhibitions to test lineups, and Houston showed its depth in the opener: Tari Eason came off the bench to lead the Rockets with 20 points. In a second game two days later, Houston has little reason to push its rotation. That holds especially for Fred VanVleet, whose opener was his first action after missing all of last season. When the starters sit, the fourth quarter becomes a game of depth players, and a 2.5-point margin is small.
2. Dallas has the stronger incentive to look at its core. The Mavericks are rebuilding around reigning Rookie of the Year Cooper Flagg. New coach Dusty May arrives from a national championship at Michigan, and he needs reps for Flagg alongside ninth overall pick Morez Johnson. A team that wants to see its new pairing in real game conditions may keep its key players on the floor longer than a 52-win team that already knows its identity.
3. Adjustment after a lopsided loss. Dallas has nearly half its roster new this offseason, plus a new front office led by Masai Ujiri. An 18-point loss gives that staff a full game of film on the same opponent. With a game plan in place and two days to install fixes, the margin could tighten. We cannot measure that effect, but it is the logic behind taking the points.
4. Setting. The Venetian Arena is owned by Las Vegas Sands Corp., whose owners also own the Mavericks. We do not assign this any measurable value, and it should not be read as a home-court effect. It is noted only for completeness.
The case against
The evidence against this pick is direct, and it is recent.
The opener was not close. Houston won by 18 and never trailed. A 2.5-point line asks Dallas to improve by roughly 16 points against the same opponent within 48 hours. Even with preseason noise, that is a large swing.
Houston's continuity is real. The Rockets come off back-to-back 52-win seasons. Both ended in first-round exits, including a six-game loss to the Los Angeles Lakers last spring, but the regular-season base is stable. They also added veteran guards Marcus Smart and Bogdan Bogdanović and drafted Bruce Thornton. Even Houston's second unit has experienced ball-handlers, so the depth argument made above can cut the other way: if both teams rest starters, the Rockets' reserves may simply be the stronger group.
Dallas is still forming. A first-time head coach, a new front office and nearly half the roster arriving from elsewhere is a recipe for uneven execution. Spacing, defensive coverage and late-game sets may not be settled in game two. The Flagg-Johnson pairing is an open question, not a proven strength.
Rotation risk runs both ways. If May decides to protect Flagg and spread minutes across his roster, Dallas loses its primary edge in effort and evaluation. Our research found no confirmed injury reports, so availability is also unknown in both directions.
How often does this lose? Without model output, we cannot give an honest percentage, and we will not invent one. Qualitatively, the opener margin and Houston's roster continuity make this a pick that could easily lose by a wide margin. Preseason variance may help Dallas, but variance does not favor either side by default. Readers should size any decision accordingly.
Bottom line
Dallas +2.5 is a lean built on preseason dynamics and Dallas's incentive to give its new core real minutes. It is not backed by a model number. The strongest evidence available, an 18-point Houston win two days ago, points the other way. We are publishing it with that disclosure attached, and the result will be recorded on our ledger either way.
21+ only. Predictions carry real risk of loss. Set limits in advance, and step away if it stops being fun.